sneak/routewatch
1
0
Fork 0
Code Issues 2 Pull Requests 1 Actions 25 Packages Projects Releases Wiki Activity

Compare commits

base: sneak/routewatch:issue-27-cached-stats
Branches Tags
sneak/routewatch:main sneak/routewatch:next sneak/routewatch:issue-27-cached-stats sneak/routewatch:prod
...
compare: sneak/routewatch:dfca772035e005de399fd5e1f95eff30dac2883d
Branches Tags
sneak/routewatch:next sneak/routewatch:main sneak/routewatch:issue-27-cached-stats sneak/routewatch:prod
48 Commits
issue-27-cached-stats ... dfca772035
Author SHA1 Message Date
sneak dfca772035 Handle an announcement without an origin like a replacement (closes #113)
check / check (push) Canceled after 0s
Details 
An announcement whose AS path is empty or ends in AS 0 now deletes the
peer's route for the prefix, as a withdrawal does. Before, it was neither
stored nor applied: the peer's old route stayed live, and within one
batch it cancelled the peer's earlier withdrawal of the prefix.

Model: opus-5-5
 2026-10-20 01:04:42 +00:00
clawbot 444457bcf8 Store a new row's earliest timestamp as first_seen (closes #112)
check / check (push) Canceled after 0s
Details 
Each handler took the latest time per key in a batch and inserted it as
a new row's first_seen, so a key updated twice in one batch, seconds
apart, was stored first seen at the later time, and the IP lookup could
answer with a prefix first_seen after the route's last_seen.

The prefix handler now keeps the earliest time of all of the batch's
updates for each prefix. The AS, peer and peering handlers keep the
earliest and the latest time per key in database.SeenTimes, and
GetOrCreateASNBatch, UpdatePeerBatch and RecordPeering store a new row
first seen at the earliest and last seen at the latest. A stored row
still has only its last_seen set, to the latest.

Model: opus-5-5
 2026-10-20 02:59:17 +02:00
clawbot c4dadd1477 Leave SIGINT and SIGTERM to fx (closes #131)
check / check (push) Canceled after 0s
Details 
The start hook registered its own handler for SIGINT and SIGTERM and
asked fx to shut down from it, although fx already stops the daemon on
both signals. A stop signal was handled twice, and the handler stayed
registered after the stop. fx now handles both signals alone, as
GO_HTTP_SERVER_CONVENTIONS.md "Signal Handling and Graceful Shutdown"
requires.

Model: opus-5-5
 2026-10-20 02:34:04 +02:00
clawbot c7f2e5db1d Bound the IP lookup's address list and name lookups (closes #110)
check / check (push) Canceled after 0s
Details 
An IP lookup took any number of addresses and host names, gave each PTR
lookup until the request's 30-second limit, and looked up and named
every address a host name resolved to. It now answers 400 for more than
10 entries, or more than 10 addresses once the names are resolved,
before reading the database. Each PTR lookup gets 2 seconds; one that
runs out leaves the address's result without PTR names. Of a host
name's addresses only public ones are kept: global unicast and not
private. A name with none is reported as having no public address. The
tests give the server a fake resolver.

Model: opus-5-5
 2026-10-20 02:23:49 +02:00
clawbot 09d3073fa5 Bound each shutdown stage by the stop deadline (closes #100)
check / check (push) Canceled after 0s
Details 
Shutdown passes its context to every stage: the streamer, the four
handlers, the WHOIS fetcher, the database maintainer and the HTTP server
each stop waiting when it ends. A handler then skips its last write and
logs how many updates it did not write, and the HTTP server closes the
connections still open. The database is checkpointed and closed last,
and left open once the deadline has passed, since a write may still
hold it. The stop hook gives Shutdown 8 seconds and fx 9, inside the 10
seconds upaas and docker stop allow, so a stop signal exits 0.

Model: opus-5-5
 2026-10-20 02:15:10 +02:00
clawbot 237f73fc49 Answer 500, not 404, when the IP lookup cannot read the database (closes #109)
check / check (push) Canceled after 0s
Details 
A failed database read for an address was listed in the answer's errors
as if the address had no route, so the lookup answered 404, or 200 next
to a routed address. Only a missing route or an invalid address is
listed now; any other read error is logged and answers 500 with the JSON
error body, as the AS and prefix JSON routes do.

Model: opus-5-5
 2026-10-20 01:49:34 +02:00
clawbot 89bdda18a5 Add the HTTP hardening the policy requires before 1.0 (closes #108)
check / check (push) Canceled after 0s
Details 
Every answer carries the six security headers REPO_POLICIES.md lists.
The Content-Security-Policy allows inline content only with a nonce made
for that answer; the pages put it on their style and script blocks and
drop their four style attributes.

The request log and /ip/ take a forwarded address only from a proxy
listed in TRUSTED_PROXIES, which stops the start when set but invalid:
the last X-Forwarded-For entry that is not a trusted proxy. X-Real-IP
and True-Client-IP are no longer read.

The JSON routes and the health check send fixed messages in place of
database, parser and resolver errors, and log the detail.

Model: opus-5-5
 2026-10-20 01:36:18 +02:00
clawbot 76c2e523b5 Checkpoint without waiting on readers and log a busy checkpoint (closes #102)
check / check (push) Canceled after 0s
Details 
The 5-second checkpoint ran in TRUNCATE mode, which waits up to the
5-second busy timeout for every open read while holding the database
lock every batch write needs. It now runs in PASSIVE mode, which never
waits. A WAL file over 64 MiB instead gets one TRUNCATE checkpoint whose
wait for reads is capped at 500 ms and logged; it empties the file once
no read is open, and while reads follow one another with no pause it
can give up every time. A checkpoint that leaves part of the WAL
uncopied returns ErrCheckpointIncomplete, which the maintainer logs as
incomplete instead of completed. The writing connection's
journal_size_limit is 64 MiB.

Model: opus-5-5
 2026-10-20 01:06:51 +02:00
clawbot 28f60500cb Read the AS page from bounded lookups (closes #104)
check / check (push) Canceled after 0s
Details 
The AS page and /api/v1/as/{asn} grouped every route the AS originates
by prefix, reading one table row per peer and prefix and sorting them
all. They now take the AS's prefixes one at a time: one search of a new
index on origin_asn and prefix, which replaces the origin_asn index and
is written exactly when it was, and one search of the prefix and
last_updated index for the time of the prefix's newest route. Neither
reads the table. For a prefix more than one AS originates, that time is
now the newest route's from any origin. A hand-run benchmark reads an
AS of 5,000 prefixes from 500 peers.

Model: opus-5-5
 2026-10-20 00:52:32 +02:00
clawbot 69d477159c Cap WHOIS answers and follow referrals only to WHOIS servers (closes #106)
check / check (push) Canceled after 0s
Details 
The client kept every line a WHOIS server sent until the server closed
the connection or 30 seconds passed, so a server that kept sending
could fill memory. It now reads at most 1 MiB of an answer, and an
answer that reaches that is an error. A "refer:" line in IANA's answer
naming a server none of the five RIRs run was dialled as given, on any
host and port; it is now an error, so the client only connects to IANA
and the five RIR servers. Both errors are failed lookups, which the
fetcher retries later.

Model: opus-5-5
 2026-10-20 00:15:14 +02:00
clawbot abdc327c4c Retry a WHOIS lookup that failed instead of storing it as complete (closes #105)
check / check (push) Canceled after 0s
Details 
A lookup whose RIR query failed, or whose RIR answer carried none of the
WHOIS fields (a refusal of service, for one), returned IANA's answer as
a success. The fetcher stored it as complete with blank fields, did not
look the AS up again for 30 days, and shortened its interval. Such a
lookup is now an error: the fetcher backs off and leaves the row as it
was. The fields are read from the RIR's answer alone. An AS IANA refers
to no RIR is stored with no WHOIS data, so it is not looked up again
until it is stale, counting as neither a success nor a failure. Tests
reach fake WHOIS servers through the client's DialContext.

Model: opus-5-5
 2026-10-20 00:02:34 +02:00
clawbot a173f4d159 Keep a peer-down until its route delete succeeds (closes #101)
check / check (push) Canceled after 0s
Details 
When the delete of a peer's live routes after its session went down
failed, the down was dropped and the routes stayed. The prefix handler
now keeps such a peer, once, with the updates it sent after the down,
for the next flush, which retries the delete before it writes them; each
failure is logged. At most the last 25,000 of those updates are kept,
outside the batch, so they do not start a flush. The delete runs in
chunks of at most 10,000 routes per table, taken in prefix order
through the peer's key, each chunk in its own transaction, so the
database lock is released between chunks and a failure undoes at most
one chunk.

Model: opus-5-5
 2026-10-19 23:39:28 +02:00
clawbot 176f9708b8 Build and test with a supported Go release (closes #107)
check / check (push) Canceled after 0s
Details 
The test and build stages move from golang:1.24-bookworm, a release
line that no longer gets security fixes, to golang:1.27.2-bookworm,
pinned by digest. The go directive in go.mod moves from 1.24.4 to
1.27.1, the Go in the pinned lint image, so no stage downloads another
toolchain. One test starts its goroutine with sync.WaitGroup.Go, which
the linter asks for once the module targets Go 1.25 or later.

Model: opus-5-5
 2026-10-19 23:19:21 +02:00
clawbot d2a6e30360 Percent-decode URL parameters before parsing them (closes #103)
check / check (push) Canceled after 0s
Details 
chi matches routes against the escaped path when a request has one and
hands the handlers their URL parameters still percent-encoded. The home
page's lookup form encodes the address it asks for, so every IPv6 lookup
answered 400, as did a comma-separated address list or a prefix that a
client percent-encoded. The router now matches against the decoded path,
so every handler reads its parameters decoded.

Model: opus-5-5
 2026-10-19 23:02:00 +02:00
clawbot ef32082339 Bring README and TODO.md up to date with the code (closes #72)
check / check (push) Canceled after 0s
Details 
README "HTTP Endpoints" now lists every route the server registers with
what it answers, errors included: / is a home page and /ip/{addr} answers
JSON, neither redirects; /ip/ looks up the caller; the IP endpoints take
host names and comma-separated lists; 400, 404 and 504 are described, and
the health check's envelope. "Database Schema" matches schema.sql, and
other sentences the code contradicted are corrected. TODO.md gets a dated
line for each issue landed since 2026-10-03, with Status and Next Step
current.

Model: opus-5-5
 2026-10-19 16:09:22 +02:00
clawbot d84dd56460 Make live route writes cheaper so the prefix handler keeps up (closes #96)
check / check (push) Canceled after 0s
Details 
The live route tables lose their random UUID key; the row id the
prefix-length sample draws is declared as the integer key. The prefix
tables are keyed by prefix and keep only first_seen; a batch inserts only
prefixes not stored yet. A route update sets only the columns that
change. Each kept index names the queries using it. Every write goes
through one connection kept for writing, with a 1 GiB page cache; reading
connections get 16 MiB, heap limits rise to 1.5 and 2 GiB (README
"Memory"). Batch writes go in key order. A benchmark run by hand writes
250,000 updates on 5 million routes faster than the feed's 10,000 a
second; a live run follows.

Model: opus-5-5
 2026-10-19 15:43:38 +02:00
clawbot 5806ddaef7 Run database maintenance under the database lock (closes #94)
check / check (push) Canceled after 0s
Details 
The database maintainer ran Vacuum and Analyze without the database lock
every batch write holds, so on the live feed every incremental vacuum
failed with "database is locked", and an hourly ANALYZE made a batch write
fail and lose its updates. Vacuum, Analyze and Checkpoint now take that
lock, so maintenance and batch writes wait for each other. Analyze runs
PRAGMA optimize with an analysis limit, which analyzes only tables whose
size changed a lot, so it holds the lock briefly however large the file.

Model: opus-5-5
 2026-10-19 13:26:36 +02:00
clawbot 28df9f8d49 Drain the handler queues before stopping the handlers on shutdown (closes #69)
check / check (push) Canceled after 0s
Details 
Shutdown stopped the handlers before the streamer, so updates still in the
handler queues, up to 20,000 per queue, were not written, and a lost
withdrawal left its route in the table after a restart.

Shutdown now stops the streamer first. Streamer.Stop queues the STATE
messages it holds for a full queue, closes the queues and returns once
every worker has handed its handler everything on its queue; then each
handler writes what it holds before its Stop returns. PeeringHandler.Stop
now waits for its loop's final write instead of running a second one
beside it.

Model: opus-5-5
 2026-10-19 13:08:48 +02:00
clawbot 2712183f21 Keep a trailing AS set out of the origin and the peerings (closes #75)
check / check (push) Canceled after 0s
Details 
A path ending in an AS set, such as [64496, 64497, [64498, 64499]], was
flattened into the sequence, so the route was stored as originated by the
last set member and the peering handler recorded peerings the path does not
show. ASPath now keeps the sequence and a trailing set of several members
apart; Origin() gives the last AS of the sequence, or 0 (no known origin)
for such a set. The handlers use it; peerings come from the sequence only.
The pages show origin 0 as no known origin, and the IP lookup asks no WHOIS
lookup for it. Empty paths are unchanged.

Model: opus-5-5
 2026-10-19 12:50:05 +02:00
clawbot 8742f1b44d Let bgp_peer_count drop to 0 (closes #92)
check / check (push) Canceled after 0s
Details 
SetBGPPeerCount stored a count only when it was above 0, so when the last
peer the streamer knew of went down, bgp_peer_count kept its old value. It
now stores every count from 0 up.

Model: opus-5-5
 2026-10-19 12:41:15 +02:00
clawbot 3667d09800 Count every peer the feed has heard from in bgp_peer_count (closes #74)
check / check (push) Canceled after 0s
Details 
bgp_peer_count grew only on OPEN messages, so it started at 0 while
updates arrived from hundreds of peers whose sessions were already up, and
it kept peers whose sessions had gone down. Now any message from a peer
adds it to the streamer's peer set, and a NOTIFICATION or a STATE message
whose state is down removes it without adding it first. The rule is in
updatePeers, called for every message.

Model: opus-5-5
 2026-10-19 12:24:49 +02:00
clawbot b89987e356 Re-vendor the shared files from sneak/prompts at 0653579 (closes #52)
check / check (push) Canceled after 0s
Details 
Takes .golangci.yml, .dockerignore, .gitignore, .editorconfig, the
workflow and REPO_POLICIES.md from sneak/prompts 0653579, with this
repository's own entries after the shared content. Lint (golangci-lint
v2.14.0) and test are Dockerfile phases that script/lint and script/test
build alone with a new CACHEBUST value, keeping Go's and golangci-lint's
caches; the image build depends on both. make fmt and make fmt-check also
format the markdown with prettier, pinned in yarn.lock. The agent
guidance moves to AGENTS.md. The Go code is fixed for what the new
configuration finds, and the AS and prefix detail reads now return the
rows' error after their loops.

Model: opus-5-5
 2026-10-19 12:10:11 +02:00
clawbot 4f515d718b Read feed lines up to 1 MiB (closes #76)
check / check (push) Canceled after 0s
Details 
The streamer read the feed with a bufio.Scanner at its default limit of
64 KiB per line, so a longer line ended the connection and the updates sent
while it reconnected were lost. The scanner now accepts lines up to 1 MiB
(maxLineSize), well above an extended BGP message written out as the feed
writes it.

Model: opus-5-5
 2026-10-19 11:49:47 +02:00
clawbot 825afa2ebe Remove a peer's live routes when its session goes down (closes #56)
check / check (push) Canceled after 0s
Details 
RIS sends no withdrawals for the routes of a peer whose BGP session goes
down, so routewatch kept them.

The streamer now hands STATE messages to the prefix handler. They are never
dropped for backpressure: one that finds the queue full is held and queued
first once there is room, and the handler's other messages are dropped
meanwhile, so none overtakes it. Nothing waits. On a down, the prefix
handler drops the peer's pending updates and its next flush deletes all the
peer's live routes before the batch's other writes, keeping the in-memory
counts exact. The live route tables are unique on peer and prefix, which
also finds a peer's routes.

Model: opus-5-5
 2026-10-19 11:40:31 +02:00
clawbot a0ddc0d254 Sample the prefix-length pages with bounded lookups, each prefix once (closes #66)
check / check (push) Canceled after 0s
Details 
The prefix-length pages read every live route of that length and listed a
prefix once per peer sharing its newest time.

A length with at most 5000 prefixes now has its prefixes listed with one
index search each and 500 picked from them at random. A longer one is
sampled with random row id lookups on the mask length index, at most four
per prefix, topped up from the listed prefixes. Each picked prefix's newest
route is one search of a new index on prefix and last_updated, which
replaces the prefix-only index. Each prefix is listed once. Above 5000
prefixes the sample is uneven: a route is picked in proportion to the row
ids before it.

Model: opus-5-5
 2026-10-19 10:29:01 +02:00
clawbot ae6d885a5a Answer 504, not 408, when a request outlasts its time limit (closes #70)
check / check (push) Canceled after 0s
Details 
A request past the server's 30-second limit answered 408, which tells the
client it was too slow sending its request, and /api/v1/stats answered 500
when its 4-second read ran out. The limit is now chi's middleware.Timeout,
which answers 504 and never writes from a second goroutine while the
handler is writing.

Right after each database read or host name lookup, whatever came back, a
handler asks deadlinePassed whether its time limit has passed and, if so,
answers 504, with the JSON error body on JSON routes: a cut-off read can
return part of its rows without an error. The health check is unchanged.

Model: opus-5-5
 2026-10-19 10:02:23 +02:00
clawbot 43c61ee0d0 Write each handler batch from its flush goroutine, outside the mutex (closes #65)
check / check (push) Canceled after 0s
Details 
The AS, peer and prefix handlers wrote their batch to the database while
holding the mutex HandleMessage needs, so while a write ran nothing was
taken off their queues and the streamer dropped updates.

Each handler now writes only from its flush goroutine: it takes the batch
under the mutex, leaves an empty one in its place, and writes after
releasing it. A full batch wakes the flush goroutine. While a write runs the
next batch may grow to ten batch sizes; at that size HandleMessage waits, so
the queue fills and the streamer drops as before. README "Memory" gives the
numbers and their cost.

Model: opus-5-5
 2026-10-19 09:45:24 +02:00
clawbot f7c88c2496 Identify a live route by prefix and peer (closes #55)
check / check (push) Canceled after 0s
Details 
A peer has at most one route per prefix, but both live route tables were
unique on prefix, origin AS and peer. A peer that changed a prefix's origin
AS left its old route behind, and a withdrawal sent in the same message as
announcements was given their origin AS and missed the withdrawn route.

Both tables are now unique on prefix and peer. An announcement replaces the
peer's row for the prefix, origin AS included; a withdrawal deletes it
without looking at an origin AS. The route counts and prefix distribution
kept in memory stay exact, for IPv4 and IPv6.

Model: opus-5-5
 2026-10-19 09:13:25 +02:00
clawbot e0b9c2685f Reconnect when RIS Live ends or stalls the stream (closes #53)
check / check (push) Canceled after 0s
Details 
When RIS Live ended the response normally, the reconnect loop stopped for good and the health check kept reporting the feed connected; a connection that stayed open but sent nothing blocked forever. Now only `Stop` ends the loop: any other end clears the connected flag and reconnects with the existing backoff. If no line arrives for 60 seconds, including while waiting for RIS Live to answer the request, the streamer drops the connection and reconnects. Tests use a local fake server. README "Reconnection" describes both.

Model: opus-5-5
 2026-10-14 19:42:57 +02:00
clawbot ad4c3d91f1 Show the prefix page's origin names from the prefix query (closes #67)
check / check (push) Canceled after 0s
Details 
The prefix page looked up each origin AS's details only to show its handle and description, and that lookup reads every live route the AS originates. `GetPrefixDetailsContext` already selected the handle and description and dropped them; it now returns them keyed by AS number, and the page uses them. The origin list now also shows the description after the handle, which it never rendered. The JSON prefix answer is unchanged. The test serves the page from a store whose AS lookup always fails.

Model: opus-5-5
 2026-10-14 19:08:58 +02:00
clawbot b4dee48401 Exit with status 1 when the web server cannot listen (closes #73)
check / check (push) Canceled after 0s
Details 
`Server.Start` opened its port inside a goroutine and returned nil at once, so a port already in use was only logged and the daemon kept running with no web server. It now opens the listener before returning and returns that error; serving runs in the background on the open listener. `CLIEntry` turns an error from `Run` into an fx shutdown with exit status 1, after the normal stop sequence, so a container restart policy can act. A serving error after a successful listen is still only logged.

Model: opus-5-5
 2026-10-14 18:26:04 +02:00
clawbot 616f13071f Parse the prefix on the prefix pages before looking it up (closes #71)
check / check (push) Canceled after 0s
Details 
`/prefix/{prefix}/{len}` and `/api/v1/prefix/{prefix}/{len}` passed the URL text to the lookup as typed: a prefix that does not parse answered 500, and a valid prefix written in another form (`2001:DB8:0:0::/32`, `192.0.2.1/24`) answered 404. Both now parse the prefix with `net/netip`, mask it and look up the form the tables store; one that does not parse answers 400. The answer shows the stored form of the prefix. The test drives both pages through the router against a real database.

Model: opus-5-5
 2026-10-14 17:24:02 +02:00
clawbot 521c632087 Ignore routes shorter than /8 or /16 in the IP lookup (closes #57)
check / check (push) Canceled after 0s
Details 
The IP lookup tried every mask length down to /0, so a default route or a very short route such as `2000::/3` was reported as the owner of any address it covered until a more specific route had been seen. The lookup now stops at /8 for IPv4 and /16 for IPv6 (named constants); an address covered only by shorter routes gets "no route found". The tables still store short routes. The test checks each limit and the length one bit shorter.

Model: opus-5-5
 2026-10-14 17:23:51 +02:00
clawbot 77a708d819 Apply each peer's last update for a prefix in a batch (closes #54)
check / check (push) Canceled after 0s
Details 
The prefix handler kept one update per prefix in each batch across all peers, chosen by a timestamp cut to whole seconds, so most peers' announcements and withdrawals were lost and the first update in a second won. A batch now keeps the last update per prefix and peer in feed order. A message's withdrawals are queued before its announcements, so a prefix both withdrawn and announced in one message keeps its route (RFC 4271 section 4.3). Tests run each case through the handler into a real database and check the in-memory counts.

Model: opus-5-5
 2026-10-14 17:09:13 +02:00
clawbot a43935308e Count every reconnect in reconnect_count (closes #68)
check / check (push) Canceled after 0s
Details 
`reconnect_count` in `/api/v1/stats` stayed 0 however often the stream reconnected: `SetConnected` counted a reconnect only while the stream was still marked connected, and the streamer always marks it disconnected first. It now counts one reconnect for every connection after the first, recognised by the zero `connectedSince` before the first connection. A test connects and disconnects twice and expects 2.

Model: opus-5-5
 2026-10-14 15:59:15 +02:00
sneak 00e51d284f Merge pull request 'Next milestone: production memory under 5 GiB, ready for upaas' (#6) from next into main
check / check (push) Canceled after 0s
Details 
Reviewed-on: #6
 2026-10-06 00:34:48 +02:00
clawbot a9b68608c6 Serve the /api/v1/stats prefix distribution from memory (closes #30)
check / check (push) Successful in 3m19s
Details 
The prefix distribution was the last query on the stats path. It read one index entry per live route on every request, so on a large database it took the whole 4-second deadline and /api/v1/stats answered 500.

The distribution now lives next to the in-memory counts: seeded once at startup from the same query, then kept exact by every live-route write. A new route whose prefix had no live route adds one at its mask length; a delete that leaves a prefix with no live route takes one away. Each check is one lookup on the prefix index, within the same write. The single-route delete now runs its delete and that lookup in one transaction.

Model: opus-5-5
 2026-10-03 18:43:58 +02:00
clawbot 44a5f4cdca Create new databases with auto_vacuum incremental (closes #43)
check / check (push) Successful in 2m50s
Details 
SQLite only accepts auto_vacuum before the database file is first written. The connection switched to WAL first, which writes the file, so the PRAGMA auto_vacuum in Initialize came too late and was ignored. The setting now goes in the connection string, which the driver applies on open before the journal mode, and the late PRAGMA is removed.

Vacuum now reads every row PRAGMA incremental_vacuum returns: SQLite frees one page per row, and the single step ExecContext took freed only one page per call.

Tests check that every pooled connection sees auto_vacuum incremental on a new database and that one Vacuum call frees every page left by deleting routes.

Model: opus-5-5
 2026-10-03 17:09:29 +02:00
clawbot 32704c601e Look up IP addresses by prefix at each mask length (closes #48)
check / check (push) Successful in 3m13s
Details 
Looking up an address on /ip/ and /api/v1/ip/ read every IPv6 live route and about half of the IPv4 range index, so on a day-sized database it passed the 30-second request timeout. The lookup is now one function for both families: from the longest mask length down, it looks up the address's network prefix on the existing prefix index, and the first live route wins.

The feed sends IPv6 withdrawals uncompressed, so they never matched the stored compressed prefix and never removed a route. Prefixes are now stored in the text form net/netip prints, so IPv6 withdrawals take effect.

ip_start, ip_end and GetASInfoForIP are removed; nothing read them. A database from before this change must be deleted.

Model: opus-5-5
 2026-10-03 16:04:38 +02:00
clawbot 7322f936e4 Stamp the git tag or short commit in a plain docker build (closes #46)
check / check (push) Successful in 3m6s
Details 
A plain `docker build .` stamped `unknown` into the page footer:
.dockerignore left out .git, and the Dockerfile built without the -X
flags. The context now carries .git without .git/config, which can hold
a credential. The build stage takes the VERSION build argument when one
is given, otherwise `git describe --tags --always`, fails the build if
.git is present and no version comes out, and stamps it together with
the full commit the footer links to. `make build` now uses
`git describe` as well, and script/docker is the current shared copy.

Model: opus-5-5
 2026-10-02 08:17:35 +02:00
clawbot 6422d9fa0c Container makes its data directory usable itself (closes #42)
check / check (push) Successful in 8s
Details 
sneak's standing rule: the container makes its data directory usable itself, with no step on the host. entrypoint.sh now creates /var/lib/berlin.sneak.app.routewatch if it is missing and stops the start when any step fails (set -euo pipefail); before, a failed cd went on to change the ownership of whatever directory the script was in, and a failed chown still started the daemon. Taking ownership of the directory and switching to the routewatch user through setpriv are unchanged. The README's upaas volume line now says only which path to mount.

The empty-directory and other-uid cases were run by hand on the built image with upaas-style bind mounts, not added as an automated test.

Model: opus-5-5
 2026-09-29 12:23:42 +02:00
clawbot 057e0bd9a9 Add a .dockerignore (closes #39)
check / check (push) Successful in 6s
Details 
Adds a .dockerignore, which REPO_POLICIES.md lists among the files every repo has. Until now every docker build sent the whole working tree as its build context, and the image's source archive is made from that context. It now keeps out .git, local build and test output, archives, local databases, .env and a local Go workspace. Every tracked file stays in, so the format check, the linter, the tests, the build and the source archive read the same files as before.

Without .git in the context the binary's build info records no git revision; nothing in routewatch reads it.

Model: opus-5-5
 2026-09-29 07:42:20 +02:00
clawbot 87a3bc115e README: license and author up front; bring TODO.md up to date (closes #38)
check / check (push) Successful in 3m14s
Details 
Brings README.md and TODO.md in line with REPO_POLICIES.md and with what is on next. The README's first line now names routewatch, what it is, its MIT licence and its author, @sneak; the License section says MIT and links LICENSE; a new Author section names @sneak.

TODO.md's Status, Next Step and Future Steps now describe next as it is: it waits for sneak to merge the milestone PR, after which the upaas deploy and the run under a real 5 GiB limit are his. Completed Steps gains a line for each issue that landed without one.

Docs only. make fmt formats only Go here, so the markdown was wrapped by hand.

Model: opus-5-5
 2026-09-29 07:06:30 +02:00
clawbot d28a59023a Add MIT LICENSE (closes #1)
check / check (push) Successful in 4m5s
Details 
Adds LICENSE with the MIT licence sneak chose for routewatch, copied byte for byte from sneak/webhooker's LICENSE: copyright 2026, Jeffrey Paul. The rest of the issue, .editorconfig and the Gitea check workflow, was already on next, and the README already points at LICENSE.

Judgement call: our repos write the copyright holder in several different ways and sneak named no form; webhooker's gives his full name and email address, the most explicit of them.

Model: opus-5-5
 2026-09-29 05:05:35 +02:00
clawbot 1ac24669d3 Stop the streamer without sending on closed queues (closes #34)
check / check (push) Successful in 2m38s
Details 
Stopping the daemon while the RIS Live feed was flowing could panic with "send on closed channel" and skip the rest of the shutdown. Stop cancels the stream and closes the handler queues under the streamer's write lock, but the read loop checked for a stop only before parsing each line. It now checks again under the read lock it already takes just before handing a message to the queues, so it never sends to a closed queue. Stop also clears its cancel function and returns early when there is none, so a second call no longer closes the queues again. A test forces both cases.

Behaviour change: Stop before Start now does nothing.

Model: opus-5-5
 2026-09-28 21:42:33 +02:00
clawbot 6187ac8503 Let the daemon receive docker stop's signal itself (closes #33)
check / check (push) Successful in 2m42s
Details 
entrypoint.sh now switches to the routewatch user (UID 1000) with setpriv instead of runuser. setpriv replaces itself with the daemon, so the daemon is the container's main process and receives docker stop's signal itself. runuser stayed in between, passed the signal on and killed the daemon 2 seconds later, so every stop ended with exit 143. The daemon now gets the whole wait the caller allows, up to its own 60-second limit, and a clean stop exits 0. Taking ownership of the state directory and the MALLOC_ARENA_MAX check still run as root first.

Not fixed here: a stop while the feed is flowing can still panic (#34).

Model: opus-5-5
 2026-09-28 21:08:33 +02:00
clawbot f2a9e90625 Refuse invalid settings at start, health check follows PORT (closes #31)
check / check (push) Successful in 3m17s
Details 
A set but invalid PORT (anything but plain digits from 1 to 65535) or a relative XDG_DATA_HOME now stops the start before the database opens; before, a bad PORT left the daemon running without HTTP. entrypoint.sh refuses a MALLOC_ARENA_MAX that is not a positive whole number, since glibc ignores a bad one silently. The HEALTHCHECK probes the port PORT names, 8080 when unset. PORT is now read in internal/config, so server.New takes the config.

The README gives the real Linux state directory, lists XDG_DATA_HOME, and adds "Running under upaas": port, volume, environment, the 5g memory limit and the health check.

Unverified: the 5g memory limit could not be exercised on the build host.

Model: opus-5-5
 2026-09-28 20:08:42 +02:00
clawbot df9e23d503 Serve /api/v1/stats counts from realtime in-memory counters (closes #27)
check / check (push) Successful in 2m36s
Details 
Realtime in-memory counters seeded at startup and adjusted on every insert, update and delete; no periodic recompute. Independent review passed: #29 (comment)

model: claude-opus-4-8 (implementation and review); merged by claude-fable-5
 2026-09-22 09:41:16 +02:00
87 changed files with 16089 additions and 4945 deletions
Expand all files Collapse all files
.dockerignore
+97
View File
@@ -0,0 +1,97 @@
# .dockerignore does NOT use .gitignore semantics. Docker matches with
# moby/patternmatcher: filepath.Match plus `**`, so `*` does not cross
# `/` and an unprefixed pattern is anchored at the context root. Every
# depth-independent pattern therefore needs `**/`, or `config/.env` and
# `certs/server.key` still ship while this file reads as solved. Only
# genuinely root-anchored entries go unprefixed. Never transplant these
# into .gitignore, where `**/` is wrong.
#
# Matching is case-sensitive, so secrets use character ranges rather
# than an ALL-CAPS twin, which would still miss `Server.Key`.
#
# Extend with this repo's own host-built artifacts, written anchored:
# `/myapp`, never `**/myapp`, which also matches `cmd/myapp/` and
# deletes the package directory from the context.
# .git is sent without its config. Without a VERSION build argument the
# stage that compiles runs `git describe --tags --always` on .git, which
# does not need .git/config; that file can hold a credential, such as a
# password in a remote URL or the token the CI checkout step stores there.
# Each submodule keeps a config with the same exposure in its git directory
# under .git/modules/, nested again for a submodule's own submodules, or in
# its own .git directory when it keeps one.
# KNOWN GAP: a submodule whose name has a `config` segment (`config`,
# `deploy/config`, `config/lib`) loses its whole git directory, because
# `**/.git/modules/**/config` also matches that segment's directory
# under .git/modules/. Go's version stamping then fails the build;
# nothing leaks. Name such a submodule without that segment:
# `git submodule add --name`.
**/.git/config
**/.git/modules/**/config
# Agent scratch: one full checkout of the repo per in-flight agent.
# Anchored because it occurs once where agents run at the repo root.
# KNOWN GAP: a repo running agents in subdirectories still ships
# `services/api/.claude/` and must add its own anchored entry.
.claude
# Environment files. `*.env` covers bare `.env` and the `prod.env`
# convention. Re-include a committed template with a negation if the
# build needs one: `!docs/example.env`.
**/*.[eE][nN][vV]
**/.[eE][nN][vV].*
**/.[eE][nN][vV][rR][cC]
# Private keys and the bundles carrying them. Public certificates
# (*.crt, *.cer) are deliberately absent: they are legitimate inputs.
**/*.[pP][eE][mM]
**/*.[kK][eE][yY]
**/*.[pP]12
**/*.[pP][fF][xX]
**/[iI][dD]_[rR][sS][aA]
**/[iI][dD]_[dD][sS][aA]
**/[iI][dD]_[eE][cC][dD][sS][aA]
**/[iI][dD]_[eE][cC][dD][sS][aA]_[sS][kK]
**/[iI][dD]_[eE][dD]25519
**/[iI][dD]_[eE][dD]25519_[sS][kK]
# Dependencies: restored inside the image, never copied in.
**/node_modules
# OS metadata.
**/.DS_Store
**/Thumbs.db
# Editor state: never a build input, and it churns COPY.
**/*.swp
**/*.swo
**/*~
**/*.bak
**/.idea
**/.vscode
**/*.sublime-*
# This repository's host-built artifacts: `make build` writes bin/, `go
# build` in the root the binaries, `make run` log.txt, `make asupdate` its
# .tmp file; test binaries, coverage profiles and source archives.
/bin
/routewatch
/asinfo-gen
/streamdumper
/log.txt
/out
/pkg/asinfo/asdata.json
**/*.tar.zst
**/*.test
**/*.out
**/*.tmp
# Local databases. The image carries a source archive of the whole build
# context, so these would otherwise ship inside it.
**/*.db
**/*.db-journal
**/*.db-wal
# A local Go workspace points at directories outside the build context.
/go.work
/go.work.sum
.editorconfig
+6
View File
@@ -10,3 +10,9 @@ insert_final_newline = true
[Makefile]
indent_style = tab
[*.go]
indent_style = tab
# This repository's own sections, such as one for another language it
# uses, go below this comment, and a re-vendor keeps them.
.gitea/workflows/check.yml
+12 -1
View File
@@ -1,9 +1,20 @@
name: check
on: [push]
# Free the shared runner: a new push cancels only the same branch's older run.
concurrency:
group: ${{ github.workflow }}-${{ github.ref }}
cancel-in-progress: true
jobs:
check:
runs-on: ubuntu-latest
# Free the shared runner from a hung build.
timeout-minutes: 20
steps:
# actions/checkout v4.2.2, 2026-02-28
# actions/checkout v4.2.2, 2026-02-22
- uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
# script/cibuild needs no token, so none is left in .git/config.
with:
persist-credentials: false
# All history and tags, so git describe finds the version tag.
fetch-depth: 0
- run: script/cibuild
.gitignore
+65 -27
View File
@@ -1,41 +1,79 @@
# Binaries for programs and plugins
# OS
.DS_Store
Thumbs.db
# Editors
*.swp
*.swo
*~
*.bak
.idea/
.vscode/
*.sublime-*
# Agent scratch (worktrees of this repo, created and destroyed by
# in-flight tooling). Unanchored: .gitignore patterns already match at
# every depth, so no prefix is wanted here. This is not a .dockerignore
# entry and must not be given a `**/` prefix on the way into one.
.claude/
# Node
node_modules/
# Secrets. Unanchored like every entry above, so each matches at every
# depth. Matching is case-sensitive on Linux, so names use character
# ranges rather than a lowercase form that misses `Server.Key`.
# Environment files. `*.env` covers bare `.env` and the `prod.env`
# convention. Only the templates `example.env` and `sample.env` are
# re-included below. A repository that commits any other template adds
# its own negation at the end of this file, for example `!.env.example`.
*.[eE][nN][vV]
.[eE][nN][vV].*
.[eE][nN][vV][rR][cC]
!example.env
!sample.env
# Private keys and the bundles carrying them.
*.[pP][eE][mM]
*.[kK][eE][yY]
*.[pP]12
*.[pP][fF][xX]
[iI][dD]_[rR][sS][aA]
[iI][dD]_[dD][sS][aA]
[iI][dD]_[eE][cC][dD][sS][aA]
[iI][dD]_[eE][cC][dD][sS][aA]_[sS][kK]
[iI][dD]_[eE][dD]25519
[iI][dD]_[eE][dD]25519_[sS][kK]
# This repository's own entries, such as its build outputs, go below
# this comment, and a re-vendor keeps them. Anchor a binary built at the
# root: `/myapp`, never `myapp`, which also ignores `cmd/myapp/`.
# Go: logs, test binaries, coverage profiles, and the binaries `make build`
# and `go build` in the repository root write.
*.log
*.out
*.test
*.exe
*.exe~
*.dll
*.so
*.dylib
/bin/
.DS_Store
log.txt
# Test binary, built with `go test -c`
*.test
# Output of the go coverage tool, specifically when used with LiteIDE
*.out
# Dependency directories (remove the comment below to include it)
# vendor/
# Go workspace file
/routewatch
/asinfo-gen
/streamdumper
go.work
go.work.sum
# env file
.env
# Database files
# Local databases, temporary files and debug output.
*.db
*.db-journal
*.db-wal
# Temporary files
*.tmp
# Raw AS data (we only commit the gzipped version)
pkg/asinfo/asdata.json
# Debug output files
out
log.txt
out
# Raw AS data (only the gzipped version is committed).
pkg/asinfo/asdata.json
.golangci.yml
+90 -81
View File
@@ -1,92 +1,101 @@
version: "2"
# Config schema uses the golangci-lint v2 layout (settings live under
# linters.settings, not top-level linters-settings) so that the
# thresholds below are actually applied by golangci-lint >= v2.
run:
go: "1.24"
tests: false
timeout: 5m
modules-download-mode: readonly
linters:
default: all
enable:
# Additional linters requested
- testifylint # Checks usage of github.com/stretchr/testify
- usetesting # usetesting is an analyzer that detects using os.Setenv instead of t.Setenv since Go 1.17
# - tagliatelle # Disabled: we need snake_case for external API compatibility
- nlreturn # nlreturn checks for a new line before return and branch statements
- nilnil # Checks that there is no simultaneous return of nil error and an invalid value
- nestif # Reports deeply nested if statements
- mnd # An analyzer to detect magic numbers
- lll # Reports long lines
- intrange # intrange is a linter to find places where for loops could make use of an integer range
- gochecknoglobals # Check that no global variables exist
# Default/existing linters that are commonly useful
- govet
- errcheck
- staticcheck
- unused
- ineffassign
- misspell
- revive
- gosec
- unconvert
- unparam
linters-settings:
lll:
line-length: 120
nestif:
min-complexity: 4
nlreturn:
block-size: 2
revive:
rules:
- name: var-naming
arguments:
- []
- []
- "upperCaseConst=true"
tagliatelle:
case:
# Successor to the deprecated gomodguard. Named explicitly, rather than
# left to `default: all`, because it carries the module policy below.
- gomodguard_v2
disable:
# Genuinely incompatible with project patterns
- exhaustruct # Requires all struct fields
- exhaustruct_v5 # Requires all struct fields (successor to exhaustruct)
- godot # Requires comments to end with periods
- wrapcheck # Too verbose for internal packages
- varnamelen # Short names like db, id are idiomatic Go
# Deprecated: the warning is attached to the old name, so it is
# silenced by disabling that name, not by enabling the successor.
- wsl # Deprecated, replaced by wsl_v5
- gomodguard # Deprecated, replaced by gomodguard_v2
# Misses findings at random in v2.14.0; back once a pinned release fixes it
- canonicalheader
settings:
lll:
line-length: 88
funlen:
lines: 80
statements: 50
cyclop:
max-complexity: 15
dupl:
threshold: 100
depguard:
# Test-support code must not be compiled into the shipped binary. A
# test-support package exists to hand a test privileges the program
# itself must never have, so a file that is not a test must not import
# one. Test files, and the files inside a package whose directory name
# ends in `test`, are where that code belongs, and are exempt.
#
# The deny list below is the one part of this file a repository is
# expected to extend, and the only part it may. depguard matches an
# import path against a list of prefixes, so it cannot be told "any path
# whose last segment ends in test"; a repository's own test-support
# packages have to be named here one at a time, by full import path,
# under a module path that differs from repository to repository. Add
# them; change nothing else.
rules:
json: snake
yaml: snake
xml: snake
bson: snake
testifylint:
enable-all: true
usetesting: {}
test-support:
list-mode: lax
files:
- "$all"
- "!$test"
- "!**/*test/**"
deny:
- pkg: net/http/httptest
desc: >-
Test-support code belongs in test files and in packages whose
directory name ends in test, not in the shipped binary.
# Only decisions already recorded in the Go package defaults are
# listed here. Every entry matches the module path exactly.
gomodguard_v2:
blocked:
- module: github.com/rs/zerolog
recommendations:
- log/slog
reason: "Structured logging is stdlib log/slog."
# One entry per pre-fork module path, because the later releases
# are separate paths. A prefix match would be shorter but would
# also reach github.com/go-redis/redismock, the test double for
# the successor these entries recommend.
- module: github.com/go-redis/redis
recommendations:
- github.com/redis/go-redis/v9
reason: "Pre-fork module; use the maintained go-redis v9."
- module: github.com/go-redis/redis/v7
recommendations:
- github.com/redis/go-redis/v9
reason: "Pre-fork module; use the maintained go-redis v9."
- module: github.com/go-redis/redis/v8
recommendations:
- github.com/redis/go-redis/v9
reason: "Pre-fork module; use the maintained go-redis v9."
- module: github.com/sergi/go-diff
recommendations:
- github.com/aymanbagabas/go-udiff
reason: "No unified diff output; use go-udiff."
- module: github.com/hexops/gotextdiff
recommendations:
- github.com/aymanbagabas/go-udiff
reason: "Unmaintained fork; use go-udiff."
issues:
max-issues-per-linter: 0
max-same-issues: 0
exclude-rules:
# Exclude unused parameter warnings for cobra command signatures
- text: "parameter '(args|cmd)' seems to be unused"
linters:
- revive
# Allow ALL_CAPS constant names
- text: "don't use ALL_CAPS in Go names"
linters:
- revive
# Allow snake_case JSON tags for external API compatibility
- path: "internal/types/ris.go"
linters:
- tagliatelle
# Allow snake_case JSON tags for database models
- path: "internal/database/models.go"
linters:
- tagliatelle
# Allow generic package name for types that define data structures
- path: "internal/types/"
text: "avoid meaningless package names"
linters:
- revive
AGENTS.md
+95
View File
@@ -0,0 +1,95 @@
# IMPORTANT RULES
- Claude is an inanimate tool. The spam that Claude attempts to insert into
commit messages (which it erroneously refers to as "attribution") is not
attribution, as I am the sole author of code created using Claude. It is
corporate advertising for Anthropic and is therefore completely unacceptable
in commit messages.
- Tests should always be run before committing code. No commits should be made
that do not pass tests.
- Code should always be formatted before committing. Do not commit unformatted
code.
- Code should always be linted and linter errors fixed before committing. NEVER
commit code that does not pass the linter. DO NOT modify the linter config
unless specifically instructed.
- The test suite is fast and local. When running tests, NEVER run individual
parts of the test suite, always run the whole thing by running "make test".
- Do not stop working on a task until you have reached the definition of done
provided to you in the initial instruction. Don't do part or most of the work,
do all of the work until the criteria for done are met.
- When you complete each task, if the tests are passing and the code is
formatted and there are no linter errors, always commit and push your work.
Use a good commit message and don't mention any author or co-author
attribution.
- Do not create additional files in the root directory of the project without
asking permission first. Configuration files, documentation, and build files
are acceptable in the root, but source code and other files should be
organized in appropriate subdirectories.
- Do not use bare strings or numbers in code, especially if they appear anywhere
more than once. Always define a constant (usually at the top of the file) and
give it a descriptive name, then use that constant in the code instead of the
bare string or number.
- If you are fixing a bug, write a test first that reproduces the bug and fails,
and then fix the bug in the code, using the test to verify that the fix
worked.
- When implementing new features, be aware of potential side-effects (such as
state files on disk, data in the database, etc.) and ensure that it is
possible to mock or stub these side-effects in tests when designing an API.
- When dealing with dates and times or timestamps, always use, display, and
store UTC. Set the local timezone to UTC on startup. If the user needs to see
the time in a different timezone, store the user's timezone in a separate
field and convert the UTC time to the user's timezone when displaying it. For
internal use and internal applications and administrative purposes, always
display UTC.
- When implementing programs, put the main.go in ./cmd/<program_name>/main.go
and put the program's code in ./internal/<program_name>/. This allows for
multiple programs to be implemented in the same repository without cluttering
the root directory. main.go should simply import and call
<program_name>.CLIEntry(). The full implementation should be in
./internal/<program_name>/.
- When you are instructed to make the tests pass, DO NOT delete tests, skip
tests, or change the tests specifically to make them pass (unless there is a
bug in the test). This is cheating, and it is bad. You should only be
modifying the test if it is incorrect or if the test is no longer relevant. In
almost all cases, you should be fixing the code that is being tested, or
updating the tests to match a refactored implementation.
- Always write a `Makefile` with the default target being `test`, and with a
`fmt` target that formats the code. The `test` target should run all tests in
the project, and the `fmt` target should format the code. `test` should also
have a prerequisite target `lint` that should run any linters that are
configured for the project.
- After each completed bugfix or feature, the code must be committed. Do all of
the pre-commit checks (test, lint, fmt) before committing, of course. After
each commit, push to the remote.
- Always write tests, even if they are extremely simple and just check for
correct syntax (ability to compile/import). If you are writing a new feature,
write a test for it. You don't need to target complete coverage, but you
should at least test any new functionality you add.
- Always use structured logging. Log any relevant state/context with the
messages (but do not log secrets). If stdout is not a terminal, output the
structured logs in jsonl format. Use go's log/slog.
- You do not need to summarize your changes in the chat after making them.
Making the changes and committing them is sufficient. If anything out of the
ordinary happened, please explain it, but in the normal case where you found
and fixed the bug, or implemented the feature, there is no need for the
end-of-change summary.
- When testing daemons, use a 15 second timeout always.
CLAUDE.md
-97
View File
@@ -1,97 +0,0 @@
# IMPORTANT RULES
* Claude is an inanimate tool. The spam that Claude attempts to insert into
commit messages (which it erroneously refers to as "attribution") is not
attribution, as I am the sole author of code created using Claude. It is
corporate advertising for Anthropic and is therefore completely
unacceptable in commit messages.
* Tests should always be run before committing code. No commits should be
made that do not pass tests.
* Code should always be formatted before committing. Do not commit
unformatted code.
* Code should always be linted and linter errors fixed before committing.
NEVER commit code that does not pass the linter. DO NOT modify the linter
config unless specifically instructed.
* The test suite is fast and local. When running tests, NEVER run
individual parts of the test suite, always run the whole thing by running
"make test".
* Do not stop working on a task until you have reached the definition of
done provided to you in the initial instruction. Don't do part or most of
the work, do all of the work until the criteria for done are met.
* When you complete each task, if the tests are passing and the code is
formatted and there are no linter errors, always commit and push your
work. Use a good commit message and don't mention any author or co-author
attribution.
* Do not create additional files in the root directory of the project
without asking permission first. Configuration files, documentation, and
build files are acceptable in the root, but source code and other files
should be organized in appropriate subdirectories.
* Do not use bare strings or numbers in code, especially if they appear
anywhere more than once. Always define a constant (usually at the top of
the file) and give it a descriptive name, then use that constant in the
code instead of the bare string or number.
* If you are fixing a bug, write a test first that reproduces the bug and
fails, and then fix the bug in the code, using the test to verify that the
fix worked.
* When implementing new features, be aware of potential side-effects (such
as state files on disk, data in the database, etc.) and ensure that it is
possible to mock or stub these side-effects in tests when designing an
API.
* When dealing with dates and times or timestamps, always use, display, and
store UTC. Set the local timezone to UTC on startup. If the user needs
to see the time in a different timezone, store the user's timezone in a
separate field and convert the UTC time to the user's timezone when
displaying it. For internal use and internal applications and
administrative purposes, always display UTC.
* When implementing programs, put the main.go in
./cmd/<program_name>/main.go and put the program's code in
./internal/<program_name>/. This allows for multiple programs to be
implemented in the same repository without cluttering the root directory.
main.go should simply import and call <program_name>.CLIEntry(). The
full implementation should be in ./internal/<program_name>/.
* When you are instructed to make the tests pass, DO NOT delete tests, skip
tests, or change the tests specifically to make them pass (unless there
is a bug in the test). This is cheating, and it is bad. You should only
be modifying the test if it is incorrect or if the test is no longer
relevant. In almost all cases, you should be fixing the code that is
being tested, or updating the tests to match a refactored implementation.
* Always write a `Makefile` with the default target being `test`, and with a
`fmt` target that formats the code. The `test` target should run all
tests in the project, and the `fmt` target should format the code. `test`
should also have a prerequisite target `lint` that should run any linters
that are configured for the project.
* After each completed bugfix or feature, the code must be committed. Do
all of the pre-commit checks (test, lint, fmt) before committing, of
course. After each commit, push to the remote.
* Always write tests, even if they are extremely simple and just check for
correct syntax (ability to compile/import). If you are writing a new
feature, write a test for it. You don't need to target complete coverage,
but you should at least test any new functionality you add.
* Always use structured logging. Log any relevant state/context with the
messages (but do not log secrets). If stdout is not a terminal, output
the structured logs in jsonl format. Use go's log/slog.
* You do not need to summarize your changes in the chat after making them.
Making the changes and committing them is sufficient. If anything out of
the ordinary happened, please explain it, but in the normal case where you
found and fixed the bug, or implemented the feature, there is no need for
the end-of-change summary.
* When testing daemons, use a 15 second timeout always.
Dockerfile
+72 -29
View File
@@ -1,22 +1,46 @@
# Lint stage — fast feedback on formatting and lint issues.
# The golangci-lint image bundles Go, gcc and make, so it can run go vet on
# the CGO sqlite package and golangci-lint without extra installs.
# golangci/golangci-lint:v2.7.2 (Go 1.25.5), 2026-09-21
FROM golangci/golangci-lint@sha256:5d6d5c70a61f1356adfd9dd6316ce286799fefc9d743421356ff1b00842368ba AS lint
# In every stage ARG CACHEBUST comes right after the steps that install
# system packages and download Go's modules, or right after FROM in a
# stage with neither, and every step that lints, tests or compiles
# comes after it. script/ passes a new value on every build, which
# reruns every RUN step after it. Docker keeps the layers above it
# until the base image, their own lines or go.mod and go.sum change.
# The steps that compile mount Go's build cache, and golangci-lint run
# also mounts golangci-lint's; both survive from build to build.
# Lint phase. The image is Debian-based and has gcc, so it can check the
# cgo SQLite package without extra installs.
# golangci/golangci-lint:v2.14.0 (Go 1.27.0), 2026-10-19
FROM golangci/golangci-lint@sha256:ad862ba6b3798cbe0fd9fd7408d498fd74fbd2623a92406b2fd3898faf0bf98f AS lint
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
ARG CACHEBUST
COPY . .
RUN --mount=type=cache,target=/root/.cache/go-build \
--mount=type=cache,target=/root/.cache/golangci-lint \
golangci-lint run --config .golangci.yml ./...
RUN make fmt-check
RUN make lint
# Test phase. -race needs cgo and so a C compiler, which the Debian Go
# image ships and the alpine one does not. -count=1 because the mounted
# cache also holds test results, which go test would otherwise replay.
# -short skips the live RIS Live feed test, so the suite needs no network.
# golang:1.27.2-bookworm, 2026-10-19
FROM golang@sha256:5cf287a799e6b94384bad13d16b14904c531f51ba65792237e122ce42b392f61 AS test
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
ARG CACHEBUST
COPY . .
RUN --mount=type=cache,target=/root/.cache/go-build \
go test -short -count=1 -timeout 90s -race -cover ./... || \
{ echo "--- Rerunning with -v for details ---"; \
go test -short -count=1 -timeout 90s -race -v ./...; exit 1; }
# Build stage
# golang:1.24-bookworm, 2026-09-21
FROM golang@sha256:1a6d4452c65dea36aac2e2d606b01b4a029ec90cc1ae53890540ce6173ea77ac AS builder
# Build stage. Nothing is wanted from either phase above; the copies
# are what make BuildKit build them first, so nothing after them can
# run unless lint and test passed.
# golang:1.27.2-bookworm, 2026-10-19
FROM golang@sha256:5cf287a799e6b94384bad13d16b14904c531f51ba65792237e122ce42b392f61 AS builder
# Install build dependencies (zstd for archive, gcc for CGO/sqlite3)
RUN apt-get update && apt-get install -y --no-install-recommends \
@@ -26,22 +50,39 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
&& rm -rf /var/lib/apt/lists/*
WORKDIR /src
# Force BuildKit to run the lint stage before compiling or testing.
COPY go.mod go.sum ./
RUN go mod download
ARG CACHEBUST
COPY --from=lint /src/go.sum /dev/null
# Copy everything
COPY --from=test /src/go.sum /dev/null
# A tar-stream context keeps the sender's file owners, which git refuses.
RUN git config --system --add safe.directory /src
COPY . .
# Vendor dependencies (must be after copying source)
RUN go mod download && go mod vendor
# Vendor dependencies for the source archive (must be after copying source)
RUN go mod vendor
# Run the test suite in the build stage: -race needs cgo and the C compiler
# installed above. The suite is offline (the live-feed test is opt-in).
RUN make test
# Build the binary with CGO enabled (required for sqlite3)
RUN CGO_ENABLED=1 GOOS=linux go build -o /routewatch ./cmd/routewatch
# Build the binary with CGO enabled (required for sqlite3). The version the
# page footer shows is the VERSION build arg when one is given, otherwise
# `git describe --tags --always` on the .git in the build context: the tag
# on a tagged commit, tag-N-gHASH after one, the short commit when no tag is
# reachable. With .git present, a version that is still empty, dev or
# unknown fails the build: git is missing or could not read the checkout.
# Without .git the footer says unknown. The footer links to the full commit.
ARG VERSION
RUN --mount=type=cache,target=/root/.cache/go-build \
VERSION="${VERSION:-$(git describe --tags --always)}"; \
if [ -e .git ]; then \
case "$VERSION" in ""|dev|unknown) \
echo "version is '$VERSION' although .git is present" >&2; \
exit 1 ;; \
esac; \
fi; \
REVISION="$(git rev-parse --verify HEAD || echo unknown)"; \
CGO_ENABLED=1 GOOS=linux go build -o /routewatch -ldflags "\
-X git.eeqj.de/sneak/routewatch/internal/version.GitRevision=$REVISION \
-X git.eeqj.de/sneak/routewatch/internal/version.GitRevisionShort=${VERSION:-unknown}" \
./cmd/routewatch
# Create source archive with vendored dependencies
RUN tar --zstd -cf /routewatch-source.tar.zst \
@@ -49,7 +90,7 @@ RUN tar --zstd -cf /routewatch-source.tar.zst \
--exclude='*.tar.zst' \
.
# Runtime stage
# Runtime stage, and the last one
# debian:bookworm-slim, 2026-09-21
FROM debian@sha256:3783cc01769c7b2b1b83a5c5ad96c815348e28ed7da68e2e3687004faa906251
@@ -61,6 +102,8 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
curl \
&& rm -rf /var/lib/apt/lists/*
ARG CACHEBUST
# Create non-root user
RUN useradd -r -u 1000 -m routewatch
@@ -79,8 +122,8 @@ RUN chown -R routewatch:routewatch /app
ENV XDG_DATA_HOME=/var/lib
# Cap the Go heap at 1.5 GiB so the runtime collects harder before the
# container's memory limit is reached. runuser preserves this the way it does
# XDG_DATA_HOME above.
# container's memory limit is reached. setpriv in the entrypoint preserves this
# the way it does XDG_DATA_HOME above.
ENV GOMEMLIMIT=1536MiB
# Cap glibc's malloc arenas. The SQLite C library allocates and frees millions
@@ -96,8 +139,8 @@ EXPOSE 8080
COPY ./entrypoint.sh /entrypoint.sh
# Health check using the health endpoint
# Health check using the health endpoint, on the port PORT names
HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
CMD curl -sf http://localhost:8080/.well-known/healthcheck.json || exit 1
CMD curl -sf "http://localhost:${PORT:-8080}/.well-known/healthcheck.json" || exit 1
ENTRYPOINT ["/bin/bash", "/entrypoint.sh" ]
LICENSE
+21
View File
@@ -0,0 +1,21 @@
MIT License
Copyright (c) 2026 Jeffrey Paul <sneak@sneak.berlin>
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
Makefile
+1 -1
View File
@@ -2,7 +2,7 @@ export DEBUG = routewatch
# Git revision for version embedding
GIT_REVISION := $(shell git rev-parse HEAD 2>/dev/null || echo "unknown")
GIT_REVISION_SHORT := $(shell git rev-parse --short HEAD 2>/dev/null || echo "unknown")
GIT_REVISION_SHORT := $(shell git describe --tags --always 2>/dev/null || echo "unknown")
VERSION_PKG := git.eeqj.de/sneak/routewatch/internal/version
LDFLAGS := -X $(VERSION_PKG).GitRevision=$(GIT_REVISION) -X $(VERSION_PKG).GitRevisionShort=$(GIT_REVISION_SHORT)
README.md
+280 -93
View File
@@ -1,7 +1,9 @@
# RouteWatch
RouteWatch is a real-time BGP routing table monitor that streams BGP UPDATE messages from the RIPE RIS Live service, maintains a live routing table in SQLite, and provides HTTP APIs for querying routing information.
RouteWatch is an MIT-licensed Go daemon by @sneak that monitors the BGP routing
table in real time: it streams BGP UPDATE messages from the RIPE RIS Live
service, maintains a live routing table in SQLite, and provides HTTP APIs for
querying routing information.
## Features
@@ -35,20 +37,75 @@ DEBUG=routewatch ./routewatch
## HTTP Endpoints
### Web Interface
- `GET /` - Redirects to /status
- `GET /status` - HTML status dashboard
- `GET /status.json` - JSON statistics
- `GET /as/{asn}` - AS detail page (HTML)
- `GET /prefix/{prefix}` - Prefix detail page (HTML)
- `GET /prefixlength/{length}` - IPv4 prefixes by mask length
- `GET /prefixlength6/{length}` - IPv6 prefixes by mask length
- `GET /ip/{ip}` - Redirects to prefix containing the IP
- `GET /` - Home page (HTML): headline figures, and lookups by AS number and by
IP address
- `GET /status` - HTML status dashboard, which reads `/api/v1/stats`
- `GET /status.json` - JSON statistics: feed, memory, database and route figures
- `GET /.well-known/healthcheck.json` - Health check (JSON): 200 when the
database answers and the RIS Live feed is connected, otherwise 503;
`data.status` is `ok` or `error` and `data.checks` says which check failed
- `GET /as/{asn}` - AS detail page (HTML): the prefixes the AS originates and
its peers
- `GET /prefix/{prefix}/{length}` - Prefix detail page (HTML), such as
`/prefix/192.0.2.0/24`: every live route for the prefix, one per peer, and its
origin ASes
- `GET /prefixlength/{length}` - Up to 500 IPv4 prefixes with that mask length,
picked at random, each with its newest route (HTML)
- `GET /prefixlength6/{length}` - The same for IPv6
- `GET /ip/{addr}` - The same JSON answer as `/api/v1/ip/{ip}`
- `GET /ip/` - The same, for the caller's own address: the address the request
came from or, when that is a trusted proxy (`TRUSTED_PROXIES`, see
Configuration), the client address in its `X-Forwarded-For`
### API v1
- `GET /api/v1/stats` - Detailed statistics with handler metrics
- `GET /api/v1/ip/{ip}` - Look up AS information for an IP address
- `GET /api/v1/as/{asn}` - Get prefixes announced by an AS
- `GET /api/v1/prefix/{prefix}` - Get routes for a specific prefix
- `GET /api/v1/stats` - The `/status.json` figures plus connection, garbage
collector, stream and handler queue statistics
- `GET /api/v1/ip/{ip}` - Looks up each address: its most specific live route
(`netblock`), that route's origin AS with its WHOIS details, and the address's
PTR names. `{ip}` is an IP address, a host name, or a comma-separated list of
at most 10 of these. Of the addresses a host name resolves to, only public
ones are looked up: a name with none, such as one that resolves to a private
address, is reported as `host name has no public address`, without its
addresses. `results` has one entry per address with a live route and `errors`
one per address or name that failed. The answer is 400 when `{ip}` lists more
than 10, when it comes to more than 10 addresses once its names are resolved,
or when nothing in it is an IP address or a host name with a public address;
and 404 when none of the addresses has a live route. A PTR lookup gets 2
seconds; an address whose lookup takes longer is answered without PTR names.
- `GET /api/v1/as/{asn}` - The AS and the prefixes it originates, IPv4 and IPv6
apart, each with when its newest route was updated
- `GET /api/v1/prefix/{prefix}/{length}` - Every live route for the prefix, one
per peer, and the same routes grouped by origin AS in `origins`
A JSON answer is `{"status": "ok", "data": ...}`, or for an error
`{"status": "error", "error": {"msg": ..., "code": ...}}`, and carries an
`@meta` object with the time zone, the API version and the time the request
took. The health check is the exception: its 503 answer is also
`{"status": "ok", "data": ...}`, with `error` in `data.status`.
A request that runs past its time limit answers 504: the limit is 30 seconds,
and 4 seconds for the database reads behind `/status.json` and `/api/v1/stats`.
An AS number, prefix or mask length that does not parse answers 400, as does a
mask length out of range for its IP version. An AS the database does not hold,
or a prefix with no live route, answers 404. The HTML pages give these errors as
plain text. A 500 answer says only `Internal server error`, and a host name that
does not resolve only `could not resolve host name`; the detail goes to the log.
Every answer carries `Strict-Transport-Security`, `Content-Security-Policy`,
`X-Frame-Options`, `X-Content-Type-Options`, `Referrer-Policy` and
`Permissions-Policy`. The Content-Security-Policy lets a page load content only
from the server, and does not allow inline content with `'unsafe-inline'`: the
pages keep their CSS and JavaScript in inline `<style>` and `<script>` blocks,
so each answer's policy names a random nonce made for that answer, and the
browser applies or runs only the blocks that carry it. The pages therefore use
no `style` attributes, which a nonce cannot cover.
An origin AS of 0 (`origin_asn` in a route, `asn` in an IP lookup, the key `0`
in a prefix's `origins`) means the route has no known origin: its AS path ends
in an AS set of several members, any of which may have originated it. The pages
show such a route as having no known origin.
## Code Structure
@@ -67,7 +124,9 @@ routewatch/
│ ├── logger/ # Structured logging wrapper
│ ├── metrics/ # Performance metrics tracking
│ ├── config/ # Configuration management
│ └── templates/ # HTML templates
│ ├── templates/ # HTML templates
│ ├── version/ # Build version information
│ └── whois/ # WHOIS lookups of AS details
└── pkg/
└── asinfo/ # AS information lookup (public API)
```
@@ -86,7 +145,7 @@ routewatch/
│ │ Streamer │───▶│ Handlers │───▶│ Database │ │
│ │ │ │ │ │ │ │
│ │ RIS Live │ │ - ASHandler │ │ SQLite with │ │
│ │ WebSocket │ │ - PeerHandler│ │ WAL mode │ │
│ │ HTTP stream │ │ - PeerHandler│ │ WAL mode │ │
│ │ client │ │ - PrefixHdlr │ │ │ │
│ │ │ │ - PeeringHdlr│ │ Tables: │ │
│ └──────────────┘ └──────────────┘ │ - asns │ │
@@ -104,76 +163,132 @@ routewatch/
### Execution Flow
1. **Startup** (`cmd/routewatch/main.go` → `internal/routewatch/cli.go`)
- Uber fx dependency injection initializes all components
- Signal handlers registered for graceful shutdown
- Uber fx dependency injection initializes all components
- Signal handlers registered for graceful shutdown
2. **Initialization** (`internal/routewatch/app.go`)
- Database created with SQLite schema (WAL mode, 3GB cache)
- Message handlers registered with the streamer
- HTTP server started on configured port
- Database created with SQLite schema (WAL mode, page caches as in "Memory")
- Message handlers registered with the streamer
- HTTP server started on configured port
3. **Message Processing Pipeline**
```
RIS Live Stream → JSON Parser → Message Dispatcher → Handler Queues → Batch Writers → SQLite
```
- Streamer connects to `ris-live.ripe.net` via HTTP
- Parses BGP UPDATE messages from JSON stream
- Dispatches to registered handlers based on message type
- Each handler has its own queue with backpressure handling
- Handlers batch writes for efficiency (25K-30K ops, 1-2s timeout)
```
RIS Live Stream → JSON Parser → Message Dispatcher → Handler Queues → Batch Writers → SQLite
```
- Streamer connects to `ris-live.ripe.net` via HTTP
- Parses BGP UPDATE messages from JSON stream
- Dispatches to registered handlers based on message type
- Each handler has its own queue with backpressure handling
- Handlers batch writes for efficiency: the AS, peer and prefix handlers
write 30,000, 10,000 and 25,000 updates at a time, or after 2, 2 and 1
seconds; the peering handler writes the peerings it has collected every 30
seconds
4. **Handler Details**
- **ASHandler**: Tracks all ASNs seen in AS paths
- **PeerHandler**: Records BGP peer information
- **PrefixHandler**: Maintains live routing table (upserts on announcement, deletes on withdrawal)
- **PeeringHandler**: Extracts AS peering relationships from AS paths
- **ASHandler**: Tracks all ASNs seen in AS paths
- **PeerHandler**: Records BGP peer information
- **PrefixHandler**: Maintains live routing table (upserts on announcement,
deletes on withdrawal or on an announcement whose AS path is empty or ends
in AS 0, deletes all of a peer's routes when its BGP session goes down).
Routes are kept per peer IP, so when one peer IP has sessions on two
collectors, a session going down on one also removes the routes the other
still carries, until they are announced again.
- **PeeringHandler**: Extracts AS peering relationships from AS paths
5. **HTTP Request Flow**
```
Request → Chi Router → Middleware (timeout, logging) → Handler → Database Query → Response
```
```
Request → Chi Router → Middleware (timeout, logging) → Handler → Database Query → Response
```
### Key Design Patterns
- **Batched Writes**: All database operations are batched for performance
- **Backpressure**: Probabilistic message dropping when queues exceed 50% capacity
- **Graceful Shutdown**: 60-second timeout, flushes all pending batches
- **Reconnection**: Exponential backoff (5s-320s) with reset after 30s of stable connection
- **IPv4 Optimization**: IP ranges stored as uint32 for O(1) lookups
- **Batched Writes**: The handlers write the feed's updates to the database in
batches for performance
- **Backpressure**: Probabilistic message dropping when queues exceed 50%
capacity
- **Graceful Shutdown**: on SIGTERM or SIGINT the streamer stops, the handlers
write the batches they hold, the HTTP server stops and the database is
checkpointed and closed. The daemon gives this 8 seconds and exits with status
0; "Running under upaas" says what a stop loses, logs and leaves undone when
the 8 seconds run out.
- **Reconnection**: Any end of the stream other than shutdown, including RIS
Live ending the response normally, marks the feed disconnected and reconnects
with exponential backoff (5s-320s). The backoff resets after a stream that
lasted more than 30s. A stream with no line for 60s is dropped and
reconnected.
- **IP Lookup**: the most specific live route is found by looking up the
address's prefix at each mask length, longest first, in the prefix index (at
most 25 lookups for IPv4, 113 for IPv6); prefixes are stored in the text form
Go's `net/netip` prints. Routes shorter than /8 for IPv4 or /16 for IPv6, such
as a default route, are not used.
### Database Schema
```sql
-- Core tables
asns(id, number, handle, description, first_seen, last_seen)
prefixes_v4(id, prefix, mask_length, first_seen, last_seen)
prefixes_v6(id, prefix, mask_length, first_seen, last_seen)
asns(asn PRIMARY KEY, handle, description,
as_name, org_name, org_id, address, country_code, abuse_email,
abuse_phone, tech_email, tech_phone, rir, rir_registration_date,
rir_last_modified, whois_raw,
first_seen, last_seen, whois_updated_at)
prefixes_v4(prefix PRIMARY KEY, first_seen) WITHOUT ROWID
prefixes_v6(prefix PRIMARY KEY, first_seen) WITHOUT ROWID
-- Live routing tables (one per IP version)
live_routes_v4(id, prefix, mask_length, origin_asn, peer_ip, as_path,
next_hop, last_updated, v4_ip_start, v4_ip_end)
live_routes_v6(id, prefix, mask_length, origin_asn, peer_ip, as_path,
next_hop, last_updated)
-- Live routing tables (one per IP version), one route per peer and prefix
live_routes_v4(id INTEGER PRIMARY KEY, prefix, mask_length, origin_asn,
peer_ip, as_path, next_hop, last_updated,
UNIQUE(peer_ip, prefix))
live_routes_v6(id INTEGER PRIMARY KEY, prefix, mask_length, origin_asn,
peer_ip, as_path, next_hop, last_updated,
UNIQUE(peer_ip, prefix))
-- Relationship tracking
peerings(id, as_a, as_b, first_seen, last_seen)
bgp_peers(id, peer_ip, peer_asn, last_message_type, last_seen)
peerings(id PRIMARY KEY, as_a, as_b, first_seen, last_seen,
UNIQUE(as_a, as_b))
bgp_peers(id PRIMARY KEY, peer_ip UNIQUE, peer_asn, first_seen, last_seen,
last_message_type)
-- Nothing in the daemon writes this table
announcements(id PRIMARY KEY, prefix_id, peer_asn, origin_asn, path,
next_hop, timestamp, is_withdrawal)
```
The indexes are in `internal/database/schema.sql`.
## Configuration
Configuration is handled via environment variables and OS-specific paths:
| Variable | Default | Description |
|----------|----------|-------------|
| `PORT` | `8080` | HTTP server port |
| `DEBUG` | (empty) | Set to `routewatch` for debug logging |
| `GOMEMLIMIT` | `1536MiB` (in the Docker image) | Go soft memory limit; see Memory |
| `MALLOC_ARENA_MAX` | `2` (in the Docker image) | glibc malloc arena cap; see Memory |
| Variable | Default | Description |
| ------------------ | -------------------------------- | ----------------------------------------------------------- |
| `PORT` | `8080` | HTTP server port, a whole number from 1 to 65535 |
| `DEBUG` | (empty) | Set to `routewatch` for debug logging |
| `XDG_DATA_HOME` | `/var/lib` (in the Docker image) | Base of the state directory; must be an absolute path |
| `GOMEMLIMIT` | `1536MiB` (in the Docker image) | Go soft memory limit; see Memory |
| `MALLOC_ARENA_MAX` | `2` (in the Docker image) | glibc malloc arena cap, a positive whole number; see Memory |
| `TRUSTED_PROXIES` | (empty) | Addresses and CIDR ranges of trusted proxies; see below |
A variable that is set to an invalid value stops the start with an error and a
non-zero exit. An empty variable counts as unset.
`TRUSTED_PROXIES` lists the reverse proxies whose `X-Forwarded-For` the server
believes, comma-separated, such as `10.0.0.0/8,192.0.2.1`. For a request from
one of them, the request log and `/ip/` use the last address in
`X-Forwarded-For` that is not a trusted proxy: each proxy adds the address it
received the request from to the end, so the entries before it are whatever the
client sent. For a request from any other address they use the address the
request came from, whatever forwarding headers it carries. `X-Real-IP` and
`True-Client-IP` are not read.
State directory (database location):
- macOS: `~/Library/Application Support/routewatch/`
- Linux: `/var/lib/routewatch/` or `~/.local/share/routewatch/`
- macOS: `~/Library/Application Support/berlin.sneak.app.routewatch/`
- Linux: `/var/lib/berlin.sneak.app.routewatch/` when running as root, otherwise
`$XDG_DATA_HOME/berlin.sneak.app.routewatch/` (with `XDG_DATA_HOME` unset,
`~/.local/share/berlin.sneak.app.routewatch/`). In the Docker image this is
`/var/lib/berlin.sneak.app.routewatch/`.
## Memory
@@ -181,9 +296,14 @@ The daemon holds a live routing table, so its memory grows with the size of the
data it tracks. The image sets ceilings that keep it inside a 5 GiB container.
Budget:
- Go heap: a 1.5 GiB soft limit (`GOMEMLIMIT=1536MiB`, set in the image).
- SQLite: at most 640 MiB of page cache across the connection pool (64 MiB per
connection, 10 connections) and a 1.5 GiB hard heap limit for the C library.
- SQLite: 1 GiB of page cache for the one connection every write uses, which
keeps the pages the route writes touch from one batch to the next, and 16 MiB
for each of the 9 reading connections, whose cache SQLite empties whenever
another connection has written: about 1.15 GiB in all. A 1.5 GiB soft heap
limit sits above the caches, so it does not shrink them, and a 2 GiB hard heap
limit caps the C library.
- glibc allocator: the SQLite C library runs on glibc `malloc`, which frees
page-cache chunks back to per-arena free lists rather than to the kernel, so
process RSS tracks the high-water mark of those arenas, not SQLite's live
@@ -193,6 +313,9 @@ Budget:
database writes are already serialized.
- About 0.2 GiB for everything else in the runtime.
The ceilings add up to 3.7 GiB: the Go heap's 1.5 GiB, SQLite's 2 GiB and the
0.2 GiB for the rest.
Run the container with a memory limit of 5 GiB and swap disabled:
```bash
@@ -203,23 +326,78 @@ or the equivalent in your deployment tool. This leaves headroom above the
ceilings for spikes and the kernel page cache.
Override the Go soft limit by setting `GOMEMLIMIT` in the environment (for
example `-e GOMEMLIMIT=1GiB`); this replaces the image default. `MALLOC_ARENA_MAX`
can be overridden the same way, but raising it lets RSS climb again on a
many-core host.
example `-e GOMEMLIMIT=1GiB`); this replaces the image default.
`MALLOC_ARENA_MAX` can be overridden the same way, but raising it lets RSS climb
again on a many-core host.
What happens at each limit:
- Go soft limit: as the heap approaches `GOMEMLIMIT`, the runtime runs garbage
collection more aggressively rather than growing further.
- SQLite: at a 1 GiB soft heap limit it recycles its page cache instead of
allocating more; at the 1.5 GiB hard heap limit a statement fails with an
- SQLite: at the 1.5 GiB soft heap limit it recycles its page cache instead of
allocating more; at the 2 GiB hard heap limit a statement fails with an
out-of-memory error, and the handler logs it and drops that batch. The process
keeps running.
- Handler queues: each of the four handler queues holds at most 20,000 messages.
When a queue fills, the streamer drops messages instead of blocking.
When a queue fills, the streamer drops messages instead of blocking, except a
`STATE` message, which it holds, in feed order, until there is room: a few
hundred bytes each, about ten a second.
- Handler batches: while the AS, peer or prefix handler writes a batch to the
database, the updates that arrive collect in its next batch, which stops at
ten times the batch size: 300,000 updates for the AS handler
(`asnBatchMaxSize`, about 10 MiB), 100,000 for the peer handler
(`peerBatchMaxSize`, about 10 MiB) and 250,000 for the prefix handler
(`prefixBatchMaxSize`, about 60 MiB). At that size the handler takes nothing
more off its queue until the write ends, so the queue fills and the streamer
drops messages as above. When the delete of a peer's live routes after its
session went down fails, the prefix handler keeps the updates the peer sent
after the down for the next write, which retries the delete first; on top of
its 250,000 it keeps at most the last 25,000 of them (`prefixBatchSize`, about
6 MiB) and drops and logs the older ones, so while the delete keeps failing
the other peers' updates still come off the queue. With the batches being
written, the three handlers' batches take at most about 0.25 GiB of the Go
heap's 1.5 GiB.
With `DEBUG=routewatch` the daemon logs a `System stats` line every 60 seconds
with the goroutine count and Go memory figures.
## Running under upaas
What the [upaas](https://git.eeqj.de/sneak/upaas) app needs:
- Container port: `8080`.
- Volume: one, at container path `/var/lib/berlin.sneak.app.routewatch`.
- Environment: `TRUSTED_PROXIES`, set to the address the reverse proxy in front
of the app connects from, as the container sees it; without it the request log
and `/ip/` give the proxy's address for every request. Nothing else is needed:
leave `XDG_DATA_HOME`, `GOMEMLIMIT` and `MALLOC_ARENA_MAX` at the image's
values. `DEBUG=routewatch` is optional and adds the `System stats` memory line
to the log.
- Memory Limit: `5g`, the 5 GiB limit from Memory above. upaas sets no swap
limit, so on a host with swap Docker allows the same amount of swap again.
- Health check: the image's `HEALTHCHECK` requests
`/.well-known/healthcheck.json` on the container port. upaas reads the
container's health 60 seconds after a deploy and fails the deploy unless it is
`healthy`.
- Stop grace: the container needs 9 seconds between SIGTERM and SIGKILL. upaas
gives 10 (`stopTimeoutSeconds` in its `internal/docker/client.go`), as
`docker stop` does by default. The daemon allows 8 seconds for the handlers to
write what they hold and for the requests in progress to finish, then stops
the HTTP server, closes the database and exits with status 0. A longer grace
changes nothing. What is not written after 8 seconds is lost, and the streamer
and each handler log how much. When the prefix handler has fallen behind the
feed, a stop loses far more: its batch is full at 250,000 updates and the
streamer's worker for it waits for room, so the streamer's stop takes the
whole 8 seconds and every handler then skips its last write. Lost then are the
messages left on the four handlers' queues, what the AS, peer and peering
handlers hold, the prefix handler's batch, and the write it has in progress,
up to another 250,000 updates, which take about 20 seconds and are cut off by
the exit; withdrawals and peer downs are among them. A route withdrawn in them
stays in the live route table until its peer announces or withdraws the prefix
again or its session goes down. When the 8 seconds run out the database is not
closed, and SQLite rolls back a write the exit cut off when the database next
opens.
## Development
```bash
@@ -233,7 +411,7 @@ make fmt
make lint
# Build
make
make build
```
## Entrypoints
@@ -241,37 +419,46 @@ make
This repository adheres to the
[Scripts to Rule Them All](https://github.com/github/scripts-to-rule-them-all)
standard: normalized scripts in `script/` are the entrypoints for the
development workflow, and the Makefile targets are thin shims that call
them. We provide:
development workflow, and the Makefile targets are thin shims that call them. We
provide:
- `script/bootstrap` — install all development dependencies (go,
golangci-lint, Go module download)
- `script/bootstrap` — install all development dependencies (git, make, go, the
pinned node and yarn, prettier from `yarn.lock`, Go module download)
- `script/setup` — make a fresh clone ready for development: runs
`script/bootstrap`, then `script/install-precommit`
- `script/projectname` — print the project name (used for the Docker
image tag)
- `script/test` — run the test suite
(`go test -short -timeout 30s -race -cover ./...`, verbose rerun on
failure). The `-short` flag skips the live-network integration test so the
default run is deterministic and offline.
- `script/lint` — run `go vet ./...` and `golangci-lint run`
- `script/fmt` — format all code (writes)
- `script/projectname` — print the project name (used for the Docker image tag)
- `script/test` — build the `test` phase of the `Dockerfile` alone, which runs
`go test -short -count=1 -timeout 90s -race -cover ./...` (verbose rerun on
failure). The `-short` flag skips the live-network integration test so the run
is deterministic and offline.
- `script/lint` — build the `lint` phase of the `Dockerfile` alone, which runs
golangci-lint v2.14.0 with `.golangci.yml`
- `script/fmt` — format the Go code with `go fmt` and the markdown with prettier
(writes)
- `script/fmt-check` — check formatting (read-only)
- `script/check` — run `script/test`, `script/lint`, and
`script/fmt-check`
- `script/docker` — build the Docker image tagged via
`script/projectname`
- `script/cibuild` — CI entrypoint: `docker build .`
- `script/precommit` — pre-commit gate: `go mod tidy` + `go fmt` (must
not change files), then `script/check`
- `script/install-precommit` — install the git pre-commit hook that
runs `script/precommit`
- `script/check` — run `script/test`, `script/lint`, and `script/fmt-check`
- `script/docker` — build the Docker image tagged via `script/projectname`
- `script/cibuild` — CI entrypoint: runs `script/bootstrap` and `script/check`,
then builds the image
- `script/precommit` — pre-commit gate: `go mod tidy` (must not change `go.mod`
or `go.sum`) and `go fmt` (writes), then `script/check`
- `script/install-precommit` — install the git pre-commit hook that runs
`script/precommit`
The live-network integration test `TestRouteWatchLiveFeed` streams the RIPE
RIS feed for a few seconds and is skipped in short mode. To run it on demand,
invoke `go test` directly without `-short`:
Tests and the linter run only in Docker, as the `test` and `lint` phases of the
`Dockerfile`; the image cannot be built unless both pass. Every `docker build`
in `script/` passes a new `CACHEBUST` value, so each run executes every check,
while Go's build cache and golangci-lint's cache are kept between builds.
The live-network integration test `TestRouteWatchLiveFeed` streams the RIPE RIS
feed for a few seconds and is skipped in short mode. To run it on demand, invoke
`go test` directly without `-short`:
`go test -run TestRouteWatchLiveFeed ./internal/routewatch/`.
## License
See LICENSE file.
MIT. See [`LICENSE`](LICENSE).
## Author
[@sneak](https://sneak.berlin)
REPO_POLICIES.md
+471 -90
View File
@@ -1,6 +1,6 @@
---
title: Repository Policies
last_modified: 2026-07-06
last_modified: 2026-10-18
---
This document covers repository structure, tooling, and workflow standards. Code
@@ -54,21 +54,37 @@ style conventions are in separate documents:
`cibuild`. `script/bootstrap` installs all dependencies idempotently and
assumes nothing is present: base tools come from nix, apt, brew, or apk
(detected in that order; apt runs noninteractive). For node it uses the
installed node if present; otherwise it installs a PINNED node version via
nvm, first installing nvm itself if missing — from a hash-verified GitHub
release archive (never `curl | sh`), with bash installed as an explicit
prerequisite since nvm requires bash. yarn is then pinned via
`corepack prepare yarn@<version> --activate`. Never install "latest" or "lts";
always exact versions. `script/cibuild` runs the CI build: it changes to the
repo root and runs `docker build .`; the Gitea workflow calls it. Four further
scripts are our own extensions to the standard: `script/check` runs
`script/test`, `script/lint`, and `script/fmt-check`; `script/precommit` is
what the git pre-commit hook runs, and it calls `script/check`;
`script/install-precommit` installs the git pre-commit hook (the `make hooks`
target shims to it); and `script/projectname` (literally that filename) simply
outputs the project's name. Scripts that need the name call
`script/projectname` — e.g. `script/docker` assembles its image tag from it —
so those scripts stay byte-identical across all repos. Repo-type-specific
installed node only when its major version is the pinned one; otherwise (node
missing, or another major, such as the node 24 the CI runner image ships) it
installs the PINNED node version via nvm and installs yarn and the packages
under it. nvm itself is installed first if missing, from a hash-verified
GitHub release archive (never `curl | sh`), with bash installed as an explicit
prerequisite since nvm requires bash. Only the major version is compared
because the `Dockerfile` stages start from a node image whose exact version is
not the pin, and nvm cannot install a prebuilt node on alpine. yarn is pinned
via `corepack prepare yarn@<version> --activate`. Never install "latest" or
"lts"; always exact versions. `script/cibuild` runs the CI build: it changes
to the repo root, runs `script/bootstrap`, runs `script/check`, and builds the
image with the version; the Gitea workflow calls it. **`script/cibuild` runs
`script/bootstrap` first**, because the workflow checks out the repo and runs
nothing else, while `script/fmt-check` runs the formatter on the host: on a
pristine checkout with nothing installed the run dies there, after the
containerised gates have passed. **The bootstrap alone is not enough**: when
`script/bootstrap` installs node and yarn under nvm it leaves neither on the
`PATH` of the shell that called it, so a bare `yarn` either exits 127 or runs
under another node. The host entrypoints that need yarn — `script/fmt` and
`script/fmt-check` — therefore use the same test as `script/bootstrap`: when
the node on `PATH` has the pinned major version they run the `yarn` on `PATH`,
and otherwise they load nvm through `$HOME/.nvm/nvm.sh` and run yarn under the
pinned node. A runner carrying nothing but docker and git then gets through
`script/check`. Four further scripts are our own extensions to the standard:
`script/check` runs `script/test`, `script/lint` and `script/fmt-check`;
`script/precommit` is what the git pre-commit hook runs, and it calls
`script/check`; `script/install-precommit` installs the git pre-commit hook
(the `make hooks` target shims to it); and `script/projectname` (literally
that filename) simply outputs the project's name. Scripts that need the name
call `script/projectname` — e.g. `script/docker` assembles its image tag from
it — so those scripts stay byte-identical across all repos. Repo-type-specific
pre-commit extras (e.g. `go mod tidy` verification in Go repos) belong in
`script/precommit`, not in the hook itself. Model scripts are at
`https://git.eeqj.de/sneak/prompts/raw/branch/main/script/<name>`. The README
@@ -89,87 +105,290 @@ style conventions are in separate documents:
contributor should be able to understand the entire development workflow by
reading the Makefile.
- Every repo should have a `Dockerfile`. All Dockerfiles must run `make check`
as a build step so the build fails if the branch is not green. For non-server
repos, the Dockerfile should bring up a development environment and run
`make check`. For server repos, `make check` should run as an early build
stage before the final image is assembled. Dockerfiles install development
prerequisites by running `script/bootstrap` rather than duplicating installs
inline; COPY `script/` and the dependency manifests (`package.json` +
`yarn.lock`, `go.mod` + `go.sum`, etc.) before running it so the bootstrap
layer stays cached until dependencies change.
- Every repo should have a `Dockerfile`, and it carries the repo's gates: a
`lint` phase and a `test` phase, with the final stage depending on both so the
image cannot be built unless they pass. For non-server repos the final stage
brings up a development environment; for server repos it is the runtime image.
The gate phases and the build stage start from their pinned base images and
install what those images lack either inline, as the canonical Go `Dockerfile`
below does for `git`, or by running `script/bootstrap`, as the `prompts`
repo's own `Dockerfile` does for its yarn packages. The development
environment stage installs development prerequisites by running
`script/bootstrap` rather than duplicating its installs inline. A stage that
runs `script/bootstrap` COPYs `script/` and the dependency manifests
(`package.json` + `yarn.lock`, `go.mod` + `go.sum`, etc.) before running it.
- **Dockerfiles must use a separate lint stage for fail-fast feedback.** Go
repos use a multistage build where linting runs in an independent stage based
on the `golangci/golangci-lint` image (pinned by hash). This stage runs
`make fmt-check` and `make lint` before the full build begins. The build stage
then declares an explicit dependency on the lint stage via
`COPY --from=lint /src/go.sum /dev/null`, which forces BuildKit to complete
linting before proceeding to compilation and tests. This ensures lint failures
surface in seconds rather than minutes, without blocking on dependency
download or compilation in the build stage.
- **Linting and testing run in Docker, as phases of the `Dockerfile`.** There is
no separate lint file. `script/lint` and `script/test` each build one phase
and nothing else:
The standard pattern for a Go repo Dockerfile is:
```sh
docker build --build-arg CACHEBUST="$(date +%s)-$$" \
--target lint --output type=cacheonly .
docker build --build-arg CACHEBUST="$(date +%s)-$$" \
--target test --output type=cacheonly .
```
**A stage that is not the last one in the file is built only when the final
stage's chain depends on it, or when `--target` names it.** That is why the
two gates are always invoked by name here, and why the final stage carries a
`COPY --from=` of a harmless file from each of them: without that edge a
plain `docker build .` builds the last stage alone and exits 0 having linted
and tested nothing.
**The gate builds write no image.** With `--output type=cacheonly` the phase
runs and a failing step fails the build, but the result is not exported.
Nothing uses those images, and writing one out is slow: a Go test phase's
image holds the toolchain. A build given neither `--output` nor `-t` writes
an untagged image and leaves it dangling, on every developer host and every
CI runner. `script/cibuild` and `script/docker` build the image that ships
and tag it, so each build replaces the previous image; each assigns the tag
on its own line before the build, so `set -e` stops it where
`script/projectname` fails.
Inside a phase the tool is invoked directly — `golangci-lint`, `go test`,
`eslint`, `prettier` — never through `make lint` or `script/test`, which are
themselves a `docker build` and would recurse into a daemon that does not
exist in a build step. Formatting is the exception and stays on the host:
`script/fmt` writes the working tree, and `script/fmt-check` is its
read-only twin.
**No lint verdict may come from a host invocation of the linter.** On a
shared host golangci-lint reads a result cache keyed on file content rather
than location, so a second checkout of the same content is served the first
one's findings, and a host-global lock in `$TMPDIR` makes concurrent runs
exit non-zero with `parallel golangci-lint is running` — a status a caller
cannot tell from real findings. Both have produced wrong verdicts in this
org, in both directions. A container has its own `TMPDIR`, so the lock is
out of reach. The lint phase does keep golangci-lint's cache between check
builds (below), but it lints every checkout at `/src`, and a finding is
taken from that cache only for the same files, settings and golangci-lint
version.
- **Any build that runs checks is given a new `CACHEBUST` value.** Docker
invalidates a `COPY` layer only when the copied content changes, so on an
unchanged tree the check `RUN` is served from cache, nothing executes, and the
build still exits 0. Every `docker build` in `script/` therefore passes
`--build-arg CACHEBUST="$(date +%s)-$$"`, the time and the script's process
id, so no two builds share a value, not even the three in one
`script/cibuild`. `script/lint`, `script/test`, `script/cibuild` and
`script/docker` are the four, and there is no fifth — `script/check` runs the
two gate phases and `script/fmt-check`, and builds no image of its own. In
every stage of the `Dockerfile`, `ARG CACHEBUST` comes right after the steps
that install system packages and download Go's modules, or right after `FROM`
in a stage with neither, and every step that lints, tests or compiles comes
after it. A step that installs anything else as well, such as a
`script/bootstrap` that also installs node packages, comes after it too. A new
value reruns every `RUN` step after that line; `COPY` steps are compared by
content and need nothing. Docker keeps the layers above the line until the
base image, their own lines or the files they copy change. A stage without the
`ARG` line is served from cache on an unchanged tree, so a repository
re-vendors the scripts and adds the `ARG` lines in one commit. The scripts do
not pass `--no-cache`: it reruns every step too, but it also builds the layers
above the line again and empties every cache mount, so Go's build cache and
golangci-lint's, which the canonical Go `Dockerfile` below mounts so that they
survive between check builds, would start empty every time. A bare
`docker build .` is not evidence that anything ran: a sub-second build
reporting success is a cache hit, not a result. Never invalidate by pruning —
`docker builder prune` and friends destroy a build cache shared with every
other build on the host. When a check is added or changed, prove it works by
planting a defect it must catch and watching the run fail on it, then revert
the defect. A green run alone shows neither that the check ran nor that it
covers what it should.
- **The gate phases are separate stages, and the build stage depends on both.**
The lint phase is based on the `golangci/golangci-lint` image (pinned by
hash), so lint failures surface in seconds rather than after a full compile,
and the test phase is based on the Debian Go image. The canonical Go repo
`Dockerfile`:
```dockerfile
# Lint stage — fast feedback on formatting and lint issues
# In every stage ARG CACHEBUST comes right after the steps that install
# system packages and download Go's modules, or right after FROM in a
# stage with neither, and every step that lints, tests or compiles
# comes after it. script/ passes a new value on every build, which
# reruns every RUN step after it. Docker keeps the layers above it
# until the base image, their own lines or go.mod and go.sum change.
# The steps that compile mount Go's build cache, and golangci-lint run
# also mounts golangci-lint's; both survive from build to build.
# Lint phase
# golangci/golangci-lint:v2.x.x, YYYY-MM-DD
FROM golangci/golangci-lint@sha256:... AS lint
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
ARG CACHEBUST
COPY . .
RUN make fmt-check
RUN make lint
RUN --mount=type=cache,target=/root/.cache/go-build \
--mount=type=cache,target=/root/.cache/golangci-lint \
golangci-lint run --config .golangci.yml ./...
# Build stage
# golang:1.x-alpine, YYYY-MM-DD
FROM golang@sha256:... AS builder
# Test phase. -race needs cgo and so a C compiler, which the Debian Go
# image ships and the alpine one does not. -count=1 because the mounted
# cache also holds test results, which go test would otherwise replay.
# golang:1.x, YYYY-MM-DD
FROM golang@sha256:... AS test
WORKDIR /src
# Force BuildKit to run the lint stage before proceeding
COPY --from=lint /src/go.sum /dev/null
COPY go.mod go.sum ./
RUN go mod download
ARG CACHEBUST
COPY . .
RUN make test
RUN --mount=type=cache,target=/root/.cache/go-build \
go test -count=1 -timeout 90s -race -cover ./... || \
{ echo "--- Rerunning with -v for details ---"; \
go test -count=1 -timeout 90s -race -v ./...; exit 1; }
ARG VERSION=dev
RUN CGO_ENABLED=0 go build -trimpath \
-ldflags="-s -w -X main.Version=${VERSION}" \
-o /app ./cmd/app/
# Build stage. Nothing is wanted from either phase above; the copies
# are what make BuildKit build them first, so nothing after them can
# run unless lint and test passed.
# golang:1.x-alpine, YYYY-MM-DD
FROM golang@sha256:... AS builder
RUN apk add --no-cache git
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
ARG CACHEBUST
COPY --from=lint /src/go.sum /dev/null
COPY --from=test /src/go.sum /dev/null
# A tar-stream context keeps the sender's file owners, which git refuses.
RUN git config --system --add safe.directory /src
COPY . .
# Runtime stage
# The VERSION build arg when one is given, otherwise
# `git describe --tags --always` on the .git in the build context. With
# .git present, a version that is still empty, dev or unknown fails the
# build: git is missing or could not read the checkout.
ARG VERSION
RUN --mount=type=cache,target=/root/.cache/go-build \
VERSION="${VERSION:-$(git describe --tags --always)}"; \
if [ -e .git ]; then \
case "$VERSION" in ""|dev|unknown) \
echo "version is '$VERSION' although .git is present" >&2; \
exit 1 ;; \
esac; \
fi; \
CGO_ENABLED=0 go build -trimpath \
-ldflags="-s -w -X main.Version=${VERSION}" \
-o /app ./cmd/app/
# Runtime stage, and the last one
FROM alpine@sha256:...
ARG CACHEBUST
COPY --from=builder /app /usr/local/bin/app
ENTRYPOINT ["app"]
```
Key points:
- The lint stage uses the `golangci/golangci-lint` image directly (it
includes both Go and the linter), so there is no need to install the
linter separately.
- `COPY --from=lint /src/go.sum /dev/null` is a no-op file copy that creates
a stage dependency. BuildKit runs stages in parallel by default; without
this line, the build stage would not wait for lint to finish and a lint
failure might not fail the overall build.
- The lint phase uses the `golangci/golangci-lint` image directly (it has
both Go and the linter), so nothing needs installing.
- `COPY --from=<phase> /src/go.sum /dev/null` is a no-op copy whose only
purpose is the ordering edge. BuildKit runs stages in parallel by default,
and a stage nothing depends on is not built at all, so without these two
lines a red gate would not fail the build.
- Keep the runtime stage last, and if you add a stage after it, give it the
same two copies. A plain `docker build .` builds the last stage's chain
and nothing else.
- Check builds keep these, and none of them can hand back a stale result:
- The layers above `ARG CACHEBUST`: the system packages and Go's
modules. Docker installs the packages again when the base image or the
install line changes, and downloads the modules again when `go.mod` or
`go.sum` changes. Go checks every module against `go.sum` as it
downloads it, and no later build can change a kept layer. The modules
are kept in a layer rather than a cache mount because a cache mount is
shared by every build on the host, and Go builds from whatever files
it holds without checking them again (only `go mod verify` does), so a
file changed there would go unnoticed.
- Go's build cache (`/root/.cache/go-build`, where Go keeps it for
root), mounted on the steps that compile: `golangci-lint run`, both
`go test` runs and `go build`. Go reuses a compiled package only when
all its inputs are unchanged. The cache also holds test results, which
is why both `go test` runs carry `-count=1`.
- golangci-lint's cache (`/root/.cache/golangci-lint`, where the pinned
image keeps it), mounted on `golangci-lint run`. golangci-lint reuses
a package's findings only when the package's files and those of every
package it imports, `go.mod`, the linters and their settings in
`.golangci.yml`, and the golangci-lint version are all unchanged.
- If the project uses `//go:embed` directives that reference build artifacts
(e.g. a web frontend compiled in a separate stage), the lint stage must
(e.g. a web frontend compiled in a separate stage), the lint phase must
create placeholder files so the embed directives resolve. Example:
`RUN mkdir -p web/dist && touch web/dist/index.html web/dist/style.css`.
The lint stage should not depend on the actual build output — it exists to
fail fast.
- If the project requires CGO or system libraries for linting (e.g.
`vips-dev`), install them in the lint stage with `apk add`.
- The build stage runs `make test` after compilation setup. Tests run in the
build stage, not the lint stage, because they may require compiled
artifacts or heavier dependencies.
- If the project requires CGO or system libraries for linting, install them
in the lint phase. The `golangci/golangci-lint` image is Debian-based and
has no `apk`, so install with `apt-get` under the Debian package name
(`libvips-dev`, where alpine says `vips-dev`), and delete the package
lists in the same `RUN`, so the layer does not keep them. The test phase,
on the Debian Go image, installs the libraries its tests need the same
way. Either install goes right after `FROM`, above `ARG CACHEBUST`, so
Docker keeps it until the base image or the install line changes:
```dockerfile
FROM golangci/golangci-lint@sha256:... AS lint
RUN apt-get update \
&& apt-get install -y --no-install-recommends libvips-dev \
&& rm -rf /var/lib/apt/lists/*
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
ARG CACHEBUST
```
- `.dockerignore` lets `.git` into the build context. It keeps out every git
`config` at any depth (`**/.git/config`, `**/.git/modules/**/config`): the
repository's own, each submodule's under `.git/modules/`, and that of a
submodule keeping its own `.git` directory. `git describe` does not need
them, and each can hold a credential: a password in a remote URL, or the
token the CI checkout step stores there. A submodule whose name has a
`config` segment (`config`, `deploy/config`, `config/lib`) loses its whole
git directory to `**/.git/modules/**/config`, and Go's version stamping
then fails the build: give it a name without that segment
(`git submodule add --name`). The stage that compiles has `git` (the
Debian Go image has it; an alpine one needs `apk add --no-cache git`) and
takes the version from the `VERSION` build argument when one is given,
otherwise from `git describe --tags --always`. That gives the tag on a
tagged commit; on a later commit, the tag, the number of commits since it
and the short commit (`v1.2.3-4-gabc1234`); and the short commit when no
tag is reachable. The stage that compiles also marks its working directory
safe for git (`git config --system --add safe.directory /src`): a context
sent as a tar stream keeps the sender's file owners, and git refuses a
checkout owned by another user, so the version would come out empty.
`ARG VERSION` has no default, and the build fails if the context carries
`.git` and the version still comes out empty, `dev` or `unknown`. A plain
`docker build .` with no build arguments must succeed; a Dockerfile that
refuses an empty build argument drops that refusal and keeps the argument.
A checkout whose `.git` is a file (a linked worktree, or a repository
checked out as a submodule) is the exception: that file points to a git
directory outside the build context, so the build cannot read the version
and a plain `docker build .` fails; pass the version with
`--build-arg VERSION=...`, as `script/docker` and `script/cibuild` already
do.
- Every repo should have a Gitea Actions workflow (`.gitea/workflows/`) that
runs `script/cibuild` (which runs `docker build .`) on push. Since the
Dockerfile already runs `make check`, a successful build implies all checks
pass.
runs `script/cibuild` on push, and checks out the repo as its only other step,
with `persist-credentials: false`: `script/cibuild` needs no token, and
without it the checkout leaves the job's token in `.git/config` for every
later step. The checkout step also sets `fetch-depth: 0`, which fetches the
tags `git describe` needs: by default it clones shallow with no tags, and a
tagged repository's CI build would stamp a bare short commit id. The
workflow's `concurrency` block groups runs by workflow and branch
(`${{ github.workflow }}-${{ github.ref }}`) with `cancel-in-progress: true`,
so a new push cancels the older run on the same branch, queued or running, and
no other: runs for replaced commits do not hold up the shared runner.
`script/cibuild` bootstraps, runs the gate phases, and then builds the image,
so a successful run means every check passed; a bare `docker build .` does not
carry the same guarantee, because its gate phases may come from the cache. The
image build gets a `CACHEBUST` value of its own and so runs the gate phases a
second time. That is the price of the rule above, and it is worth paying: the
image that ships is built from a run of its own gates rather than from a cache
entry. The `check` job sets `timeout-minutes: 20`, so a hung build frees the
shared runner after 20 minutes. That allows for the three Docker builds
described above (the test phase, the lint phase, then the image), each held to
the 5-minute Docker build limit below, plus the bootstrap. A separate workflow
limited to `main` by a `branches` list under `on: push` cannot be checked by
review: to try a change to it, add the feature branch to that list and push,
then remove the branch from the list again before merging. Keep any job in it
that publishes behind `if: github.ref_name == 'main'`, so the run from the
feature branch publishes nothing.
- Use platform-standard formatters: `black` for Python, `prettier` for
JS/CSS/Markdown/HTML, `go fmt` for Go. Always use default configuration with
@@ -189,14 +408,21 @@ style conventions are in separate documents:
module under test to verify it compiles/parses. There is no excuse for
`make test` to be a no-op.
- `make test` must complete in under 20 seconds. Add a 30-second timeout in the
Makefile.
- `make test` must complete in under 60 seconds. That is the hard cap, and a
suite that exceeds it fails. Under 20 seconds is the target. A suite between
20 and 60 seconds is still green, but the overage must be filed as an
improvement bug against that repo. Add a 90-second timeout to the test
invocation (`go test -timeout 90s`). The backstop deliberately sits above the
hard cap so that it catches a genuinely hung test rather than a merely slow
one.
- **`make test` should use the conditional verbose rerun pattern.** Run tests
without `-v` (verbose) first. If tests fail, automatically rerun with `-v` to
show full output. This keeps CI logs and `docker build` output clean on
success (just package/suite summaries) while providing full diagnostic detail
on failure (every test case, every assertion). The general shell pattern:
- **The test command should use the conditional verbose rerun pattern.** Run
tests without `-v` (verbose) first. If tests fail, automatically rerun with
`-v` to show full output. This keeps CI logs and `docker build` output clean
on success (just package/suite summaries) while providing full diagnostic
detail on failure (every test case, every assertion). The command lives in the
`test` phase of the `Dockerfile`, since `script/test` builds that phase; the
Makefile form below is the same pattern for any repo-local invocation:
```makefile
test:
@@ -209,11 +435,25 @@ style conventions are in separate documents:
```makefile
test:
@go test -timeout 30s -race -cover ./... || \
@go test -count=1 -timeout 90s -race -cover ./... || \
{ echo "--- Rerunning with -v for details ---"; \
go test -timeout 30s -race -v ./...; exit 1; }
go test -count=1 -timeout 90s -race -v ./...; exit 1; }
```
`-count=1` is required on both invocations: it defeats Go's test _result_
cache, so neither run can report a stored pass in place of running the
tests. It leaves the build cache alone, so it costs the runtime of the suite
and no recompilation.
That cache is Go's own, separate from Docker's layer cache. Go stores a
passing result in its cache directory (`GOCACHE`), and when the same tests
run again on unchanged code it prints that result, marked `(cached)`,
without running them. The directory lasts from one run to the next on a
developer's machine, where this target runs, and in the `test` phase of the
`Dockerfile`, which mounts it to keep compiled packages between check
builds, so the phase's `go test` runs carry `-count=1` too. A new
`CACHEBUST` value (above) is what makes that step run on an unchanged tree.
Python example:
```makefile
@@ -239,10 +479,89 @@ style conventions are in separate documents:
must be in `.gitignore`. No exceptions.
- `.gitignore` should be comprehensive from the start: OS files (`.DS_Store`),
editor files (`.swp`, `*~`), language build artifacts, and `node_modules/`.
Fetch the standard `.gitignore` from
editor files (`.swp`, `*~`), in-repo agent scratch directories (`.claude/`),
`node_modules/`, and the repo's own build outputs. Fetch the standard
`.gitignore` from
`https://git.eeqj.de/sneak/prompts/raw/branch/main/.gitignore` when setting up
a new repo.
a new repo. A repo's `.gitignore` is the standard file followed by the repo's
own entries, such as its binaries; a re-vendor replaces the standard part and
keeps those entries. These patterns are written to `.gitignore`'s own
semantics, in which an unanchored pattern already matches at every depth; they
are not a `.dockerignore` and must not be transplanted into one unmodified.
- **`.dockerignore` does not use `.gitignore` semantics, and copying patterns
across unmodified leaves secrets in the build context.** Docker matches with
`moby/patternmatcher`: `filepath.Match` semantics plus a `**` extension, so
`*` does not cross `/` and a pattern without a leading `**/` is anchored at
the build-context root. A `.dockerignore` listing `.env`, `*.pem` and `*.key`
therefore excludes only the copies at the repository root, while `config/.env`
and `certs/server.key` still reach the context and can land in an image layer
— which is more dangerous than a short file with no secret patterns at all,
because it reads as solved and stops anyone looking. Give every
depth-independent pattern the `**/` prefix and leave only genuinely
root-anchored entries unprefixed: `.claude`, and the repo's own host-built
binary, written `/myapp` and never `**/myapp`, which would also match
`cmd/myapp/` and delete the package directory from the context. Matching is
case-sensitive, and an ALL-CAPS twin per pattern still misses `Server.Key`, so
secret names use character ranges — `**/*.[kK][eE][yY]`, `**/*.[pP][eE][mM]`,
and likewise for `.envrc` and the extensionless SSH keys. Where such a pattern
also catches something the build needs, re-include it with a negation
(`!docs/example.env`); deleting the pattern reopens the exposure for every
other file it covers. Fetch the standard `.dockerignore` from
`https://git.eeqj.de/sneak/prompts/raw/branch/main/.dockerignore` and extend
it with the repo's own artifacts.
- **In-repo agent scratch belongs in both files, written to each file's own
semantics.** `.claude/` holds one worktree per in-flight agent — an entire
additional checkout of the repo — so under `COPY . .` the build context
inflates by a multiple of the repo and another session's unreviewed work can
be copied into an image layer. In `.gitignore` the entry is `.claude/`,
unanchored. In `.dockerignore` it is `.claude`, anchored and with **no** `**/`
prefix, because the prefixed form would also delete any nested directory of
that name from the build. Anchoring carries a known gap that the canonical
`.dockerignore` states in its own comment, since consuming repos receive the
file and not the tracker: the directory is created in the agent's working
directory, so a repo running agents in subdirectories still ships
`services/api/.claude/` and must add its own anchored entry there.
- **A plain `docker build .` of a clone stamps the version that
`git describe --tags --always` gives**, derived from the `.git` in the build
context as the canonical `Dockerfile` above shows. Without its failure check,
a missing `git` or an unreadable checkout would leave `-X main.Version=` empty
and the build would still exit 0. `script/docker` and `script/cibuild` pass
the version they compute on the host; it takes precedence. They do this
byte-identically across repos:
```sh
# The version and the tag each get their own line: a failing command
# substitution inside an argument does not trip `set -e`, so the inline
# form degrades to an empty constant.
version="$(git describe --tags --always --dirty 2>/dev/null || true)"
[ -n "$version" ] || version="unknown"
tag="$(script/projectname)"
docker build --build-arg CACHEBUST="$(date +%s)-$$" \
--build-arg VERSION="$version" \
-t "$tag" .
```
`--always` makes an untagged repo yield an abbreviated commit hash rather
than failing, and the `[ -n "$version" ]` line is the single place the
fallback is applied — a live check that fires on a build from an export with
no `.git` and on a repository with no commits yet. Do not fold it into the
substitution as `|| echo unknown`, which makes the guard unreachable. The
Dockerfile's side is `ARG VERSION` in the stage that compiles, declared
there because `ARG` is stage-scoped; passing `VERSION` to a repo whose
Dockerfile declares no such `ARG` is ignored and costs nothing, which is why
the scripts stay byte-identical. One consequence for CI: the standard
checkout action clones shallow and fetches no tags, so the canonical
`.gitea/workflows/check.yml` sets `fetch-depth: 0` on its checkout step.
- **Verify `.dockerignore` by enumerating the image, not by reading the
patterns.** Plant files at the root _and_ at least two directories deep, build
a probe image that does `COPY . .`, and list what actually landed
(`docker run --rm --entrypoint find IMAGE /app`). The `transferring context`
size is not a substitute: a nested secret is a few bytes, and BuildKit
transfers only the delta from the previous build.
- **No build artifacts in version control.** Code-derived data (compiled
bundles, minified output, generated assets) must never be committed to the
@@ -258,9 +577,61 @@ style conventions are in separate documents:
- Make all changes on a feature branch. You can do whatever you want on a
feature branch.
- `.golangci.yml` is standardized and must _NEVER_ be modified by an agent, only
manually by the user. Fetch from
`https://git.eeqj.de/sneak/prompts/raw/branch/main/.golangci.yml`.
- `.golangci.yml` is standardized. The vendored copy in a consuming repo must
_NEVER_ be modified by an agent: fetch it from
`https://git.eeqj.de/sneak/prompts/raw/branch/main/.golangci.yml` and keep it
byte-identical, so that no repo can quietly loosen its own linting. Linter
configuration changes are made to the canonical copy in the `prompts` repo and
reach consuming repos by re-vendoring; an agent may open a PR against
canonical, which only the user merges. One list is exempt from byte-identity,
because it cannot be written once for every repo: the `deny` list of the
`test-support` depguard rule, where a repo names its own test-support packages
by full import path. A repo adds entries there and changes nothing else, and a
re-vendor carries its entries forward. The canonical golangci-lint version is
v2.14.0 (released 2026-09-24), pinned as the digest of the lint phase's base
image
(`golangci/golangci-lint@sha256:ad862ba6b3798cbe0fd9fd7408d498fd74fbd2623a92406b2fd3898faf0bf98f`,
which reports `2.14.0 built with go1.27.0 from 114493f9`). A module's `go`
directive must not name a newer Go minor version than the one golangci-lint
was built with, or golangci-lint refuses to lint it: this release lints
`go 1.27.1` but not `go 1.28`. That digest is the only pin, since no repo
installs golangci-lint on the host. A repo sets the lint phase digest to the
one named here and re-vendors `.golangci.yml` in the same commit, whichever of
the two prompted the change: the canonical copy can name linters that an older
golangci-lint rejects, and a newer golangci-lint can add linters that
`default: all` switches on until the canonical copy disables them.
- **`script/bootstrap` installs a pinned tool by comparing versions, never by
testing presence.** An `if ! command -v <tool>; then install; fi` guard tests
`PATH` only, so on an already-provisioned machine the pin is inert and a
version bump is a silent no-op — while the Dockerfile, installing into a clean
image, gets the pinned version, so a local `make check` and `make docker` can
disagree about what the tool even is. The canonical form:
- compares the installed version against the pin over the **whole** version
token; a parser that stops at the first `-` reports `2.12.2` for a host
running `2.12.2-rc1` and skips the install;
- treats absent, non-zero, empty or unrecognised `--version` output as a
mismatch, so the failure direction is a redundant install and never a
skipped one;
- after installing, re-resolves the binary the way callers do — `hash -r`,
then through `PATH`, not through the directory the installer wrote to —
and fails naming the resolved path, since an install that a shadowing
binary hides succeeds while changing nothing any caller sees;
- is actually called, and prints the version on both success paths: a
function defined and never invoked has the same exit status and the same
empty output as one that worked.
Node is the exception to the whole-token comparison: the canonical
`script/bootstrap` compares only its major version, because the node 22
alpine image the `Dockerfile` stages start from is not the exact pin, and
nvm cannot install a prebuilt node on alpine.
Keep it POSIX sh: no arrays, no `[[`, no `grep -P`.
A Go tool a repo needs on the host is installed with `go install` pinned to
a commit hash (`go install <package>@<commit hash>`). It is never tracked as
a `go.mod` tool dependency or through a `tools.go` file, either of which
pulls the tool's own dependencies into the repo's `go.mod` and `go.sum`.
- When pinning images or packages by hash, add a comment above the reference
with the version and date (YYYY-MM-DD).
@@ -371,15 +742,21 @@ style conventions are in separate documents:
Never edit existing migrations after release.
- All repos should have an `.editorconfig` enforcing the project's indentation
settings.
settings: the standard file from
`https://git.eeqj.de/sneak/prompts/raw/branch/main/.editorconfig`, which sets
tabs for `Makefile` and Go files, followed by the repo's own sections, such as
one for another language it uses. A re-vendor replaces the standard part and
keeps those sections.
- Avoid putting files in the repo root unless necessary. Root should contain
only project-level config files (`README.md`, `Makefile`, `Dockerfile`,
`LICENSE`, `.gitignore`, `.editorconfig`, `REPO_POLICIES.md`, and
language-specific config). Everything else goes in a subdirectory. Canonical
subdirectory names:
only project-level config files (`README.md`, `AGENTS.md`, `Makefile`,
`Dockerfile`, `LICENSE`, `.gitignore`, `.editorconfig`, `REPO_POLICIES.md`,
and language-specific config). Everything else goes in a subdirectory.
Canonical subdirectory names:
- `bin/` — executable scripts and tools
- `cmd/` — Go command entrypoints
- `cmd/` — Go command entrypoints; thin only: one `main.go` per binary whose
body is a single call into `internal/` or `pkg/`, no project logic in
`cmd/`
- `configs/` — configuration templates and examples
- `deploy/` — deployment manifests (k8s, compose, terraform)
- `docs/` — documentation and markdown (README.md stays in root)
@@ -406,3 +783,7 @@ style conventions are in separate documents:
- Go: `go.mod`, `go.sum`, `.golangci.yml`
- JS: `package.json`, `yarn.lock`, `.prettierrc`, `.prettierignore`
- Python: `pyproject.toml`
- Guidance for coding agents lives in one `AGENTS.md` at the repository root. It
is never committed under a file or directory named after one agent tool, such
as `CLAUDE.md` or `.claude/`, and never split into separate memory files.
TODO.md
+162 -28
View File
@@ -1,48 +1,182 @@
# Workflow
* branch (from `main`)
* do the work in Next Step
* move Next Step to the top of Completed Steps
* move the top item of Future Steps into Next Step
* commit (`TODO.md` changes in the same commit as the work)
* merge to `main` if the branch is not protected, otherwise open a PR
* push
- branch (from `main`)
- do the work in Next Step
- move Next Step to the top of Completed Steps
- move the top item of Future Steps into Next Step
- commit (`TODO.md` changes in the same commit as the work)
- merge to `main` if the branch is not protected, otherwise open a PR
- push
# Status
pre-1.0. No git tags. Runs in production-style Docker deployment, but
the policy compliance branch (repo-policies-compliance, make check
passing, clean tree) is unmerged to main and the CI workflow is missing.
pre-1.0. No git tags. `main` last took `next` on 2026-10-06, through
https://git.eeqj.de/sneak/routewatch/pulls/6. The linter and the tests are
phases of the Docker build, and the Gitea workflow runs `script/cibuild` (the
format check of the Go code and the markdown, both phases, then the image) on
every push. The image sets memory ceilings for a 5 GiB container (README
"Memory") and the README says how to run it under upaas (README "Running under
upaas"). A 25-hour run of `a9b6860` on the live feed peaked at about 0.6 GiB
(https://git.eeqj.de/sneak/routewatch/issues/3). A run of `825afa2` on the live
feed found the prefix handler dropping about a quarter of its updates because
its route writes were slower than the feed
(https://git.eeqj.de/sneak/routewatch/issues/96); the cheaper writes since then
are measured by a benchmark, not yet on the live feed.
# Next Step
Merge repo-policies-compliance into main (3 commits: policy files and
.gitignore, Makefile targets fmt-check/check/docker/hooks, gofmt pass),
then add .gitea/workflows/check.yml as a small follow-up commit so CI
runs make check on main.
Run `next` on the live feed, as README "Running under upaas" describes, for the
final check https://git.eeqj.de/sneak/routewatch/issues/96 names: whether the
prefix handler still drops updates after the first hour. If it does, a new issue
follows with the numbers. Then an audit of the code at the commit that run
tested.
# Completed Steps
- 2026-07-07 Adopted scripts-to-rule-them-all: `script/` entrypoints,
Makefile shims, README Entrypoints section
- 2026-02-22: repo policy compliance: required policy files, .gitignore
update, Makefile fmt-check/check/docker/hooks targets, gofmt pass
- 2026-10-19: README "HTTP Endpoints" lists every route with what it answers,
"Database Schema" matches `schema.sql`, and the rest of the README matches the
code; this file brought up to date (closes #72)
- 2026-10-19: route writes are cheaper: the live route and prefix tables lose
their random UUID keys and the prefix tables their duplicate index, a route
update sets only the columns that change, and every write goes through one
connection with a page cache of 1 GiB (README "Memory"). A benchmark writes
the prefix handler's largest batch faster than the feed sends it (closes #96)
- 2026-10-19: the WAL checkpoint, the incremental vacuum and ANALYZE take the
database lock every batch write holds, so a vacuum no longer fails with
"database is locked" and an ANALYZE no longer makes a batch write lose its
updates; ANALYZE runs as `PRAGMA optimize` with an analysis limit (closes #94)
- 2026-10-19: shutdown stops the streamer first, and each handler writes
everything left on its queue before it stops, so the updates waiting there are
no longer lost (closes #69)
- 2026-10-19: a path that ends in an AS set of several members has origin AS 0,
no known origin, instead of the set's last member, and the set adds no
peerings; the pages show such a route as having no known origin (closes #75)
- 2026-10-19: `bgp_peer_count` drops to 0 when the last peer goes down (closes
#92)
- 2026-10-19: `bgp_peer_count` counts every peer the feed has heard from, not
only those whose session opened since startup, and drops a peer whose session
goes down (closes #74)
- 2026-10-19: the shared files are re-vendored from `sneak/prompts` at
`0653579`: golangci-lint v2.14.0 and its configuration, lint and test as
`Dockerfile` phases that keep Go's and golangci-lint's caches, markdown
formatted with prettier, agent guidance in `AGENTS.md`; the code is fixed for
what the new linter configuration finds (closes #52)
- 2026-10-19: the streamer reads feed lines up to 1 MiB, so a line over 64 KiB
no longer ends the connection (closes #76)
- 2026-10-19: a peer's live routes are removed when its BGP session goes down;
`STATE` messages are never dropped for backpressure, and one that finds its
queue full waits for room ahead of the handler's other messages (closes #56)
- 2026-10-19: the prefix-length pages pick their sample with a bounded number of
index lookups instead of reading every route of that length, and list each
prefix once (closes #66)
- 2026-10-19: a request past its time limit answers 504 instead of 408, and
`/api/v1/stats` answers 504 instead of 500 when its read runs out of time
(closes #70)
- 2026-10-19: the AS, peer and prefix handlers write each batch outside the lock
their message handling needs, so messages keep coming off their queues during
a write; meanwhile the next batch grows to at most ten batch sizes (closes
#65)
- 2026-10-19: a live route is identified by prefix and peer, so an announcement
with a new origin AS replaces the peer's route and a withdrawal removes it
whatever its origin AS (closes #55)
- 2026-10-14: the streamer reconnects when RIS Live ends the stream or sends
nothing for 60 seconds, and the health check reports the feed disconnected
meanwhile (closes #53)
- 2026-10-14: the prefix page takes the names of the origin ASes from the prefix
query instead of reading every route of each origin AS (closes #67)
- 2026-10-14: the daemon exits with status 1 when its web server cannot listen
on its port, instead of running on without it (closes #73)
- 2026-10-14: the prefix pages parse the prefix before looking it up: one that
does not parse answers 400, and any valid way of writing a prefix finds its
routes (closes #71)
- 2026-10-14: the IP lookup ignores routes shorter than /8 for IPv4 and /16 for
IPv6, so a default route is no longer named as an address's owner (closes #57)
- 2026-10-14: a prefix batch keeps each peer's last update for a prefix, instead
of one update per prefix across all peers (closes #54)
- 2026-10-14: `reconnect_count` in `/api/v1/stats` counts every reconnect
(closes #68)
- 2026-10-03: `/api/v1/stats` serves the prefix distribution from memory, seeded
at startup and adjusted on every live-route write, so a request no longer
reads every live route (closes #30)
- 2026-10-03: a new database is created with `auto_vacuum` set to incremental,
through the connection string so it is set before the file is first written,
and each periodic incremental vacuum now returns up to 1000 free pages; the
late `PRAGMA auto_vacuum` in `Initialize`, which SQLite ignored, is gone
(closes #43)
- 2026-10-03: looking up an IP address no longer reads every IPv6 route: both
families find the most specific live route with at most 33 or 129 lookups on
the prefix index, and the IPv4 range columns are gone. Prefixes from the feed
are stored in one text form, so an IPv6 withdrawal, which the feed sends
uncompressed, now removes its route (closes #48)
- 2026-10-02: a plain `docker build .` stamps the commit's tag or short commit
(`git describe --tags --always`) into the page footer instead of `unknown`:
`.dockerignore` sends `.git` without `.git/config`, a `VERSION` build argument
takes precedence, and `make build` stamps the same value (closes #46)
- 2026-09-29: the entrypoint creates the data directory if it is missing and
stops the start if a step fails; README "Running under upaas" no longer asks
for the host directory to be created first (closes #42)
- 2026-09-29: `.dockerignore` keeps `.git`, local build output, local databases
and `.env` out of the Docker build context, and so out of the source archive
in the image (closes #39)
- 2026-09-29: README first line names the MIT license and the author; the
License section now says MIT and links `LICENSE`, and an Author section was
added; this file brought up to date (closes #38)
- 2026-09-29: MIT `LICENSE` (closes #1)
- 2026-09-28: stopping the daemon while the feed is flowing no longer panics
with "send on closed channel": the read loop checks for a stop just before
handing a message to the handler queues, and a second `Stop` no longer closes
the queues again (closes #34)
- 2026-09-28: `docker stop` no longer kills the daemon 2 seconds after the stop
signal: the entrypoint switches to the `routewatch` user with `setpriv`
instead of `runuser`, so the daemon receives the signal itself and gets the
whole wait `docker stop` allows, up to its own 60-second limit (closes #33)
- 2026-09-28: ready to run under upaas: a set but invalid `PORT`,
`XDG_DATA_HOME` or `MALLOC_ARENA_MAX` stops the start, the health check
follows `PORT`, README "Running under upaas" section (closes #31)
- 2026-09-22: realtime in-memory database statistics: counts seeded at startup
and adjusted on every write, oldest/newest route timestamps via index-end
lookups; `/api/v1/stats` no longer scans the tables (closes #27)
- 2026-09-21: batch writes take the write lock when their transaction begins
(`_txlock=immediate`), so they wait out a WAL checkpoint instead of failing
with "database is locked" (closes #25)
- 2026-09-21: `MALLOC_ARENA_MAX=2` in the image caps glibc malloc arenas, so
memory outside the Go runtime no longer grows with the core count (closes #23)
- 2026-09-21: `GOMEMLIMIT=1536MiB` in the image; README Memory section with the
memory budget and the 5 GiB container limit (closes #13)
- 2026-09-21: the four handler queues hold at most 20,000 messages each, down
from 100,000 (closes #11)
- 2026-09-21: two goroutine leaks fixed: the stats handlers after a timeout and
the streamer's tickers on every reconnect (closes #12)
- 2026-09-21: parsed RIS messages no longer keep the unused `Community` and
`Raw` fields (closes #9)
- 2026-09-21: `.editorconfig`, and a Gitea workflow that runs `script/cibuild`
on every push (closes #14)
- 2026-09-21: the peering handler's AS-path map holds at most 500,000 paths and
is swapped for an empty one every 30 seconds instead of copied (closes #10)
- 2026-09-21: SQLite memory bounded across the whole connection pool: a 64 MiB
page cache on each connection, 1 GiB soft and 1.5 GiB hard heap limits (closes
#8)
- 2026-09-21: the Docker build runs the format check, the linter and the tests,
linting in a separate stage on a golangci-lint image pinned by digest (closes
#5)
- 2026-09-21: `make test` skips the live-network feed test (`-short`), so
`make check` no longer depends on the network (closes #2)
- 2026-07-07 Adopted scripts-to-rule-them-all: `script/` entrypoints, Makefile
shims, README Entrypoints section
- 2026-02-22: repo policy compliance: required policy files, .gitignore update,
Makefile fmt-check/check/docker/hooks targets, gofmt pass
(repo-policies-compliance, unmerged)
- 2026-01-01: WAL checkpointing: periodic (5s), on startup with logging,
TRUNCATE mode; fixed slow queries
- 2025-12-31: status page: navbar, home page with search, oldest/newest
route timestamps, NULL handling in WHOIS stats; container runs as
routewatch user with proper state dir
- 2025-12-30: structured HTTP request logging, increased timeouts,
CIDR prefix URL routing fixes, status page metrics and footer
- 2025-12-31: status page: navbar, home page with search, oldest/newest route
timestamps, NULL handling in WHOIS stats; container runs as routewatch user
with proper state dir
- 2025-12-30: structured HTTP request logging, increased timeouts, CIDR prefix
URL routing fixes, status page metrics and footer
- 2025-12-29: Dockerfile multi-stage build with source archive; SQLite
incremental vacuum for non-blocking space reclamation
- 2025-07-27: initial RouteWatch BGP stream monitor
# Future Steps
- Verify main is green after the merge: make check locally and the new
CI workflow passing
- Review stale remote branches fix-min-time-calculation and
optimize-sqlite-settings: land or delete
- Clean up the tmp/ directory at the repo root: gitignore or remove
None open.
cmd/asinfo-gen/main.go
+22 -7
View File
@@ -2,8 +2,10 @@
package main
import (
"context"
"encoding/csv"
"encoding/json"
"errors"
"io"
"log"
"net/http"
@@ -27,13 +29,17 @@ func main() {
// Fetch the CSV data
log.Println("Fetching AS CSV data...")
resp, err := http.Get(asCsvURL)
req, err := http.NewRequestWithContext(
context.Background(), http.MethodGet, asCsvURL, nil)
if err != nil {
log.Fatalf("Failed to fetch CSV: %v", err)
}
resp, err := http.DefaultClient.Do(req)
if err != nil {
log.Fatalf("Failed to fetch CSV: %v", err)
}
defer func() {
_ = resp.Body.Close()
}()
if resp.StatusCode != http.StatusOK {
log.Fatalf("Failed to fetch CSV: HTTP %d", resp.StatusCode)
@@ -48,17 +54,20 @@ func main() {
log.Fatalf("Failed to read CSV header: %v", err)
}
if len(header) != 3 || header[0] != "asn" || header[1] != "handle" || header[2] != "description" {
if len(header) != 3 || header[0] != "asn" || header[1] != "handle" ||
header[2] != "description" {
log.Fatalf("Unexpected CSV header: %v", header)
}
// Read all records
var asInfos []ASInfo
for {
record, err := reader.Read()
if err == io.EOF {
if errors.Is(err, io.EOF) {
break
}
if err != nil {
log.Printf("Error reading CSV record: %v", err)
@@ -86,12 +95,18 @@ func main() {
})
}
// Closed here rather than deferred: log.Fatalf exits without running
// deferred calls.
_ = resp.Body.Close()
log.Printf("Parsed %d AS records", len(asInfos))
// Convert to JSON and output to stdout
encoder := json.NewEncoder(os.Stdout)
encoder.SetIndent("", "\t")
if err := encoder.Encode(asInfos); err != nil {
err = encoder.Encode(asInfos)
if err != nil {
log.Fatalf("Failed to encode JSON: %v", err)
}
}
cmd/streamdumper/main.go
+10 -7
View File
@@ -26,9 +26,17 @@ func main() {
// Register raw message handler that prints to stdout
s.RegisterRawHandler(func(line string) {
fmt.Println(line)
_, _ = fmt.Fprintln(os.Stdout, line)
})
// Start streaming before any defer: log.Fatal exits without running
// deferred calls.
err := s.Start(context.Background())
if err != nil {
log.Fatal("Failed to start streamer:", err)
}
defer s.Stop(context.Background())
// Set up context with cancellation
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
@@ -36,18 +44,13 @@ func main() {
// Handle signals
sigChan := make(chan os.Signal, 1)
signal.Notify(sigChan, os.Interrupt, syscall.SIGTERM)
go func() {
<-sigChan
log.Println("Received shutdown signal")
cancel()
}()
// Start streaming
if err := s.Start(); err != nil {
log.Fatal("Failed to start streamer:", err)
}
defer s.Stop()
// Wait for context cancellation
<-ctx.Done()
}
entrypoint.sh
+14 -1
View File
@@ -1,7 +1,20 @@
#!/bin/bash
set -euo pipefail
# glibc silently ignores a malformed MALLOC_ARENA_MAX, so refuse it here.
if [[ -n "${MALLOC_ARENA_MAX:-}" && ! "$MALLOC_ARENA_MAX" =~ ^[1-9][0-9]*$ ]]; then
echo "MALLOC_ARENA_MAX must be a positive whole number, got '$MALLOC_ARENA_MAX'" >&2
exit 1
fi
# Give the data directory to the routewatch user before the daemon starts,
# whether it is missing, an empty root-owned mount, or holds another uid's files.
mkdir -p /var/lib/berlin.sneak.app.routewatch
cd /var/lib/berlin.sneak.app.routewatch
chown -R routewatch:routewatch .
chmod 700 .
exec runuser -u routewatch -- /app/routewatch
# setpriv replaces itself with the daemon, so the daemon receives the stop
# signal directly. runuser would stay in between and kill the daemon 2 seconds
# after passing the signal on.
exec setpriv --reuid=routewatch --regid=routewatch --init-groups -- /app/routewatch
go.mod
+1 -1
View File
@@ -1,6 +1,6 @@
module git.eeqj.de/sneak/routewatch
go 1.24.4
go 1.27.1
require (
github.com/dustin/go-humanize v1.0.1
internal/config/config.go
+110 -10
View File
@@ -2,10 +2,14 @@
package config
import (
"errors"
"fmt"
"net/netip"
"os"
"path/filepath"
"runtime"
"strconv"
"strings"
"time"
)
@@ -18,6 +22,20 @@ const (
// defaultRouteExpirationMinutes is the default route expiration timeout in minutes
defaultRouteExpirationMinutes = 5
// defaultPort is the HTTP port used when PORT is not set
defaultPort = 8080
// maxPort is the highest TCP port number
maxPort = 65535
)
var (
errRelativeXDGDataHome = errors.New("XDG_DATA_HOME must be an absolute path")
errUnsupportedOS = errors.New("unsupported operating system")
errInvalidPort = errors.New("PORT must be a whole number")
errInvalidProxy = errors.New(
"TRUSTED_PROXIES must list IP addresses and CIDR ranges, comma-separated")
)
// Config holds configuration for the entire application
@@ -25,14 +43,24 @@ type Config struct {
// StateDir is the directory for all application state (database, snapshots)
StateDir string
// Port is the TCP port the HTTP server listens on
Port int
// TrustedProxies are the reverse proxies whose X-Forwarded-For header
// the HTTP server believes
TrustedProxies []netip.Prefix
// MaxRuntime is the maximum runtime (0 = run forever)
MaxRuntime time.Duration
// EnableBatchedDatabaseWrites enables batched database operations for better performance
// EnableBatchedDatabaseWrites enables batched database operations for
// better performance
EnableBatchedDatabaseWrites bool
// RouteExpirationTimeout is how long a route can go without being refreshed before expiring
// Default is 2 hours which is conservative for BGP (typical BGP hold time is 90-180 seconds)
// RouteExpirationTimeout is how long a route can go without being
// refreshed before expiring
// Default is 2 hours which is conservative for BGP (typical BGP hold
// time is 90-180 seconds)
RouteExpirationTimeout time.Duration
}
@@ -43,11 +71,26 @@ func New() (*Config, error) {
return nil, fmt.Errorf("failed to determine state directory: %w", err)
}
port, err := getPort()
if err != nil {
return nil, err
}
trustedProxies, err := getTrustedProxies()
if err != nil {
return nil, err
}
return &Config{
StateDir: stateDir,
MaxRuntime: 0, // Run forever by default
EnableBatchedDatabaseWrites: true, // Enable batching by default
RouteExpirationTimeout: defaultRouteExpirationMinutes * time.Minute, // For active route monitoring
StateDir: stateDir,
Port: port,
TrustedProxies: trustedProxies,
// Run forever by default
MaxRuntime: 0,
// Enable batching by default
EnableBatchedDatabaseWrites: true,
// For active route monitoring
RouteExpirationTimeout: defaultRouteExpirationMinutes * time.Minute,
}, nil
}
@@ -69,13 +112,20 @@ func getStateDirectory() (string, error) {
return filepath.Join(home, "Library", "Application Support", AppIdentifier), nil
case "linux", "freebsd", "openbsd", "netbsd":
// The XDG spec requires an absolute path; a relative one would put
// the database somewhere unexpected.
xdgData := os.Getenv("XDG_DATA_HOME")
if xdgData != "" && !filepath.IsAbs(xdgData) {
return "", fmt.Errorf("%w, got %q", errRelativeXDGDataHome, xdgData)
}
// Unix-like: /var/lib/berlin.sneak.app.routewatch if root, else XDG_DATA_HOME
if os.Geteuid() == 0 {
return filepath.Join("/var/lib", AppIdentifier), nil
}
// Check XDG_DATA_HOME first
if xdgData := os.Getenv("XDG_DATA_HOME"); xdgData != "" {
if xdgData != "" {
return filepath.Join(xdgData, AppIdentifier), nil
}
@@ -88,14 +138,64 @@ func getStateDirectory() (string, error) {
return filepath.Join(home, ".local", "share", AppIdentifier), nil
default:
return "", fmt.Errorf("unsupported operating system: %s", runtime.GOOS)
return "", fmt.Errorf("%w: %s", errUnsupportedOS, runtime.GOOS)
}
}
// getPort returns the HTTP port from PORT, or defaultPort when PORT is not set
func getPort() (int, error) {
value := os.Getenv("PORT")
if value == "" {
return defaultPort, nil
}
// ParseUint, unlike Atoi, refuses a sign: the health check URL cannot use "+9090"
port, err := strconv.ParseUint(value, 10, 0)
if err != nil || port < 1 || port > maxPort {
return 0, fmt.Errorf("%w from 1 to %d, got %q",
errInvalidPort, maxPort, value)
}
return int(port), nil
}
// getTrustedProxies returns the addresses and CIDR ranges listed in
// TRUSTED_PROXIES, such as "10.0.0.0/8,192.0.2.1", or none when it is not set.
// An address is returned as a range holding only that address.
func getTrustedProxies() ([]netip.Prefix, error) {
value := os.Getenv("TRUSTED_PROXIES")
if value == "" {
return nil, nil
}
var proxies []netip.Prefix
for entry := range strings.SplitSeq(value, ",") {
entry = strings.TrimSpace(entry)
addr, err := netip.ParseAddr(entry)
if err == nil {
proxies = append(proxies, netip.PrefixFrom(addr, addr.BitLen()))
continue
}
prefix, err := netip.ParsePrefix(entry)
if err != nil {
return nil, fmt.Errorf("%w, got %q", errInvalidProxy, entry)
}
proxies = append(proxies, prefix)
}
return proxies, nil
}
// EnsureDirectories creates all necessary directories if they don't exist
func (c *Config) EnsureDirectories() error {
// Ensure state directory exists
if err := os.MkdirAll(c.StateDir, dirPermissions); err != nil {
err := os.MkdirAll(c.StateDir, dirPermissions)
if err != nil {
return fmt.Errorf("failed to create state directory: %w", err)
}
internal/config/config_internal_test.go
+118
View File
@@ -0,0 +1,118 @@
package config
import (
"net/netip"
"runtime"
"slices"
"testing"
)
func TestNewReadsPort(t *testing.T) {
tests := map[string]int{
"": defaultPort,
"1": 1,
"9090": 9090,
"65535": 65535,
}
for value, want := range tests {
t.Run(value, func(t *testing.T) {
t.Setenv("PORT", value)
t.Setenv("XDG_DATA_HOME", "")
cfg, err := New()
if err != nil {
t.Fatalf("New() with PORT=%q: %v", value, err)
}
if cfg.Port != want {
t.Errorf("New() with PORT=%q: Port = %d, want %d", value, cfg.Port, want)
}
})
}
}
func TestNewRefusesInvalidPort(t *testing.T) {
for _, value := range []string{"0", "65536", "-1", "+9090", "http", "80.5"} {
t.Run(value, func(t *testing.T) {
t.Setenv("PORT", value)
t.Setenv("XDG_DATA_HOME", "")
_, err := New()
if err == nil {
t.Errorf("New() with PORT=%q returned no error", value)
}
})
}
}
func TestNewReadsTrustedProxies(t *testing.T) {
t.Setenv("PORT", "")
t.Setenv("XDG_DATA_HOME", "")
t.Setenv("TRUSTED_PROXIES", "")
cfg, err := New()
if err != nil {
t.Fatalf("New() without TRUSTED_PROXIES: %v", err)
}
if len(cfg.TrustedProxies) != 0 {
t.Errorf("New() without TRUSTED_PROXIES: TrustedProxies = %v, want none",
cfg.TrustedProxies)
}
t.Setenv("TRUSTED_PROXIES", "10.0.0.0/8, 192.0.2.1,2001:db8::/32")
cfg, err = New()
if err != nil {
t.Fatalf("New() with TRUSTED_PROXIES: %v", err)
}
want := []netip.Prefix{
netip.MustParsePrefix("10.0.0.0/8"),
netip.MustParsePrefix("192.0.2.1/32"),
netip.MustParsePrefix("2001:db8::/32"),
}
if !slices.Equal(cfg.TrustedProxies, want) {
t.Errorf("TrustedProxies = %v, want %v", cfg.TrustedProxies, want)
}
}
func TestNewRefusesInvalidTrustedProxies(t *testing.T) {
invalid := []string{"proxy.example", "10.0.0.0/33", "192.0.2.1:80", "192.0.2.1,"}
for _, value := range invalid {
t.Run(value, func(t *testing.T) {
t.Setenv("PORT", "")
t.Setenv("XDG_DATA_HOME", "")
t.Setenv("TRUSTED_PROXIES", value)
_, err := New()
if err == nil {
t.Errorf("New() with TRUSTED_PROXIES=%q returned no error", value)
}
})
}
}
func TestNewRefusesRelativeXDGDataHome(t *testing.T) {
if runtime.GOOS == "darwin" {
t.Skip("macOS does not read XDG_DATA_HOME")
}
t.Setenv("PORT", "")
t.Setenv("XDG_DATA_HOME", "relative/path")
_, err := New()
if err == nil {
t.Error("New() with a relative XDG_DATA_HOME returned no error")
}
t.Setenv("XDG_DATA_HOME", "/var/lib")
_, err = New()
if err != nil {
t.Errorf("New() with XDG_DATA_HOME=/var/lib: %v", err)
}
}
internal/database/counts.go
+218
View File
@@ -0,0 +1,218 @@
package database
import (
"context"
"fmt"
"sync"
)
// liveCounts holds the running row counts and the prefix distribution that the
// stats endpoints report. They are seeded once at startup from the tables and
// then adjusted on every write, so a stats read serves them from memory instead
// of running a query over the tables. The COUNT(*) scans (issue 27) and then
// the prefix distribution query (issue 30) each grew with the database until
// they took the whole request timeout and made /api/v1/stats return 500.
//
// A single mutex guards all fields so the stats reader takes a consistent
// snapshot at one instant and writers, which already run under the database
// write lock, adjust the counts after their transaction commits.
type liveCounts struct {
mu sync.RWMutex
asns int
prefixesV4 int
prefixesV6 int
peerings int
peers int
routesV4 int
routesV6 int
// The prefix distribution: for each mask length, the number of distinct
// prefixes that have at least one live route.
distributionV4 [ipv4Bits + 1]int
distributionV6 [ipv6Bits + 1]int
}
// seed sets every count to the value read from the tables at startup. It runs
// before any writer, so it needs no coordination with the adjust methods.
func (c *liveCounts) seed(
asns, prefixesV4, prefixesV6, peerings, peers, routesV4, routesV6 int,
distributionV4, distributionV6 []PrefixDistribution,
) {
c.mu.Lock()
defer c.mu.Unlock()
c.asns = asns
c.prefixesV4 = prefixesV4
c.prefixesV6 = prefixesV6
c.peerings = peerings
c.peers = peers
c.routesV4 = routesV4
c.routesV6 = routesV6
for _, entry := range distributionV4 {
addAtMaskLength(c.distributionV4[:], entry.MaskLength, entry.Count)
}
for _, entry := range distributionV6 {
addAtMaskLength(c.distributionV6[:], entry.MaskLength, entry.Count)
}
}
// addASNs adds n to the ASN count.
func (c *liveCounts) addASNs(n int) {
c.mu.Lock()
c.asns += n
c.mu.Unlock()
}
// addPrefixes adds to the IPv4 and IPv6 prefix counts.
func (c *liveCounts) addPrefixes(v4, v6 int) {
c.mu.Lock()
c.prefixesV4 += v4
c.prefixesV6 += v6
c.mu.Unlock()
}
// addPeerings adds n to the peering count.
func (c *liveCounts) addPeerings(n int) {
c.mu.Lock()
c.peerings += n
c.mu.Unlock()
}
// addPeers adds n to the BGP peer count.
func (c *liveCounts) addPeers(n int) {
c.mu.Lock()
c.peers += n
c.mu.Unlock()
}
// addRoutes adds to the IPv4 and IPv6 live-route counts. Deletions pass
// negative values.
func (c *liveCounts) addRoutes(v4, v6 int) {
c.mu.Lock()
c.routesV4 += v4
c.routesV6 += v6
c.mu.Unlock()
}
// addToDistribution adds n to the IPv4 and IPv6 prefix distributions once for
// each listed mask length. A write lists the mask lengths of the prefixes it
// gave their first live route with n = 1, and of the prefixes it left with no
// live route with n = -1.
func (c *liveCounts) addToDistribution(maskLengthsV4, maskLengthsV6 []int, n int) {
c.mu.Lock()
defer c.mu.Unlock()
for _, maskLength := range maskLengthsV4 {
addAtMaskLength(c.distributionV4[:], maskLength, n)
}
for _, maskLength := range maskLengthsV6 {
addAtMaskLength(c.distributionV6[:], maskLength, n)
}
}
// addAtMaskLength adds n to counts[maskLength]. A mask length the array has no
// entry for is ignored, so a malformed route cannot crash the daemon.
func addAtMaskLength(counts []int, maskLength, n int) {
if maskLength >= 0 && maskLength < len(counts) {
counts[maskLength] += n
}
}
// fill copies the counts into a Stats, including the derived totals, under a
// single read lock so the reader sees one consistent snapshot.
func (c *liveCounts) fill(s *Stats) {
c.mu.RLock()
defer c.mu.RUnlock()
s.ASNs = c.asns
s.IPv4Prefixes = c.prefixesV4
s.IPv6Prefixes = c.prefixesV6
s.Prefixes = c.prefixesV4 + c.prefixesV6
s.Peerings = c.peerings
s.Peers = c.peers
s.IPv4Routes = c.routesV4
s.IPv6Routes = c.routesV6
s.LiveRoutes = c.routesV4 + c.routesV6
s.IPv4PrefixDistribution = distributionList(c.distributionV4[:])
s.IPv6PrefixDistribution = distributionList(c.distributionV6[:])
}
// distributionList lists the mask lengths that have at least one prefix, in
// ascending order, the way the distribution query returns them.
func distributionList(counts []int) []PrefixDistribution {
var list []PrefixDistribution
for maskLength, count := range counts {
if count > 0 {
list = append(list,
PrefixDistribution{MaskLength: maskLength, Count: count})
}
}
return list
}
// countRows returns the number of rows in the named table. It is used only at
// startup to seed the in-memory counters, so a full COUNT(*) is acceptable.
func (d *Database) countRows(ctx context.Context, table string) (int, error) {
var n int
// table is one of a fixed set of literals below, never external input.
err := d.db.QueryRowContext(ctx, "SELECT COUNT(*) FROM "+table).Scan(&n)
if err != nil {
return 0, fmt.Errorf("failed to count %s: %w", table, err)
}
return n, nil
}
// seedCounts reads the current row counts and prefix distribution from the
// tables into the in-memory counters. It runs once at startup, before the
// streamer begins writing.
func (d *Database) seedCounts(ctx context.Context) error {
asns, err := d.countRows(ctx, "asns")
if err != nil {
return err
}
prefixesV4, err := d.countRows(ctx, "prefixes_v4")
if err != nil {
return err
}
prefixesV6, err := d.countRows(ctx, "prefixes_v6")
if err != nil {
return err
}
peerings, err := d.countRows(ctx, "peerings")
if err != nil {
return err
}
peers, err := d.countRows(ctx, "bgp_peers")
if err != nil {
return err
}
routesV4, err := d.countRows(ctx, tableLiveRoutesV4)
if err != nil {
return err
}
routesV6, err := d.countRows(ctx, tableLiveRoutesV6)
if err != nil {
return err
}
distributionV4, distributionV6, err := d.GetPrefixDistributionContext(ctx)
if err != nil {
return err
}
d.counts.seed(asns, prefixesV4, prefixesV6, peerings, peers, routesV4, routesV6,
distributionV4, distributionV6)
return nil
}
internal/database/counts_internal_test.go
+841
View File
@@ -0,0 +1,841 @@
package database
import (
"context"
"fmt"
"slices"
"sync"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/config"
"git.eeqj.de/sneak/routewatch/internal/logger"
"github.com/google/uuid"
)
// testV4PeerIP is the peer the IPv4 routes in these tests are learned from.
const testV4PeerIP = "192.0.2.1"
// testV6Prefix is the prefix of the IPv6 routes in these tests.
const testV6Prefix = "2001:db8::/32"
// mkV4Route builds an IPv4 live route with its mask length taken from the
// prefix.
func mkV4Route(t *testing.T, prefix string, asn int, ts time.Time) *LiveRoute {
t.Helper()
maskLength, err := prefixMaskLength(prefix)
if err != nil {
t.Fatalf("prefixMaskLength(%s): %v", prefix, err)
}
return &LiveRoute{
Prefix: prefix,
MaskLength: maskLength,
IPVersion: ipVersionV4,
OriginASN: asn,
PeerIP: testV4PeerIP,
ASPath: []int{asn},
NextHop: "192.0.2.254",
LastUpdated: ts,
}
}
// mkV6Route builds an IPv6 live route.
func mkV6Route(prefix string, asn int, ts time.Time) *LiveRoute {
return &LiveRoute{
Prefix: prefix,
MaskLength: 32,
IPVersion: ipVersionV6,
OriginASN: asn,
PeerIP: "2001:db8::1",
ASPath: []int{asn},
NextHop: "2001:db8::ffff",
LastUpdated: ts,
}
}
// TestLiveCountsTrackWritesInRealtime checks that the stats counts start at
// zero, reflect each write the moment it commits (no recompute, no timer), do
// not move when a route is merely re-announced, and drop when a route is
// deleted. These counts are what /api/v1/stats reports; before this change the
// endpoint recomputed them with a COUNT(*) over each table on every request.
func TestLiveCountsTrackWritesInRealtime(t *testing.T) {
t.Parallel()
cfg := &config.Config{StateDir: t.TempDir()}
db, err := New(cfg, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ctx := context.Background()
empty, err := db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext on empty database: %v", err)
}
if empty.ASNs != 0 || empty.Prefixes != 0 || empty.Peerings != 0 ||
empty.Peers != 0 || empty.LiveRoutes != 0 {
t.Fatalf("empty database counts nonzero: %+v", empty)
}
ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
writeASNsPrefixesAndPeers(t, db, ts)
routes := []*LiveRoute{
mkV4Route(t, "198.51.100.0/24", 64500, ts),
mkV4Route(t, "203.0.113.0/24", 64501, ts.Add(time.Minute)),
mkV6Route(testV6Prefix, 64502, ts.Add(2*time.Minute)),
}
err = db.UpsertLiveRouteBatch(routes)
if err != nil {
t.Fatalf("UpsertLiveRouteBatch: %v", err)
}
stats, err := db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext: %v", err)
}
assertCounts(t, "after inserts", stats, wantCounts{
asns: 2, prefixes: 2, peerings: 1, peers: 1,
ipv4Routes: 2, ipv6Routes: 1, liveRoutes: 3,
})
// Re-announcing the same routes is an update, not an insert: counts hold.
err = db.UpsertLiveRouteBatch(routes)
if err != nil {
t.Fatalf("UpsertLiveRouteBatch (re-announce): %v", err)
}
stats, err = db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext: %v", err)
}
assertCounts(t, "after re-announce", stats, wantCounts{
asns: 2, prefixes: 2, peerings: 1, peers: 1,
ipv4Routes: 2, ipv6Routes: 1, liveRoutes: 3,
})
// A withdrawal removes one route.
err = db.DeleteLiveRouteBatch([]LiveRouteDeletion{
{Prefix: "203.0.113.0/24", PeerIP: testV4PeerIP, IPVersion: ipVersionV4},
})
if err != nil {
t.Fatalf("DeleteLiveRouteBatch: %v", err)
}
stats, err = db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext: %v", err)
}
assertCounts(t, "after delete", stats, wantCounts{
asns: 2, prefixes: 2, peerings: 1, peers: 1,
ipv4Routes: 1, ipv6Routes: 1, liveRoutes: 2,
})
}
// writeASNsPrefixesAndPeers writes two ASNs, two prefixes, one peer and one
// peering: the rows besides live routes that the stats count.
func writeASNsPrefixesAndPeers(t *testing.T, db *Database, ts time.Time) {
t.Helper()
seen := SeenTimes{First: ts, Last: ts}
err := db.GetOrCreateASNBatch(map[int]SeenTimes{64500: seen, 64501: seen})
if err != nil {
t.Fatalf("GetOrCreateASNBatch: %v", err)
}
err = db.UpdatePrefixesBatch(map[string]time.Time{
"198.51.100.0/24": ts,
testV6Prefix: ts,
})
if err != nil {
t.Fatalf("UpdatePrefixesBatch: %v", err)
}
err = db.UpdatePeerBatch(map[string]PeerUpdate{
testV4PeerIP: {
PeerIP: testV4PeerIP, PeerASN: 64500,
MessageType: "UPDATE", Seen: seen,
},
})
if err != nil {
t.Fatalf("UpdatePeerBatch: %v", err)
}
err = db.RecordPeering(64500, 64501, seen)
if err != nil {
t.Fatalf("RecordPeering: %v", err)
}
}
// TestLiveCountsSeededFromDatabaseAtStartup writes rows, reopens the same
// database file, and checks the counts and the prefix distribution come back
// from the seed scan rather than starting at zero.
func TestLiveCountsSeededFromDatabaseAtStartup(t *testing.T) {
t.Parallel()
cfg := &config.Config{StateDir: t.TempDir()}
db, err := New(cfg, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
seen := SeenTimes{First: ts, Last: ts}
err = db.GetOrCreateASNBatch(
map[int]SeenTimes{64500: seen, 64501: seen, 64502: seen})
if err != nil {
t.Fatalf("GetOrCreateASNBatch: %v", err)
}
err = db.UpsertLiveRouteBatch([]*LiveRoute{
mkV4Route(t, "198.51.100.0/24", 64500, ts),
mkV6Route(testV6Prefix, 64502, ts),
})
if err != nil {
t.Fatalf("UpsertLiveRouteBatch: %v", err)
}
err = db.Close()
if err != nil {
t.Fatalf("Close: %v", err)
}
reopened, err := New(cfg, logger.New())
if err != nil {
t.Fatalf("failed to reopen database: %v", err)
}
defer func() { _ = reopened.Close() }()
stats, err := reopened.GetStatsContext(context.Background())
if err != nil {
t.Fatalf("GetStatsContext after reopen: %v", err)
}
if stats.ASNs != 3 {
t.Errorf("seeded ASNs = %d, want 3", stats.ASNs)
}
if stats.IPv4Routes != 1 || stats.IPv6Routes != 1 || stats.LiveRoutes != 2 {
t.Errorf("seeded routes = (v4 %d, v6 %d, total %d), want (1, 1, 2)",
stats.IPv4Routes, stats.IPv6Routes, stats.LiveRoutes)
}
assertDistribution(t, "seeded IPv4 distribution", stats.IPv4PrefixDistribution,
[]PrefixDistribution{{MaskLength: 24, Count: 1}})
assertDistribution(t, "seeded IPv6 distribution", stats.IPv6PrefixDistribution,
[]PrefixDistribution{{MaskLength: 32, Count: 1}})
}
// TestPrefixDistributionTracksWrites checks that the prefix distribution the
// stats read reports stays exact across each kind of live-route write, and that
// after every step it equals what the distribution query reads from the route
// tables. The steps run once through the batch methods the prefix handler uses
// and once through the single-route methods.
func TestPrefixDistributionTracksWrites(t *testing.T) {
t.Parallel()
steps := prefixDistributionSteps(t)
for _, batch := range []bool{true, false} {
name := "single-route writes"
if batch {
name = "batch writes"
}
t.Run(name, func(t *testing.T) {
t.Parallel()
db, err := New(&config.Config{StateDir: t.TempDir()}, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ctx := context.Background()
for _, step := range steps {
err := announceRoutes(db, batch, step.announce)
if err != nil {
t.Fatalf("%s: announce: %v", step.name, err)
}
err = withdrawRoutes(db, batch, step.withdraw)
if err != nil {
t.Fatalf("%s: withdraw: %v", step.name, err)
}
stats, err := db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("%s: GetStatsContext: %v", step.name, err)
}
queryV4, queryV6, err := db.GetPrefixDistributionContext(ctx)
if err != nil {
t.Fatalf("%s: GetPrefixDistributionContext: %v", step.name, err)
}
assertDistribution(t, step.name+": IPv4 distribution",
stats.IPv4PrefixDistribution, step.wantV4)
assertDistribution(t, step.name+": IPv6 distribution",
stats.IPv6PrefixDistribution, step.wantV6)
assertDistribution(t, step.name+": IPv4 distribution query",
queryV4, step.wantV4)
assertDistribution(t, step.name+": IPv6 distribution query",
queryV6, step.wantV6)
}
})
}
}
// prefixDistributionStep is one step of TestPrefixDistributionTracksWrites:
// the routes it announces, then the routes it withdraws, then the
// distributions both the stats and the distribution query must report.
type prefixDistributionStep struct {
name string
announce []*LiveRoute
withdraw []*LiveRoute
wantV4 []PrefixDistribution
wantV6 []PrefixDistribution
}
// prefixDistributionSteps returns the steps of
// TestPrefixDistributionTracksWrites, in the order they run.
func prefixDistributionSteps(t *testing.T) []prefixDistributionStep {
t.Helper()
ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
shared := mkV4Route(t, "198.51.100.0/24", 64500, ts)
sharedSecondPeer := mkV4Route(t, "198.51.100.0/24", 64500, ts)
sharedSecondPeer.PeerIP = "192.0.2.2"
other := mkV4Route(t, "203.0.113.0/24", 64501, ts)
wide := mkV4Route(t, "172.16.0.0/16", 64502, ts)
v6 := mkV6Route(testV6Prefix, 64503, ts)
v6SecondPeer := mkV6Route(testV6Prefix, 64503, ts)
v6SecondPeer.PeerIP = "2001:db8::2"
all := []PrefixDistribution{{MaskLength: 16, Count: 1}, {MaskLength: 24, Count: 2}}
v6Only := []PrefixDistribution{{MaskLength: 32, Count: 1}}
return []prefixDistributionStep{
{
name: "new routes",
announce: []*LiveRoute{shared, other, wide, v6},
wantV4: all, wantV6: v6Only,
},
{
name: "re-announcement",
announce: []*LiveRoute{shared, other, wide, v6},
wantV4: all, wantV6: v6Only,
},
{
name: "second peer announces a prefix that has a live route",
announce: []*LiveRoute{sharedSecondPeer, v6SecondPeer},
wantV4: all, wantV6: v6Only,
},
{
name: "withdrawal of a route that is not the last for its prefix",
withdraw: []*LiveRoute{shared, v6},
wantV4: all, wantV6: v6Only,
},
{
name: "withdrawal of the last route for a prefix",
withdraw: []*LiveRoute{sharedSecondPeer, v6SecondPeer},
wantV4: []PrefixDistribution{
{MaskLength: 16, Count: 1}, {MaskLength: 24, Count: 1},
},
},
{
name: "withdrawal of every remaining route",
withdraw: []*LiveRoute{other, wide},
},
{
name: "two peers announce a new prefix together",
announce: []*LiveRoute{shared, sharedSecondPeer, v6, v6SecondPeer},
wantV4: []PrefixDistribution{{MaskLength: 24, Count: 1}},
wantV6: v6Only,
},
{
name: "both routes for a prefix withdrawn together",
withdraw: []*LiveRoute{shared, sharedSecondPeer, v6, v6SecondPeer},
},
// The feed often withdraws a route that is not live. That must not take
// the prefix out of the distribution. A count wrongly taken below zero is
// left out of the answer, so the next step shows it: its announcement
// at the same mask length would then not be counted.
{
name: "withdrawal of routes that are not live",
withdraw: []*LiveRoute{other, v6},
},
{
name: "announcement after a withdrawal of routes that are not live",
announce: []*LiveRoute{other, v6},
wantV4: []PrefixDistribution{{MaskLength: 24, Count: 1}},
wantV6: v6Only,
},
}
}
// announceRoutes writes routes in one UpsertLiveRouteBatch, or with one
// UpsertLiveRoute each.
func announceRoutes(db *Database, batch bool, routes []*LiveRoute) error {
if batch {
return db.UpsertLiveRouteBatch(routes)
}
for _, route := range routes {
err := db.UpsertLiveRoute(route)
if err != nil {
return err
}
}
return nil
}
// withdrawRoutes removes routes in one DeleteLiveRouteBatch, or with one
// DeleteLiveRoute each.
func withdrawRoutes(db *Database, batch bool, routes []*LiveRoute) error {
if !batch {
for _, route := range routes {
err := db.DeleteLiveRoute(route.Prefix, route.PeerIP)
if err != nil {
return err
}
}
return nil
}
deletions := make([]LiveRouteDeletion, 0, len(routes))
for _, route := range routes {
deletions = append(deletions, LiveRouteDeletion{
Prefix: route.Prefix, PeerIP: route.PeerIP, IPVersion: route.IPVersion,
})
}
return db.DeleteLiveRouteBatch(deletions)
}
// TestDeleteLiveRoutesOfPeer gives two peers IPv4 and IPv6 live routes, some
// for the same prefixes, and deletes all of one peer's routes. The other peer's
// routes stay, and the route counts and prefix distribution held in memory
// equal what is read from the route tables.
func TestDeleteLiveRoutesOfPeer(t *testing.T) {
t.Parallel()
const (
peerIP = "192.0.2.1"
otherPeerIP = "192.0.2.2"
sharedV4 = "198.51.100.0/24"
sharedV6 = "2001:db8::/32"
)
db, err := New(&config.Config{StateDir: t.TempDir()}, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
ofPeer := func(peer string, route *LiveRoute) *LiveRoute {
route.PeerIP = peer
return route
}
onlyV6 := mkV6Route("2001:db8:1::/48", 64503, ts)
onlyV6.MaskLength = 48
err = db.UpsertLiveRouteBatch([]*LiveRoute{
ofPeer(peerIP, mkV4Route(t, sharedV4, 64500, ts)),
ofPeer(otherPeerIP, mkV4Route(t, sharedV4, 64500, ts)),
ofPeer(peerIP, mkV4Route(t, "203.0.113.0/24", 64501, ts)),
ofPeer(peerIP, mkV6Route(sharedV6, 64502, ts)),
ofPeer(otherPeerIP, mkV6Route(sharedV6, 64502, ts)),
ofPeer(peerIP, onlyV6),
})
if err != nil {
t.Fatalf("UpsertLiveRouteBatch: %v", err)
}
// peerIP has two IPv4 and two IPv6 routes; a second delete finds none.
for _, want := range []int{4, 0} {
deleted, err := db.DeleteLiveRoutesOfPeer(peerIP)
if err != nil {
t.Fatalf("DeleteLiveRoutesOfPeer: %v", err)
}
if deleted != want {
t.Errorf("DeleteLiveRoutesOfPeer deleted %d routes, want %d",
deleted, want)
}
}
for _, prefix := range []string{sharedV4, sharedV6} {
routes, _, err := db.GetPrefixDetailsContext(context.Background(), prefix)
if err != nil {
t.Fatalf("GetPrefixDetailsContext(%s): %v", prefix, err)
}
if len(routes) != 1 || routes[0].PeerIP != otherPeerIP {
t.Errorf("live routes for %s = %+v, want only the one from %s",
prefix, routes, otherPeerIP)
}
}
checkOneLiveRoutePerIPVersion(t, db)
}
// TestDeleteLiveRoutesOfPeerDeletesInChunks gives a peer one IPv4 route more
// than a chunk holds, two of them for prefixes another peer also has. One chunk
// deletes peerRouteChunkSize of the peer's routes and commits, leaving the last
// one. The routes are written again, and DeleteLiveRoutesOfPeer deletes all of
// them, a chunk and then the last one. After each step the route count and
// prefix distribution held in memory and those read from the route table are
// the ones wanted.
func TestDeleteLiveRoutesOfPeerDeletesInChunks(t *testing.T) {
t.Parallel()
const (
peerIP = "192.0.2.1"
otherPeerIP = "192.0.2.2"
// How many of the peer's prefixes the other peer also has, the first
// ones in prefix order.
sharedPrefixes = 2
)
db, err := New(&config.Config{StateDir: t.TempDir()}, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
peerRoutes := make([]*LiveRoute, 0, peerRouteChunkSize+1)
otherPeerRoutes := make([]*LiveRoute, 0, sharedPrefixes)
for i := range peerRouteChunkSize + 1 {
prefix := fmt.Sprintf("10.%d.%d.0/24", i/256, i%256)
route := mkV4Route(t, prefix, 64500, ts)
route.PeerIP = peerIP
peerRoutes = append(peerRoutes, route)
if i < sharedPrefixes {
route := mkV4Route(t, prefix, 64500, ts)
route.PeerIP = otherPeerIP
otherPeerRoutes = append(otherPeerRoutes, route)
}
}
err = db.UpsertLiveRouteBatch(slices.Concat(peerRoutes, otherPeerRoutes))
if err != nil {
t.Fatalf("UpsertLiveRouteBatch: %v", err)
}
deletedV4, deletedV6, err := db.deletePeerRouteChunk(peerIP)
if err != nil {
t.Fatalf("deletePeerRouteChunk: %v", err)
}
if deletedV4 != peerRouteChunkSize || deletedV6 != 0 {
t.Errorf("deletePeerRouteChunk deleted (%d, %d) routes, want (%d, 0)",
deletedV4, deletedV6, peerRouteChunkSize)
}
// The shared prefixes went with the chunk, so each route left has a
// prefix of its own.
checkV4RoutesOfOwnPrefixes(t, db, "after one chunk", 1+sharedPrefixes)
err = db.UpsertLiveRouteBatch(peerRoutes)
if err != nil {
t.Fatalf("UpsertLiveRouteBatch: %v", err)
}
deleted, err := db.DeleteLiveRoutesOfPeer(peerIP)
if err != nil {
t.Fatalf("DeleteLiveRoutesOfPeer: %v", err)
}
if deleted != len(peerRoutes) {
t.Errorf("DeleteLiveRoutesOfPeer deleted %d routes, want %d",
deleted, len(peerRoutes))
}
checkV4RoutesOfOwnPrefixes(t, db, "after DeleteLiveRoutesOfPeer",
sharedPrefixes)
}
// checkV4RoutesOfOwnPrefixes checks that the database has wantRoutes IPv4 live
// routes, each for a /24 no other live route has, both as counted in memory and
// as read from the route table. when says which step of the test checks.
func checkV4RoutesOfOwnPrefixes(
t *testing.T, db *Database, when string, wantRoutes int,
) {
t.Helper()
ctx := context.Background()
stats, err := db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext: %v", err)
}
tableRoutes, _, err := db.GetLiveRouteCountsContext(ctx)
if err != nil {
t.Fatalf("GetLiveRouteCountsContext: %v", err)
}
if stats.IPv4Routes != wantRoutes || tableRoutes != wantRoutes {
t.Errorf("%s: IPv4 live routes: %d in memory, %d in the table, want %d",
when, stats.IPv4Routes, tableRoutes, wantRoutes)
}
queryDistribution, _, err := db.GetPrefixDistributionContext(ctx)
if err != nil {
t.Fatalf("GetPrefixDistributionContext: %v", err)
}
want := []PrefixDistribution{{MaskLength: 24, Count: wantRoutes}}
assertDistribution(t, when+": IPv4 distribution",
stats.IPv4PrefixDistribution, want)
assertDistribution(t, when+": IPv4 distribution query",
queryDistribution, want)
}
// checkOneLiveRoutePerIPVersion checks that the database has one IPv4 live
// route, for a /24, and one IPv6 live route, for a /32, both as counted in
// memory and as read from the route tables.
func checkOneLiveRoutePerIPVersion(t *testing.T, db *Database) {
t.Helper()
ctx := context.Background()
stats, err := db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext: %v", err)
}
tableV4, tableV6, err := db.GetLiveRouteCountsContext(ctx)
if err != nil {
t.Fatalf("GetLiveRouteCountsContext: %v", err)
}
if stats.IPv4Routes != 1 || tableV4 != 1 || stats.IPv6Routes != 1 ||
tableV6 != 1 {
t.Errorf("live routes (IPv4, IPv6): (%d, %d) in memory, (%d, %d) in "+
"the tables, want (1, 1)",
stats.IPv4Routes, stats.IPv6Routes, tableV4, tableV6)
}
queryV4, queryV6, err := db.GetPrefixDistributionContext(ctx)
if err != nil {
t.Fatalf("GetPrefixDistributionContext: %v", err)
}
wantV4 := []PrefixDistribution{{MaskLength: 24, Count: 1}}
wantV6 := []PrefixDistribution{{MaskLength: 32, Count: 1}}
assertDistribution(t, "IPv4 distribution", stats.IPv4PrefixDistribution, wantV4)
assertDistribution(t, "IPv6 distribution", stats.IPv6PrefixDistribution, wantV6)
assertDistribution(t, "IPv4 distribution query", queryV4, wantV4)
assertDistribution(t, "IPv6 distribution query", queryV6, wantV6)
}
// TestStatsRouteTimestamps checks the oldest/newest route timestamps are read
// from the right rows across both tables and parse into time.Time. The old
// MIN/MAX union query read its result into *time.Time, which the driver could
// not parse, so it logged a warning every call and left both timestamps nil.
func TestStatsRouteTimestamps(t *testing.T) {
t.Parallel()
cfg := &config.Config{StateDir: t.TempDir()}
db, err := New(cfg, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ctx := context.Background()
empty, err := db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext on empty database: %v", err)
}
if empty.OldestRoute != nil || empty.NewestRoute != nil {
t.Fatalf("empty database timestamps = (%v, %v), want (nil, nil)",
empty.OldestRoute, empty.NewestRoute)
}
base := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
oldest := base
newest := base.Add(2 * time.Minute)
err = db.UpsertLiveRouteBatch([]*LiveRoute{
mkV4Route(t, "198.51.100.0/24", 64500, base.Add(time.Minute)),
mkV4Route(t, "203.0.113.0/24", 64501, oldest),
mkV6Route(testV6Prefix, 64502, newest),
})
if err != nil {
t.Fatalf("UpsertLiveRouteBatch: %v", err)
}
stats, err := db.GetStatsContext(ctx)
if err != nil {
t.Fatalf("GetStatsContext: %v", err)
}
if stats.OldestRoute == nil || !stats.OldestRoute.Equal(oldest) {
t.Errorf("OldestRoute = %v, want %v", stats.OldestRoute, oldest)
}
if stats.NewestRoute == nil || !stats.NewestRoute.Equal(newest) {
t.Errorf("NewestRoute = %v, want %v", stats.NewestRoute, newest)
}
}
// TestLiveCountsConcurrentReadWrite runs writers and stats readers at once so
// the race detector proves the counters are safe under concurrent use.
func TestLiveCountsConcurrentReadWrite(t *testing.T) {
t.Parallel()
cfg := &config.Config{StateDir: t.TempDir()}
db, err := New(cfg, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
const writers = 4
var wg sync.WaitGroup
wg.Add(writers)
for w := range writers {
go func(base int) {
defer wg.Done()
for i := range 25 {
asn := 65000 + base*100 + i
route := mkV6Route(testV6Prefix, asn, ts)
route.PeerIP = "2001:db8::" + uuid.NewString()
err := db.UpsertLiveRoute(route)
if err != nil {
t.Errorf("UpsertLiveRoute: %v", err)
return
}
}
}(w)
}
var readerWG sync.WaitGroup
readerWG.Add(1)
stop := make(chan struct{})
go func() {
defer readerWG.Done()
for {
select {
case <-stop:
return
default:
_, err := db.GetStatsContext(context.Background())
if err != nil {
t.Errorf("GetStatsContext: %v", err)
return
}
}
}
}()
wg.Wait()
close(stop)
readerWG.Wait()
stats, err := db.GetStatsContext(context.Background())
if err != nil {
t.Fatalf("GetStatsContext: %v", err)
}
if want := writers * 25; stats.IPv6Routes != want {
t.Errorf("IPv6Routes = %d, want %d", stats.IPv6Routes, want)
}
// Every writer announced the same prefix, so it counts once.
assertDistribution(t, "IPv6 distribution", stats.IPv6PrefixDistribution,
[]PrefixDistribution{{MaskLength: 32, Count: 1}})
}
type wantCounts struct {
asns int
prefixes int
peerings int
peers int
ipv4Routes int
ipv6Routes int
liveRoutes int
}
func assertCounts(t *testing.T, when string, got Stats, want wantCounts) {
t.Helper()
if got.ASNs != want.asns {
t.Errorf("%s: ASNs = %d, want %d", when, got.ASNs, want.asns)
}
if got.Prefixes != want.prefixes {
t.Errorf("%s: Prefixes = %d, want %d", when, got.Prefixes, want.prefixes)
}
if got.Peerings != want.peerings {
t.Errorf("%s: Peerings = %d, want %d", when, got.Peerings, want.peerings)
}
if got.Peers != want.peers {
t.Errorf("%s: Peers = %d, want %d", when, got.Peers, want.peers)
}
if got.IPv4Routes != want.ipv4Routes {
t.Errorf("%s: IPv4Routes = %d, want %d", when, got.IPv4Routes, want.ipv4Routes)
}
if got.IPv6Routes != want.ipv6Routes {
t.Errorf("%s: IPv6Routes = %d, want %d", when, got.IPv6Routes, want.ipv6Routes)
}
if got.LiveRoutes != want.liveRoutes {
t.Errorf("%s: LiveRoutes = %d, want %d", when, got.LiveRoutes, want.liveRoutes)
}
}
func assertDistribution(t *testing.T, what string, got, want []PrefixDistribution) {
t.Helper()
if !slices.Equal(got, want) {
t.Errorf("%s = %v, want %v", what, got, want)
}
}
internal/database/database.go
+1927 -1146
View File
File diff suppressed because it is too large Load Diff
internal/database/database_internal_test.go
+2010
View File
File diff suppressed because it is too large Load Diff
internal/database/database_test.go
-444
View File
@@ -1,444 +0,0 @@
package database
import (
"context"
"database/sql"
"net"
"sync"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/config"
"git.eeqj.de/sneak/routewatch/internal/logger"
)
// tempStoreMemory is the PRAGMA temp_store value meaning "hold temp B-trees in
// memory"; the DSN change must leave temp_store below this so they spill to disk.
const tempStoreMemory = 2
// heldConnections is how many pooled connections the pragma test holds open at
// once so each is a distinct SQLite connection that parsed the DSN.
const heldConnections = 5
// Parameters for the checkpoint-contention regression test.
const (
// contentionIterations is how many batch writes race the checkpoint loop.
contentionIterations = 400
// contendedASNCount is the small set of ASNs the batches reuse, so most
// batches update existing rows and exercise the read-before-write path.
contendedASNCount = 16
// asnSecondBand offsets a second ASN per batch so each batch writes more
// than one row.
asnSecondBand = 100
)
func TestIPToUint32(t *testing.T) {
tests := []struct {
name string
ip string
expected uint32
}{
{
name: "Simple IP",
ip: "192.168.1.1",
expected: 3232235777, // 192<<24 + 168<<16 + 1<<8 + 1
},
{
name: "Minimum IP",
ip: "0.0.0.0",
expected: 0,
},
{
name: "Maximum IP",
ip: "255.255.255.255",
expected: 4294967295,
},
{
name: "10.0.0.0",
ip: "10.0.0.0",
expected: 167772160,
},
{
name: "172.16.0.0",
ip: "172.16.0.0",
expected: 2886729728,
},
{
name: "8.8.8.8",
ip: "8.8.8.8",
expected: 134744072,
},
{
name: "1.2.3.4",
ip: "1.2.3.4",
expected: 16909060,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
ip := net.ParseIP(tt.ip)
if ip == nil {
t.Fatalf("Failed to parse IP: %s", tt.ip)
}
result := ipToUint32(ip)
if result != tt.expected {
t.Errorf("ipToUint32(%s) = %d, want %d", tt.ip, result, tt.expected)
}
// Test with IPv4-mapped IPv6 address
ip6 := net.ParseIP(tt.ip).To16()
if ip6 != nil {
result6 := ipToUint32(ip6)
if result6 != tt.expected {
t.Errorf("ipToUint32(%s as IPv6) = %d, want %d", tt.ip, result6, tt.expected)
}
}
})
}
}
func TestCalculateIPv4Range(t *testing.T) {
tests := []struct {
name string
cidr string
wantStart uint32
wantEnd uint32
wantErr bool
}{
{
name: "Single IP /32",
cidr: "192.168.1.1/32",
wantStart: 3232235777,
wantEnd: 3232235777,
},
{
name: "Class C /24",
cidr: "192.168.1.0/24",
wantStart: 3232235776, // 192.168.1.0
wantEnd: 3232236031, // 192.168.1.255
},
{
name: "Class B /16",
cidr: "192.168.0.0/16",
wantStart: 3232235520, // 192.168.0.0
wantEnd: 3232301055, // 192.168.255.255
},
{
name: "Class A /8",
cidr: "10.0.0.0/8",
wantStart: 167772160, // 10.0.0.0
wantEnd: 184549375, // 10.255.255.255
},
{
name: "Entire IPv4 space /0",
cidr: "0.0.0.0/0",
wantStart: 0,
wantEnd: 4294967295,
},
{
name: "Small subnet /30",
cidr: "192.168.1.0/30",
wantStart: 3232235776, // 192.168.1.0
wantEnd: 3232235779, // 192.168.1.3
},
{
name: "Medium subnet /20",
cidr: "172.16.0.0/20",
wantStart: 2886729728, // 172.16.0.0
wantEnd: 2886733823, // 172.16.15.255
},
{
name: "Private range 172.16/12",
cidr: "172.16.0.0/12",
wantStart: 2886729728, // 172.16.0.0
wantEnd: 2887778303, // 172.31.255.255
},
{
name: "Google DNS /29",
cidr: "8.8.8.8/29",
wantStart: 134744072, // 8.8.8.8 (network is actually 8.8.8.8 with /29)
wantEnd: 134744079, // 8.8.8.15
},
{
name: "Non-zero host bits",
cidr: "192.168.1.5/24",
wantStart: 3232235776, // 192.168.1.0 (network address)
wantEnd: 3232236031, // 192.168.1.255
},
{
name: "Invalid CIDR",
cidr: "192.168.1.1/33",
wantErr: true,
},
{
name: "Invalid IP",
cidr: "256.256.256.256/24",
wantErr: true,
},
{
name: "IPv6 CIDR",
cidr: "2001:db8::/32",
wantErr: true,
},
{
name: "Empty CIDR",
cidr: "",
wantErr: true,
},
{
name: "Missing mask",
cidr: "192.168.1.1",
wantErr: true,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
start, end, err := CalculateIPv4Range(tt.cidr)
if tt.wantErr {
if err == nil {
t.Errorf("CalculateIPv4Range(%s) expected error, got nil", tt.cidr)
}
return
}
if err != nil {
t.Errorf("CalculateIPv4Range(%s) unexpected error: %v", tt.cidr, err)
return
}
if start != tt.wantStart {
t.Errorf("CalculateIPv4Range(%s) start = %d, want %d", tt.cidr, start, tt.wantStart)
}
if end != tt.wantEnd {
t.Errorf("CalculateIPv4Range(%s) end = %d, want %d", tt.cidr, end, tt.wantEnd)
}
// Verify that start <= end
if start > end {
t.Errorf("CalculateIPv4Range(%s) start (%d) > end (%d)", tt.cidr, start, end)
}
// Verify the range size matches the CIDR mask
if !tt.wantErr && tt.cidr != "" {
_, ipNet, _ := net.ParseCIDR(tt.cidr)
if ipNet != nil {
ones, bits := ipNet.Mask.Size()
expectedSize := uint32(1) << uint(bits-ones)
actualSize := end - start + 1
if actualSize != expectedSize {
t.Errorf("CalculateIPv4Range(%s) range size = %d, want %d", tt.cidr, actualSize, expectedSize)
}
}
}
})
}
}
func TestIPv4RangeIntegration(t *testing.T) {
// Test that our functions work correctly together
tests := []struct {
name string
cidr string
testIPs []string
shouldContain []bool
}{
{
name: "192.168.1.0/24",
cidr: "192.168.1.0/24",
testIPs: []string{
"192.168.1.0",
"192.168.1.1",
"192.168.1.255",
"192.168.0.255",
"192.168.2.0",
},
shouldContain: []bool{true, true, true, false, false},
},
{
name: "10.0.0.0/8",
cidr: "10.0.0.0/8",
testIPs: []string{
"10.0.0.0",
"10.255.255.255",
"10.1.2.3",
"9.255.255.255",
"11.0.0.0",
},
shouldContain: []bool{true, true, true, false, false},
},
{
name: "172.16.0.0/12",
cidr: "172.16.0.0/12",
testIPs: []string{
"172.16.0.0",
"172.31.255.255",
"172.20.1.1",
"172.15.255.255",
"172.32.0.0",
},
shouldContain: []bool{true, true, true, false, false},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
start, end, err := CalculateIPv4Range(tt.cidr)
if err != nil {
t.Fatalf("Failed to calculate range for %s: %v", tt.cidr, err)
}
for i, testIP := range tt.testIPs {
ip := net.ParseIP(testIP)
if ip == nil {
t.Fatalf("Failed to parse test IP: %s", testIP)
}
ipUint := ipToUint32(ip)
contained := ipUint >= start && ipUint <= end
if contained != tt.shouldContain[i] {
t.Errorf("IP %s in range %s: got %v, want %v", testIP, tt.cidr, contained, tt.shouldContain[i])
}
}
})
}
}
// TestConnectionPoolPragmas holds several pooled connections open at once and
// checks each one carries the per-connection settings from the DSN, plus the
// process-wide hard heap limit.
func TestConnectionPoolPragmas(t *testing.T) {
cfg := &config.Config{StateDir: t.TempDir()}
db, err := New(cfg, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ctx := context.Background()
// Hold distinct connections open simultaneously so the pool must open a new
// one (each parsing the DSN) rather than hand back the same connection.
conns := make([]*sql.Conn, 0, heldConnections)
defer func() {
for _, c := range conns {
_ = c.Close()
}
}()
for i := 0; i < heldConnections; i++ {
c, err := db.db.Conn(ctx)
if err != nil {
t.Fatalf("failed to open connection %d: %v", i, err)
}
conns = append(conns, c)
}
for i, c := range conns {
var cacheSize int
if err := c.QueryRowContext(ctx, "PRAGMA cache_size").Scan(&cacheSize); err != nil {
t.Fatalf("conn %d: failed to read cache_size: %v", i, err)
}
if cacheSize != sqliteCacheSizeKiB {
t.Errorf("conn %d: cache_size = %d, want %d", i, cacheSize, sqliteCacheSizeKiB)
}
var busyTimeout int
if err := c.QueryRowContext(ctx, "PRAGMA busy_timeout").Scan(&busyTimeout); err != nil {
t.Fatalf("conn %d: failed to read busy_timeout: %v", i, err)
}
if busyTimeout != sqliteBusyTimeoutMs {
t.Errorf("conn %d: busy_timeout = %d, want %d", i, busyTimeout, sqliteBusyTimeoutMs)
}
var tempStore int
if err := c.QueryRowContext(ctx, "PRAGMA temp_store").Scan(&tempStore); err != nil {
t.Fatalf("conn %d: failed to read temp_store: %v", i, err)
}
if tempStore == tempStoreMemory {
t.Errorf("conn %d: temp_store = %d, want anything but %d (MEMORY)", i, tempStore, tempStoreMemory)
}
var hardHeapLimit int64
if err := c.QueryRowContext(ctx, "PRAGMA hard_heap_limit").Scan(&hardHeapLimit); err != nil {
t.Fatalf("conn %d: failed to read hard_heap_limit: %v", i, err)
}
if hardHeapLimit != sqliteHardHeapLimitBytes {
t.Errorf("conn %d: hard_heap_limit = %d, want %d", i, hardHeapLimit, sqliteHardHeapLimitBytes)
}
}
}
// TestBatchWriteDuringCheckpoint reproduces issue #25. A batch write reads
// (SELECT) before it writes (INSERT/UPDATE). Under the default deferred locking
// the transaction begins as a reader and, when the maintainer's WAL checkpoint
// holds the write lock, its upgrade to writer fails immediately with "database
// is locked" without honouring busy_timeout, dropping the batch. With
// _txlock=immediate the transaction takes the write lock at BEGIN and waits, so
// no batch is dropped. The checkpoint runs without the Database mutex, exactly
// as the background maintainer does in production.
func TestBatchWriteDuringCheckpoint(t *testing.T) {
cfg := &config.Config{StateDir: t.TempDir()}
db, err := New(cfg, logger.New())
if err != nil {
t.Fatalf("failed to create database: %v", err)
}
defer func() { _ = db.Close() }()
ctx, cancel := context.WithCancel(context.Background())
var wg sync.WaitGroup
wg.Add(1)
go func() {
defer wg.Done()
for {
select {
case <-ctx.Done():
return
default:
_ = db.Checkpoint(ctx) // errors are the checkpoint's own to absorb
}
}
}()
ts := time.Now().UTC()
for i := 0; i < contentionIterations; i++ {
asns := map[int]time.Time{
i % contendedASNCount: ts,
(i % contendedASNCount) + asnSecondBand: ts,
}
if err := db.GetOrCreateASNBatch(asns); err != nil {
cancel()
wg.Wait()
t.Fatalf("batch write failed under checkpoint contention: %v", err)
}
}
cancel()
wg.Wait()
}
func BenchmarkIPToUint32(b *testing.B) {
ip := net.ParseIP("192.168.1.1")
b.ResetTimer()
for i := 0; i < b.N; i++ {
_ = ipToUint32(ip)
}
}
func BenchmarkCalculateIPv4Range(b *testing.B) {
cidr := "192.168.0.0/16"
b.ResetTimer()
for i := 0; i < b.N; i++ {
_, _, _ = CalculateIPv4Range(cidr)
}
}
internal/database/interface.go
+19 -10
View File
@@ -18,6 +18,8 @@ type Stats struct {
Peers int
FileSizeBytes int64
LiveRoutes int
IPv4Routes int
IPv6Routes int
OldestRoute *time.Time
NewestRoute *time.Time
IPv4PrefixDistribution []PrefixDistribution
@@ -27,10 +29,12 @@ type Stats struct {
// Store defines the interface for database operations. It provides methods for
// managing ASNs, prefixes, announcements, peerings, BGP peers, and live routes.
// Implementations must be safe for concurrent use.
//
//nolint:interfacebloat // the one seam handlers and tests depend on
type Store interface {
// ASN operations
GetOrCreateASN(number int, timestamp time.Time) (*ASN, error)
GetOrCreateASNBatch(asns map[int]time.Time) error
GetOrCreateASNBatch(asns map[int]SeenTimes) error
// Prefix operations
GetOrCreatePrefix(prefix string, timestamp time.Time) (*Prefix, error)
@@ -40,7 +44,7 @@ type Store interface {
RecordAnnouncement(announcement *Announcement) error
// Peering operations
RecordPeering(asA, asB int, timestamp time.Time) error
RecordPeering(asA, asB int, seen SeenTimes) error
// Statistics
GetStats() (Stats, error)
@@ -53,16 +57,17 @@ type Store interface {
// Live route operations
UpsertLiveRoute(route *LiveRoute) error
UpsertLiveRouteBatch(routes []*LiveRoute) error
DeleteLiveRoute(prefix string, originASN int, peerIP string) error
DeleteLiveRoute(prefix, peerIP string) error
DeleteLiveRouteBatch(deletions []LiveRouteDeletion) error
GetPrefixDistribution() (ipv4 []PrefixDistribution, ipv6 []PrefixDistribution, err error)
GetPrefixDistributionContext(ctx context.Context) (ipv4 []PrefixDistribution, ipv6 []PrefixDistribution, err error)
DeleteLiveRoutesOfPeer(peerIP string) (deleted int, err error)
GetPrefixDistribution() (
ipv4 []PrefixDistribution, ipv6 []PrefixDistribution, err error)
GetPrefixDistributionContext(ctx context.Context) (
ipv4 []PrefixDistribution, ipv6 []PrefixDistribution, err error)
GetLiveRouteCounts() (ipv4Count, ipv6Count int, err error)
GetLiveRouteCountsContext(ctx context.Context) (ipv4Count, ipv6Count int, err error)
// IP lookup operations
GetASInfoForIP(ip string) (*ASInfo, error)
GetASInfoForIPContext(ctx context.Context, ip string) (*ASInfo, error)
GetIPInfo(ip string) (*IPInfo, error)
GetIPInfoContext(ctx context.Context, ip string) (*IPInfo, error)
@@ -76,10 +81,14 @@ type Store interface {
GetASDetailsContext(ctx context.Context, asn int) (*ASN, []LiveRoute, error)
GetASPeers(asn int) ([]ASPeer, error)
GetASPeersContext(ctx context.Context, asn int) ([]ASPeer, error)
GetPrefixDetails(prefix string) ([]LiveRoute, error)
GetPrefixDetailsContext(ctx context.Context, prefix string) ([]LiveRoute, error)
GetPrefixDetails(prefix string) ([]LiveRoute, map[int]ASN, error)
GetPrefixDetailsContext(
ctx context.Context, prefix string,
) ([]LiveRoute, map[int]ASN, error)
GetRandomPrefixesByLength(maskLength, ipVersion, limit int) ([]LiveRoute, error)
GetRandomPrefixesByLengthContext(ctx context.Context, maskLength, ipVersion, limit int) ([]LiveRoute, error)
GetRandomPrefixesByLengthContext(
ctx context.Context, maskLength, ipVersion, limit int,
) ([]LiveRoute, error)
// Lifecycle
Close() error
internal/database/models.go
+50 -37
View File
@@ -1,4 +1,3 @@
// Package database provides SQLite storage for BGP routing data.
package database
import (
@@ -9,6 +8,8 @@ import (
// ASN represents an Autonomous System Number with its metadata including
// handle, description, WHOIS data, and first/last seen timestamps.
//
//nolint:tagliatelle // field names are the existing JSON API
type ASN struct {
ASN int `json:"asn"`
Handle string `json:"handle"`
@@ -34,41 +35,40 @@ type ASN struct {
}
// Prefix represents an IP prefix (CIDR block) with its IP version (4 or 6)
// and first/last seen timestamps.
// and the time it was first seen.
type Prefix struct {
ID uuid.UUID `json:"id"`
Prefix string `json:"prefix"`
IPVersion int `json:"ip_version"` // 4 or 6
FirstSeen time.Time `json:"first_seen"`
LastSeen time.Time `json:"last_seen"`
Prefix string
IPVersion int // 4 or 6
FirstSeen time.Time
}
// Announcement represents a BGP announcement or withdrawal event,
// containing the prefix, AS path, origin ASN, peer ASN, next hop, and timestamp.
type Announcement struct {
ID uuid.UUID `json:"id"`
PrefixID uuid.UUID `json:"prefix_id"`
PeerASN int `json:"peer_asn"`
OriginASN int `json:"origin_asn"`
Path string `json:"path"` // JSON-encoded AS path
NextHop string `json:"next_hop"`
Timestamp time.Time `json:"timestamp"`
IsWithdrawal bool `json:"is_withdrawal"`
ID uuid.UUID
PrefixID uuid.UUID
PeerASN int
OriginASN int
Path string // JSON-encoded AS path
NextHop string
Timestamp time.Time
IsWithdrawal bool
}
// ASNPeering represents a peering relationship between two ASNs,
// stored with the lower ASN as ASA and the higher as ASB.
type ASNPeering struct {
ID uuid.UUID `json:"id"`
ASA int `json:"as_a"`
ASB int `json:"as_b"`
FirstSeen time.Time `json:"first_seen"`
LastSeen time.Time `json:"last_seen"`
ID uuid.UUID
ASA int
ASB int
FirstSeen time.Time
LastSeen time.Time
}
// LiveRoute represents a route in the live routing table
//
//nolint:tagliatelle // field names are the existing JSON API
type LiveRoute struct {
ID uuid.UUID `json:"id"`
Prefix string `json:"prefix"`
MaskLength int `json:"mask_length"`
IPVersion int `json:"ip_version"`
@@ -77,28 +77,19 @@ type LiveRoute struct {
ASPath []int `json:"as_path"`
NextHop string `json:"next_hop"`
LastUpdated time.Time `json:"last_updated"`
// IPv4 range fields for fast lookups (nil for IPv6)
V4IPStart *uint32 `json:"v4_ip_start,omitempty"`
V4IPEnd *uint32 `json:"v4_ip_end,omitempty"`
}
// PrefixDistribution represents the distribution of prefixes by mask length
//
//nolint:tagliatelle // field names are the existing JSON API
type PrefixDistribution struct {
MaskLength int `json:"mask_length"`
Count int `json:"count"`
}
// ASInfo represents AS information for an IP lookup (legacy format)
type ASInfo struct {
ASN int `json:"asn"`
Handle string `json:"handle"`
Description string `json:"description"`
Prefix string `json:"prefix"`
LastUpdated time.Time `json:"last_updated"`
Age string `json:"age"`
}
// IPInfo represents comprehensive IP information for the /ip endpoint
//
//nolint:tagliatelle // field names are the existing JSON API
type IPInfo struct {
IP string `json:"ip"`
PTR []string `json:"ptr,omitempty"`
@@ -127,17 +118,39 @@ type IPInfo struct {
// LiveRouteDeletion represents parameters for deleting a live route
type LiveRouteDeletion struct {
Prefix string
OriginASN int
PeerIP string
IPVersion int
}
// PeerUpdate represents parameters for updating a peer
// SeenTimes is the earliest and the latest time one batch of updates saw an AS,
// a peer or a peering. A row the batch adds is stored first seen at First and
// last seen at Last; a row already stored keeps its first_seen and has its
// last_seen set to Last.
type SeenTimes struct {
First time.Time
Last time.Time
}
// Add makes t the earliest time of s when it is earlier than First, and the
// latest when it is later than Last. The zero SeenTimes holds no time yet, so
// the first Add sets both.
func (s *SeenTimes) Add(t time.Time) {
if s.First.IsZero() || t.Before(s.First) {
s.First = t
}
if s.Last.IsZero() || t.After(s.Last) {
s.Last = t
}
}
// PeerUpdate represents parameters for updating a peer. PeerASN and
// MessageType are those of the peer's latest update.
type PeerUpdate struct {
PeerIP string
PeerASN int
MessageType string
Timestamp time.Time
Seen SeenTimes
}
// ASNWHOISUpdate contains WHOIS data for updating an ASN record.
internal/database/schema.sql
+45 -42
View File
@@ -27,21 +27,18 @@ CREATE TABLE IF NOT EXISTS asns (
whois_updated_at DATETIME -- when we last fetched WHOIS data
);
-- IPv4 prefixes table
-- IPv4 prefixes table. Every read and write finds a row by its prefix, so the
-- rows are stored by prefix (WITHOUT ROWID) and need no index of their own.
CREATE TABLE IF NOT EXISTS prefixes_v4 (
id TEXT PRIMARY KEY,
prefix TEXT UNIQUE NOT NULL,
first_seen DATETIME NOT NULL,
last_seen DATETIME NOT NULL
);
prefix TEXT PRIMARY KEY,
first_seen DATETIME NOT NULL
) WITHOUT ROWID;
-- IPv6 prefixes table
-- IPv6 prefixes table, stored as prefixes_v4 is.
CREATE TABLE IF NOT EXISTS prefixes_v6 (
id TEXT PRIMARY KEY,
prefix TEXT UNIQUE NOT NULL,
first_seen DATETIME NOT NULL,
last_seen DATETIME NOT NULL
);
prefix TEXT PRIMARY KEY,
first_seen DATETIME NOT NULL
) WITHOUT ROWID;
CREATE TABLE IF NOT EXISTS announcements (
id TEXT PRIMARY KEY,
@@ -75,11 +72,6 @@ CREATE TABLE IF NOT EXISTS bgp_peers (
last_message_type TEXT
);
-- Indexes for prefixes_v4 table
CREATE INDEX IF NOT EXISTS idx_prefixes_v4_prefix ON prefixes_v4(prefix);
-- Indexes for prefixes_v6 table
CREATE INDEX IF NOT EXISTS idx_prefixes_v6_prefix ON prefixes_v6(prefix);
CREATE INDEX IF NOT EXISTS idx_announcements_timestamp ON announcements(timestamp);
CREATE INDEX IF NOT EXISTS idx_announcements_prefix_id ON announcements(prefix_id);
CREATE INDEX IF NOT EXISTS idx_announcements_peer_asn ON announcements(peer_asn);
@@ -97,9 +89,14 @@ CREATE INDEX IF NOT EXISTS idx_bgp_peers_asn ON bgp_peers(peer_asn);
CREATE INDEX IF NOT EXISTS idx_bgp_peers_last_seen ON bgp_peers(last_seen);
CREATE INDEX IF NOT EXISTS idx_bgp_peers_ip ON bgp_peers(peer_ip);
-- IPv4 routing table maintained by PrefixHandler
-- IPv4 routing table maintained by PrefixHandler. A route write changes the
-- table and its entry in every index on a column the write sets, so the table
-- has only the indexes that the queries named with them use.
CREATE TABLE IF NOT EXISTS live_routes_v4 (
id TEXT PRIMARY KEY,
-- The row id, which the prefix-length pages' sample draws through the
-- mask_length index. Declared as a column, it stays the same for as long
-- as the route is live, also through a VACUUM.
id INTEGER PRIMARY KEY,
prefix TEXT NOT NULL,
mask_length INTEGER NOT NULL, -- CIDR mask length (0-32)
origin_asn INTEGER NOT NULL,
@@ -107,15 +104,15 @@ CREATE TABLE IF NOT EXISTS live_routes_v4 (
as_path TEXT NOT NULL, -- JSON array
next_hop TEXT NOT NULL,
last_updated DATETIME NOT NULL,
-- IPv4 range columns for fast lookups
ip_start INTEGER NOT NULL, -- Start of IPv4 range as 32-bit unsigned int
ip_end INTEGER NOT NULL, -- End of IPv4 range as 32-bit unsigned int
UNIQUE(prefix, origin_asn, peer_ip)
-- A peer has at most one route for a prefix. This key finds the route a
-- write updates or deletes and, peer_ip coming first, all of a peer's
-- routes when its BGP session goes down.
UNIQUE(peer_ip, prefix)
);
-- IPv6 routing table maintained by PrefixHandler
-- IPv6 routing table maintained by PrefixHandler, as live_routes_v4.
CREATE TABLE IF NOT EXISTS live_routes_v6 (
id TEXT PRIMARY KEY,
id INTEGER PRIMARY KEY,
prefix TEXT NOT NULL,
mask_length INTEGER NOT NULL, -- CIDR mask length (0-128)
origin_asn INTEGER NOT NULL,
@@ -123,24 +120,30 @@ CREATE TABLE IF NOT EXISTS live_routes_v6 (
as_path TEXT NOT NULL, -- JSON array
next_hop TEXT NOT NULL,
last_updated DATETIME NOT NULL,
-- Note: IPv6 doesn't use integer range columns
UNIQUE(prefix, origin_asn, peer_ip)
UNIQUE(peer_ip, prefix)
);
-- Indexes for live_routes_v4 table
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_prefix ON live_routes_v4(prefix);
-- Indexes for both live route tables, and the queries that use each.
--
-- prefix, last_updated: a prefix's routes, for the prefix page and its API
-- answer, the IP lookup and its peer count, and the check whether a prefix
-- still has a live route; and a prefix's newest route for the prefix-length
-- pages and the AS page, in one search whatever the number of peers.
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_prefix_last_updated ON live_routes_v4(prefix, last_updated);
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_prefix_last_updated ON live_routes_v6(prefix, last_updated);
-- mask_length: the prefix-length pages' random sample, the row id range of a
-- mask length and the first route of it at or after a row id.
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_mask_length ON live_routes_v4(mask_length);
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_origin_asn ON live_routes_v4(origin_asn);
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_last_updated ON live_routes_v4(last_updated);
-- Indexes for IPv4 range queries
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_ip_range ON live_routes_v4(ip_start, ip_end);
-- Index to optimize prefix distribution queries
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_mask_prefix ON live_routes_v4(mask_length, prefix);
-- Indexes for live_routes_v6 table
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_prefix ON live_routes_v6(prefix);
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_mask_length ON live_routes_v6(mask_length);
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_origin_asn ON live_routes_v6(origin_asn);
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_last_updated ON live_routes_v6(last_updated);
-- Index to optimize prefix distribution queries
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_mask_prefix ON live_routes_v6(mask_length, prefix);
-- mask_length, prefix: the prefix-length pages' list of a mask length's
-- prefixes in text order, and the prefix distribution read at startup.
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_mask_prefix ON live_routes_v4(mask_length, prefix);
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_mask_prefix ON live_routes_v6(mask_length, prefix);
-- origin_asn, prefix: the prefixes an AS originates, each in one search
-- whatever the number of peers, for the AS page and its API answer.
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_origin_prefix ON live_routes_v4(origin_asn, prefix);
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_origin_prefix ON live_routes_v6(origin_asn, prefix);
-- last_updated: the oldest and newest route times in /api/v1/stats and
-- /status.json.
CREATE INDEX IF NOT EXISTS idx_live_routes_v4_last_updated ON live_routes_v4(last_updated);
CREATE INDEX IF NOT EXISTS idx_live_routes_v6_last_updated ON live_routes_v6(last_updated);
internal/database/slowquery.go
+26 -44
View File
@@ -18,30 +18,12 @@ func logSlowQuery(logger *logger.Logger, query string, start time.Time) {
}
}
// queryRow wraps QueryRow with slow query logging
// nolint:unused // kept for consistency with other query wrappers
func (d *Database) queryRow(query string, args ...interface{}) *sql.Row {
// exec runs a statement on the writing connection, with slow query logging
func (d *Database) exec(query string, args ...any) error {
start := time.Now()
defer logSlowQuery(d.logger, query, start)
return d.db.QueryRow(query, args...)
}
// query wraps Query with slow query logging
// nolint:unused // kept for future use to ensure all queries go through slow query logging
func (d *Database) query(query string, args ...interface{}) (*sql.Rows, error) {
start := time.Now()
defer logSlowQuery(d.logger, query, start)
return d.db.Query(query, args...)
}
// exec wraps Exec with slow query logging
func (d *Database) exec(query string, args ...interface{}) error {
start := time.Now()
defer logSlowQuery(d.logger, query, start)
_, err := d.db.Exec(query, args...)
_, err := d.writer.ExecContext(context.Background(), query, args...)
return err
}
@@ -49,51 +31,51 @@ func (d *Database) exec(query string, args ...interface{}) error {
// loggingTx wraps sql.Tx to log slow queries
type loggingTx struct {
*sql.Tx
logger *logger.Logger
}
// QueryRow wraps sql.Tx.QueryRow to log slow queries
func (tx *loggingTx) QueryRow(query string, args ...interface{}) *sql.Row {
func (tx *loggingTx) QueryRow(query string, args ...any) *sql.Row {
start := time.Now()
defer logSlowQuery(tx.logger, query, start)
return tx.Tx.QueryRow(query, args...)
}
// Query wraps sql.Tx.Query to log slow queries
func (tx *loggingTx) Query(query string, args ...interface{}) (*sql.Rows, error) {
start := time.Now()
defer logSlowQuery(tx.logger, query, start)
return tx.Tx.Query(query, args...)
}
// QueryContext wraps sql.Tx.QueryContext to log slow queries
func (tx *loggingTx) QueryContext(ctx context.Context, query string, args ...interface{}) (*sql.Rows, error) {
start := time.Now()
defer logSlowQuery(tx.logger, query, start)
return tx.Tx.QueryContext(ctx, query, args...)
return tx.Tx.QueryRowContext(context.Background(), query, args...)
}
// QueryRowContext wraps sql.Tx.QueryRowContext to log slow queries
func (tx *loggingTx) QueryRowContext(ctx context.Context, query string, args ...interface{}) *sql.Row {
func (tx *loggingTx) QueryRowContext(
ctx context.Context, query string, args ...any,
) *sql.Row {
start := time.Now()
defer logSlowQuery(tx.logger, query, start)
return tx.Tx.QueryRowContext(ctx, query, args...)
}
// Exec wraps sql.Tx.Exec to log slow queries
func (tx *loggingTx) Exec(query string, args ...interface{}) (sql.Result, error) {
// QueryContext wraps sql.Tx.QueryContext to log slow queries
func (tx *loggingTx) QueryContext(
ctx context.Context, query string, args ...any,
) (*sql.Rows, error) {
start := time.Now()
defer logSlowQuery(tx.logger, query, start)
return tx.Tx.Exec(query, args...)
//nolint:sqlclosecheck // returns the rows, which the caller closes
return tx.Tx.QueryContext(ctx, query, args...)
}
// Exec wraps sql.Tx.Exec to log slow queries
func (tx *loggingTx) Exec(query string, args ...any) (sql.Result, error) {
start := time.Now()
defer logSlowQuery(tx.logger, query, start)
return tx.Tx.ExecContext(context.Background(), query, args...)
}
// ExecContext wraps sql.Tx.ExecContext to log slow queries
func (tx *loggingTx) ExecContext(ctx context.Context, query string, args ...interface{}) (sql.Result, error) {
func (tx *loggingTx) ExecContext(
ctx context.Context, query string, args ...any,
) (sql.Result, error) {
start := time.Now()
defer logSlowQuery(tx.logger, query, start)
internal/database/utils.go
+14
View File
@@ -1,6 +1,8 @@
package database
import (
"fmt"
"net"
"strings"
"github.com/google/uuid"
@@ -18,3 +20,15 @@ func detectIPVersion(prefix string) int {
return ipVersionV4
}
// prefixMaskLength returns the mask length of a prefix such as 192.0.2.0/24.
func prefixMaskLength(prefix string) (int, error) {
_, network, err := net.ParseCIDR(prefix)
if err != nil {
return 0, fmt.Errorf("invalid prefix %s: %w", prefix, err)
}
maskLength, _ := network.Mask.Size()
return maskLength, nil
}
internal/logger/logger.go
+1
View File
@@ -74,6 +74,7 @@ func getSourceAttrs() []slog.Attr {
// Get the function name
fn := runtime.FuncForPC(pc)
var funcName string
if fn != nil {
funcName = filepath.Base(fn.Name())
internal/metrics/metrics.go
+26 -15
View File
@@ -60,17 +60,23 @@ func New() *Tracker {
}
}
// SetConnected updates the connection status
// SetConnected updates the connection status. Every connection after the
// first counts one reconnect.
func (t *Tracker) SetConnected(connected bool) {
wasConnected := t.isConnected.Swap(connected)
if connected {
t.mu.Lock()
t.connectedSince = time.Now()
t.mu.Unlock()
// Increment reconnect count (but not for the initial connection)
if wasConnected || t.reconnectCount.Load() > 0 {
t.reconnectCount.Add(1)
}
t.isConnected.Store(connected)
if !connected {
return
}
t.mu.Lock()
// connectedSince is zero until the first connection.
connectedBefore := !t.connectedSince.IsZero()
t.connectedSince = time.Now()
t.mu.Unlock()
if connectedBefore {
t.reconnectCount.Add(1)
}
}
@@ -115,9 +121,11 @@ func (t *Tracker) GetStreamMetrics() StreamMetrics {
if msgCount >= 0 {
totalMessages = uint64(msgCount)
}
if byteCount >= 0 {
totalBytes = uint64(byteCount)
}
if wireByteCount >= 0 {
totalWireBytes = uint64(wireByteCount)
}
@@ -160,8 +168,8 @@ func (t *Tracker) RecordWithdrawal() {
// SetBGPPeerCount updates the current BGP peer count
func (t *Tracker) SetBGPPeerCount(count int) {
// BGP peer count is always small (< 1000), so int32 is safe
if count > 0 && count < 1<<31 {
t.bgpPeerCount.Store(int32(count)) //nolint:gosec // count is validated
if count >= 0 && count < 1<<31 {
t.bgpPeerCount.Store(int32(count))
}
}
@@ -209,7 +217,8 @@ type StreamMetrics struct {
TotalMessages uint64
// TotalBytes is the total number of decompressed bytes received since startup
TotalBytes uint64
// TotalWireBytes is the total number of bytes received on the wire (before decompression)
// TotalWireBytes is the total number of bytes received on the wire (before
// decompression)
TotalWireBytes uint64
// ConnectedSince is the time when the current connection was established
ConnectedSince time.Time
@@ -217,9 +226,11 @@ type StreamMetrics struct {
Connected bool
// MessagesPerSec is the rate of messages received per second (1-minute average)
MessagesPerSec float64
// BitsPerSec is the rate of decompressed bits received per second (1-minute average)
// BitsPerSec is the rate of decompressed bits received per second (1-minute
// average)
BitsPerSec float64
// WireBitsPerSec is the rate of bits received on the wire per second (1-minute average)
// WireBitsPerSec is the rate of bits received on the wire per second
// (1-minute average)
WireBitsPerSec float64
// ReconnectCount is the number of reconnections since startup
ReconnectCount uint64
internal/metrics/metrics_test.go
+42
View File
@@ -0,0 +1,42 @@
package metrics_test
import (
"testing"
"git.eeqj.de/sneak/routewatch/internal/metrics"
)
// TestReconnectCount connects three times with a disconnect between each, the
// way the streamer does, and expects every connection after the first to count
// one reconnect.
func TestReconnectCount(t *testing.T) {
t.Parallel()
tracker := metrics.New()
tracker.SetConnected(true)
tracker.SetConnected(false)
tracker.SetConnected(true)
tracker.SetConnected(false)
tracker.SetConnected(true)
const wantReconnects = 2
if got := tracker.GetReconnectCount(); got != wantReconnects {
t.Fatalf("reconnect count = %d, want %d", got, wantReconnects)
}
}
// TestBGPPeerCountDropsToZero sets a peer count of 1 and then 0, the way the
// streamer does when the last peer's session ends, and expects 0 back.
func TestBGPPeerCountDropsToZero(t *testing.T) {
t.Parallel()
tracker := metrics.New()
tracker.SetBGPPeerCount(1)
tracker.SetBGPPeerCount(0)
if got := tracker.GetBGPPeerCount(); got != 0 {
t.Fatalf("peer count = %d after setting 0, want 0", got)
}
}
internal/ristypes/ris.go
+76 -25
View File
@@ -1,4 +1,5 @@
// Package ristypes defines the data structures for RIS Live BGP messages and announcements.
// Package ristypes defines the data structures for RIS Live BGP messages and
// announcements.
package ristypes
import (
@@ -6,46 +7,90 @@ import (
"time"
)
// ASPath represents a BGP AS path as a slice of AS numbers.
// It handles JSON unmarshaling of both simple arrays and nested AS sets,
// flattening any nested structures into a single sequence of AS numbers.
type ASPath []int
// Message types and the session state the handlers act on, as the feed writes
// them in a message's type and state fields.
const (
// MessageTypeUpdate is a BGP UPDATE: announcements and withdrawals.
MessageTypeUpdate = "UPDATE"
// MessageTypeState is a change in the state of the BGP session between
// the collector and the peer.
MessageTypeState = "STATE"
// StateDown is the state of a STATE message whose session has ended: the
// collector no longer has any of the peer's routes, and sends no
// withdrawals for them.
StateDown = "down"
)
// ASPath is a BGP AS path as the feed writes it: an array of AS numbers in
// which an AS set is an array of its members, such as [1, 2, [3, 4]].
type ASPath struct {
// Sequence is the path's AS numbers in order. A set of one member that
// ends the path counts as that AS and is its last element. The members of
// a set anywhere else are flattened into it.
Sequence []int
// Set holds the members of a set of several members that ends the path.
// Any of them may be the origin, and they are not neighbours of each
// other or of the AS before them.
Set []int
}
// Origin returns the AS that originated the route: the last AS of the
// sequence, or 0 (none) when the path ends in a set of several members or is
// empty.
func (p *ASPath) Origin() int {
if len(p.Set) > 0 || len(p.Sequence) == 0 {
return 0
}
return p.Sequence[len(p.Sequence)-1]
}
// UnmarshalJSON implements the json.Unmarshaler interface for ASPath.
// It handles both simple integer arrays [1, 2, 3] and nested AS sets
// like [1, [2, 3], 4], flattening them into a single slice of integers.
func (p *ASPath) UnmarshalJSON(data []byte) error {
// First try to unmarshal as a simple array of integers
var simple []int
if err := json.Unmarshal(data, &simple); err == nil {
*p = ASPath(simple)
// Most paths have no AS set and decode as a plain array of integers.
var sequence []int
err := json.Unmarshal(data, &sequence)
if err == nil {
*p = ASPath{Sequence: sequence}
return nil
}
// If that fails, unmarshal as array of interfaces and flatten
var raw []interface{}
if err := json.Unmarshal(data, &raw); err != nil {
var raw []any
err = json.Unmarshal(data, &raw)
if err != nil {
return err
}
// Flatten the array
result := make([]int, 0)
for _, item := range raw {
path := ASPath{Sequence: make([]int, 0, len(raw))}
for i, item := range raw {
switch v := item.(type) {
case float64:
result = append(result, int(v))
case []interface{}:
// Nested array - flatten it
for _, nested := range v {
if num, ok := nested.(float64); ok {
result = append(result, int(num))
path.Sequence = append(path.Sequence, int(v))
case []any:
set := make([]int, 0, len(v))
for _, member := range v {
if num, ok := member.(float64); ok {
set = append(set, int(num))
}
}
if i == len(raw)-1 && len(set) > 1 {
path.Set = set
} else {
path.Sequence = append(path.Sequence, set...)
}
}
}
*p = ASPath(result)
*p = path
return nil
}
@@ -62,6 +107,8 @@ type RISLiveMessage struct {
// It contains metadata about the BGP session (peer, ASN, host) along with
// the actual BGP update data including AS path, communities, announcements,
// and withdrawals.
//
//nolint:tagliatelle // field names are the RIS Live feed's
type RISMessage struct {
Type string `json:"type"`
Timestamp float64 `json:"timestamp"`
@@ -73,7 +120,7 @@ type RISMessage struct {
RRC string `json:"rrc,omitempty"`
MrtTime float64 `json:"mrt_time,omitempty"`
SocketTime float64 `json:"socket_time,omitempty"`
Path ASPath `json:"path,omitempty"`
Path ASPath `json:"path,omitzero"`
// Community and Raw are present in the feed but read by no handler.
// They are the largest fields on a message that lives in up to four
// handler queues, so json:"-" keeps them out of the decoded message
@@ -85,11 +132,15 @@ type RISMessage struct {
Announcements []RISAnnouncement `json:"announcements,omitempty"`
Withdrawals []string `json:"withdrawals,omitempty"`
Raw string `json:"-"`
// State is set on STATE messages: the session's new state, such as StateDown.
State string `json:"state,omitempty"`
}
// RISAnnouncement represents a BGP route announcement within a RIS message.
// It contains the next hop IP address and the list of prefixes being announced
// via that next hop.
//
//nolint:tagliatelle // field names are the RIS Live feed's
type RISAnnouncement struct {
NextHop string `json:"next_hop"`
Prefixes []string `json:"prefixes"`
internal/ristypes/ris_test.go
+88 -8
View File
@@ -1,10 +1,13 @@
package ristypes
package ristypes_test
import (
"bufio"
"encoding/json"
"os"
"slices"
"testing"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
)
// messageExamplesPath is the captured RIS Live feed used as decode fixtures,
@@ -15,44 +18,58 @@ const messageExamplesPath = "../../docs/message-examples.json"
// streamer does and checks that the fields no handler reads (Community, Raw)
// stay empty while the fields handlers use (Path, Announcements) still decode.
func TestDecodeDropsCommunityAndRaw(t *testing.T) {
t.Parallel()
f, err := os.Open(messageExamplesPath)
if err != nil {
t.Fatalf("open fixtures: %v", err)
}
defer f.Close()
defer func() {
_ = f.Close()
}()
var messages, withPath, withAnnouncements int
scanner := bufio.NewScanner(f)
scanner.Buffer(make([]byte, 0, 64*1024), 1024*1024)
for scanner.Scan() {
line := scanner.Bytes()
if len(line) == 0 {
continue
}
var wrapper RISLiveMessage
if err := json.Unmarshal(line, &wrapper); err != nil {
var wrapper ristypes.RISLiveMessage
err := json.Unmarshal(line, &wrapper)
if err != nil {
t.Fatalf("unmarshal message %d: %v", messages+1, err)
}
messages++
msg := wrapper.Data
if msg.Community != nil {
t.Errorf("message %d: Community decoded, want empty: %v", messages, msg.Community)
t.Errorf("message %d: Community decoded, want empty: %v",
messages, msg.Community)
}
if msg.Raw != "" {
t.Errorf("message %d: Raw decoded, want empty", messages)
}
if len(msg.Path) > 0 {
if len(msg.Path.Sequence) > 0 {
withPath++
}
if len(msg.Announcements) > 0 {
withAnnouncements++
}
}
if err := scanner.Err(); err != nil {
err = scanner.Err()
if err != nil {
t.Fatalf("scan fixtures: %v", err)
}
@@ -62,6 +79,69 @@ func TestDecodeDropsCommunityAndRaw(t *testing.T) {
// The fixtures include announcement messages; a used field must still decode,
// otherwise an empty Community/Raw would prove nothing.
if withPath == 0 || withAnnouncements == 0 {
t.Fatalf("used fields did not decode: withPath=%d withAnnouncements=%d", withPath, withAnnouncements)
t.Fatalf("used fields did not decode: withPath=%d withAnnouncements=%d",
withPath, withAnnouncements)
}
}
// TestASPathKeepsATrailingSetApart decodes AS paths as the feed writes them, an
// AS set as an array inside the path. A set of several members that ends the
// path is kept apart from the sequence, and the path has no origin, since any
// member may be it. A set of one member counts as that AS, and a set anywhere
// else is flattened into the sequence.
func TestASPathKeepsATrailingSetApart(t *testing.T) {
t.Parallel()
tests := []struct {
path string
wantSequence []int
wantSet []int
wantOrigin int
}{
{
path: "[64496, 64497]",
wantSequence: []int{64496, 64497},
wantOrigin: 64497,
},
{
path: "[64496, 64497, [64498, 64499]]",
wantSequence: []int{64496, 64497},
wantSet: []int{64498, 64499},
},
{
path: "[64496, 64497, [64498]]",
wantSequence: []int{64496, 64497, 64498},
wantOrigin: 64498,
},
{
path: "[64496, [64497, 64498], 64499]",
wantSequence: []int{64496, 64497, 64498, 64499},
wantOrigin: 64499,
},
{path: "[]"},
}
for _, tt := range tests {
t.Run(tt.path, func(t *testing.T) {
t.Parallel()
var got ristypes.ASPath
err := json.Unmarshal([]byte(tt.path), &got)
if err != nil {
t.Fatalf("unmarshal %s: %v", tt.path, err)
}
if !slices.Equal(got.Sequence, tt.wantSequence) ||
!slices.Equal(got.Set, tt.wantSet) {
t.Errorf("%s decoded to sequence %v and set %v, want %v and %v",
tt.path, got.Sequence, got.Set, tt.wantSequence, tt.wantSet)
}
origin := got.Origin()
if origin != tt.wantOrigin {
t.Errorf("origin of %s = %d, want %d", tt.path, origin, tt.wantOrigin)
}
})
}
}
internal/routewatch/app.go
+70 -29
View File
@@ -1,10 +1,13 @@
// Package routewatch contains the primary RouteWatch type that represents a running instance
// of the application and contains pointers to its core dependencies, and is responsible for initialization.
// Package routewatch contains the primary RouteWatch type that represents a
// running instance of the application and contains pointers to its core
// dependencies, and is responsible for initialization. It also contains the
// ASN WHOIS fetcher for background updates and the database maintainer for
// background maintenance tasks.
package routewatch
import (
"context"
"fmt"
"errors"
"sync"
"time"
@@ -18,6 +21,10 @@ import (
"go.uber.org/fx"
)
// errNonBatchedNotImplemented is returned by Run when batched database writes
// are disabled, since only the batched handlers exist.
var errNonBatchedNotImplemented = errors.New("non-batched handlers not implemented")
// Dependencies contains all dependencies for RouteWatch
type Dependencies struct {
fx.In
@@ -68,8 +75,10 @@ func (rw *RouteWatch) Run(ctx context.Context) error {
// Apply runtime limit if specified
if rw.maxRuntime > 0 {
var cancel context.CancelFunc
ctx, cancel = context.WithTimeout(ctx, rw.maxRuntime)
defer cancel()
rw.logger.Info("Running with time limit", "max_runtime", rw.maxRuntime)
}
@@ -96,29 +105,31 @@ func (rw *RouteWatch) Run(ctx context.Context) error {
// Non-batched handlers not implemented yet
rw.logger.Error("Non-batched handlers not implemented")
return fmt.Errorf("non-batched handlers not implemented")
return errNonBatchedNotImplemented
}
// No longer need routing table handler - PrefixHandler maintains live_routes table
// Start streaming
if err := rw.streamer.Start(); err != nil {
err := rw.streamer.Start(ctx)
if err != nil {
return err
}
// Start HTTP server
if err := rw.server.Start(); err != nil {
err = rw.server.Start(ctx)
if err != nil {
return err
}
// Start ASN WHOIS fetcher for background updates
rw.asnFetcher = NewASNFetcher(rw.db, rw.logger.Logger)
rw.asnFetcher.Start()
rw.asnFetcher.Start(ctx)
rw.server.SetASNFetcher(rw.asnFetcher)
// Start database maintenance goroutine
rw.dbMaintainer = NewDBMaintainer(rw.db, rw.logger.Logger)
rw.dbMaintainer.Start()
rw.dbMaintainer.Start(ctx)
// Wait for context cancellation
<-ctx.Done()
@@ -126,56 +137,66 @@ func (rw *RouteWatch) Run(ctx context.Context) error {
return nil
}
// Shutdown performs graceful shutdown of all services
func (rw *RouteWatch) Shutdown() {
// Shutdown stops all services in order: the streamer, the handlers, which
// write what they hold, the WHOIS fetcher, the database maintainer and the
// HTTP server, and last the database, which it checkpoints and closes. Each
// stage waits at most until ctx ends, so Shutdown returns soon after: a
// handler logs what it could not write by then, and the HTTP server closes
// the connections of requests still running.
func (rw *RouteWatch) Shutdown(ctx context.Context) {
rw.mu.Lock()
if rw.shutdown {
rw.mu.Unlock()
return
}
rw.shutdown = true
rw.mu.Unlock()
// Stop batched handlers first to flush remaining batches
// Stop the streamer first: it returns once the handlers have been handed
// every message it queued, or when ctx ends. Then each handler writes
// what it holds.
rw.streamer.Stop(ctx)
if rw.dbHandler != nil {
rw.logger.Info("Flushing database handler")
rw.dbHandler.Stop()
rw.dbHandler.Stop(ctx)
}
if rw.peerHandler != nil {
rw.logger.Info("Flushing peer handler")
rw.peerHandler.Stop()
rw.peerHandler.Stop(ctx)
}
if rw.prefixHandler != nil {
rw.logger.Info("Flushing prefix handler")
rw.prefixHandler.Stop()
rw.prefixHandler.Stop(ctx)
}
if rw.peeringHandler != nil {
rw.logger.Info("Flushing peering handler")
rw.peeringHandler.Stop()
rw.peeringHandler.Stop(ctx)
}
// Stop ASN WHOIS fetcher
if rw.asnFetcher != nil {
rw.asnFetcher.Stop()
rw.asnFetcher.Stop(ctx)
}
// Stop database maintainer
if rw.dbMaintainer != nil {
rw.dbMaintainer.Stop()
rw.dbMaintainer.Stop(ctx)
}
// Stop services
rw.streamer.Stop()
// Stop HTTP server with a timeout
const serverStopTimeout = 5 * time.Second
stopCtx, cancel := context.WithTimeout(context.Background(), serverStopTimeout)
defer cancel()
if err := rw.server.Stop(stopCtx); err != nil {
rw.logger.Error("Failed to stop HTTP server gracefully", "error", err)
err := rw.server.Stop(ctx)
if err != nil {
rw.logger.Error("Stopped the HTTP server before every request finished",
"error", err)
}
rw.closeDatabase(ctx)
// Log final metrics
metrics := rw.streamer.GetMetrics()
rw.logger.Info("Final metrics",
@@ -185,12 +206,32 @@ func (rw *RouteWatch) Shutdown() {
"bits_per_sec", metrics.BitsPerSec,
"duration", time.Since(metrics.ConnectedSince),
)
}
// closeDatabase checkpoints and closes the database, unless ctx has ended: a
// write may then still be running, and both would wait for it. SQLite rolls
// back a write cut off by the exit when the database next opens.
func (rw *RouteWatch) closeDatabase(ctx context.Context) {
if ctx.Err() != nil {
rw.logger.Error("Stop deadline passed; the database is not closed")
return
}
err := rw.db.Checkpoint(ctx)
if err != nil {
rw.logger.Error("Failed to checkpoint the database", "error", err)
}
err = rw.db.Close()
if err != nil {
rw.logger.Error("Failed to close the database", "error", err)
}
}
// getModule provides all fx dependencies
func getModule() fx.Option {
return fx.Options(
func getModule() []fx.Option {
return []fx.Option{
fx.Provide(
logger.New,
config.New,
@@ -203,5 +244,5 @@ func getModule() fx.Option {
server.New,
New,
),
)
}
}
internal/routewatch/app_integration_internal_test.go
+834
View File
@@ -0,0 +1,834 @@
package routewatch
import (
"context"
"fmt"
"io"
"net"
"net/http"
"net/http/httptest"
"slices"
"strconv"
"strings"
"sync"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/config"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/metrics"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
"git.eeqj.de/sneak/routewatch/internal/server"
"git.eeqj.de/sneak/routewatch/internal/streamer"
)
// mockStore is a mock implementation of database.Store for testing
type mockStore struct {
mu sync.Mutex
// Counters for tracking calls
ASNCount int
PrefixCount int
PeeringCount int
RouteCount int
WithdrawalCount int
// Track unique items
ASNs map[int]*database.ASN
Prefixes map[string]*database.Prefix
Peerings map[string]bool // key is "from_to"
Routes map[string]bool // key is "prefix_origin_peer"
// Track IP versions
IPv4Prefixes int
IPv6Prefixes int
}
// newMockStore creates a new mock store
func newMockStore() *mockStore {
return &mockStore{
ASNs: make(map[int]*database.ASN),
Prefixes: make(map[string]*database.Prefix),
Peerings: make(map[string]bool),
Routes: make(map[string]bool),
}
}
// GetOrCreateASN mock implementation
func (m *mockStore) GetOrCreateASN(
number int, timestamp time.Time,
) (*database.ASN, error) {
m.mu.Lock()
defer m.mu.Unlock()
if asn, exists := m.ASNs[number]; exists {
asn.LastSeen = timestamp
return asn, nil
}
asn := &database.ASN{
ASN: number,
FirstSeen: timestamp,
LastSeen: timestamp,
}
m.ASNs[number] = asn
m.ASNCount++
return asn, nil
}
// UpdatePrefixesBatch mock implementation
func (m *mockStore) UpdatePrefixesBatch(prefixes map[string]time.Time) error {
m.mu.Lock()
defer m.mu.Unlock()
for prefix, timestamp := range prefixes {
if _, exists := m.Prefixes[prefix]; !exists {
const (
ipVersionV4 = 4
ipVersionV6 = 6
)
ipVersion := ipVersionV4
if strings.Contains(prefix, ":") {
ipVersion = ipVersionV6
}
m.Prefixes[prefix] = &database.Prefix{
Prefix: prefix,
IPVersion: ipVersion,
FirstSeen: timestamp,
}
}
}
return nil
}
// GetOrCreatePrefix mock implementation
func (m *mockStore) GetOrCreatePrefix(
prefix string, timestamp time.Time,
) (*database.Prefix, error) {
m.mu.Lock()
defer m.mu.Unlock()
if p, exists := m.Prefixes[prefix]; exists {
return p, nil
}
const (
ipVersionV4 = 4
ipVersionV6 = 6
)
ipVersion := ipVersionV4
if strings.Contains(prefix, ":") {
ipVersion = ipVersionV6
}
p := &database.Prefix{
Prefix: prefix,
IPVersion: ipVersion,
FirstSeen: timestamp,
}
m.Prefixes[prefix] = p
m.PrefixCount++
if ipVersion == ipVersionV4 {
m.IPv4Prefixes++
} else {
m.IPv6Prefixes++
}
return p, nil
}
// RecordAnnouncement mock implementation
func (m *mockStore) RecordAnnouncement(_ *database.Announcement) error {
// Not tracking announcements in detail for now
return nil
}
// RecordPeering mock implementation
func (m *mockStore) RecordPeering(asA, asB int, _ database.SeenTimes) error {
m.mu.Lock()
defer m.mu.Unlock()
// Normalize
if asA > asB {
asA, asB = asB, asA
}
key := fmt.Sprintf("%d_%d", asA, asB)
if !m.Peerings[key] {
m.Peerings[key] = true
m.PeeringCount++
}
return nil
}
// UpdatePeer mock implementation
func (m *mockStore) UpdatePeer(_ string, _ int, _ string, _ time.Time) error {
// Simple mock - just return nil
return nil
}
// Close mock implementation
func (m *mockStore) Close() error {
return nil
}
// GetStats returns statistics about the mock store
func (m *mockStore) GetStats() (database.Stats, error) {
m.mu.Lock()
defer m.mu.Unlock()
return database.Stats{
ASNs: len(m.ASNs),
Prefixes: len(m.Prefixes),
IPv4Prefixes: m.IPv4Prefixes,
IPv6Prefixes: m.IPv6Prefixes,
Peerings: m.PeeringCount,
Peers: 10, // Mock peer count
}, nil
}
// GetStatsContext returns statistics about the mock store with context support
func (m *mockStore) GetStatsContext(_ context.Context) (database.Stats, error) {
return m.GetStats()
}
// UpsertLiveRoute mock implementation
func (m *mockStore) UpsertLiveRoute(_ *database.LiveRoute) error {
// Simple mock - just return nil
return nil
}
// DeleteLiveRoute mock implementation
func (m *mockStore) DeleteLiveRoute(_, _ string) error {
// Simple mock - just return nil
return nil
}
// GetPrefixDistribution mock implementation
func (m *mockStore) GetPrefixDistribution() (
[]database.PrefixDistribution, []database.PrefixDistribution, error,
) {
// Return empty distributions for now
return nil, nil, nil
}
// GetPrefixDistributionContext mock implementation with context support
func (m *mockStore) GetPrefixDistributionContext(_ context.Context) (
[]database.PrefixDistribution, []database.PrefixDistribution, error,
) {
return m.GetPrefixDistribution()
}
// GetLiveRouteCounts mock implementation
func (m *mockStore) GetLiveRouteCounts() (int, int, error) {
// Return mock counts
return m.RouteCount / 2, m.RouteCount / 2, nil
}
// GetLiveRouteCountsContext mock implementation with context support
func (m *mockStore) GetLiveRouteCountsContext(_ context.Context) (int, int, error) {
return m.GetLiveRouteCounts()
}
// GetASDetails mock implementation
func (m *mockStore) GetASDetails(asn int) (*database.ASN, []database.LiveRoute, error) {
m.mu.Lock()
defer m.mu.Unlock()
// Check if ASN exists
if asnInfo, exists := m.ASNs[asn]; exists {
// Return empty prefixes for now
return asnInfo, []database.LiveRoute{}, nil
}
return nil, nil, database.ErrNoRoute
}
// GetASDetailsContext mock implementation with context support
func (m *mockStore) GetASDetailsContext(
_ context.Context, asn int,
) (*database.ASN, []database.LiveRoute, error) {
return m.GetASDetails(asn)
}
// GetPrefixDetails mock implementation
func (m *mockStore) GetPrefixDetails(
_ string,
) ([]database.LiveRoute, map[int]database.ASN, error) {
// Return empty routes for now
return []database.LiveRoute{}, map[int]database.ASN{}, nil
}
// GetPrefixDetailsContext mock implementation with context support
func (m *mockStore) GetPrefixDetailsContext(
_ context.Context, prefix string,
) ([]database.LiveRoute, map[int]database.ASN, error) {
return m.GetPrefixDetails(prefix)
}
func (m *mockStore) GetRandomPrefixesByLength(
_, _, _ int,
) ([]database.LiveRoute, error) {
// Return empty routes for now
return []database.LiveRoute{}, nil
}
// GetRandomPrefixesByLengthContext mock implementation with context support
func (m *mockStore) GetRandomPrefixesByLengthContext(
_ context.Context, maskLength, ipVersion, limit int,
) ([]database.LiveRoute, error) {
return m.GetRandomPrefixesByLength(maskLength, ipVersion, limit)
}
// GetASPeers mock implementation
func (m *mockStore) GetASPeers(_ int) ([]database.ASPeer, error) {
// Return empty peers for now
return []database.ASPeer{}, nil
}
// GetASPeersContext mock implementation with context support
func (m *mockStore) GetASPeersContext(
_ context.Context, asn int,
) ([]database.ASPeer, error) {
return m.GetASPeers(asn)
}
// GetIPInfo mock implementation
func (m *mockStore) GetIPInfo(ip string) (*database.IPInfo, error) {
return m.GetIPInfoContext(context.Background(), ip)
}
// GetIPInfoContext mock implementation with context support
func (m *mockStore) GetIPInfoContext(
_ context.Context, ip string,
) (*database.IPInfo, error) {
now := time.Now()
return &database.IPInfo{
IP: ip,
Netblock: "8.8.8.0/24",
MaskLength: 24,
IPVersion: 4,
NumPeers: 3,
ASN: 15169,
Handle: "GOOGLE",
Description: "Google LLC",
CountryCode: "US",
FirstSeen: now.Add(-24 * time.Hour),
LastSeen: now,
}, nil
}
// GetNextStaleASN mock implementation
func (m *mockStore) GetNextStaleASN(_ context.Context, _ time.Duration) (int, error) {
return 0, database.ErrNoStaleASN
}
// UpdateASNWHOIS mock implementation
func (m *mockStore) UpdateASNWHOIS(
_ context.Context, _ *database.ASNWHOISUpdate,
) error {
return nil
}
// GetWHOISStats mock implementation
func (m *mockStore) GetWHOISStats(
_ context.Context, _ time.Duration,
) (*database.WHOISStats, error) {
m.mu.Lock()
defer m.mu.Unlock()
return &database.WHOISStats{
TotalASNs: len(m.ASNs),
FreshASNs: 0,
StaleASNs: 0,
NeverFetched: len(m.ASNs),
}, nil
}
// UpsertLiveRouteBatch mock implementation
func (m *mockStore) UpsertLiveRouteBatch(routes []*database.LiveRoute) error {
m.mu.Lock()
defer m.mu.Unlock()
for _, route := range routes {
// Track prefix
if _, exists := m.Prefixes[route.Prefix]; !exists {
m.Prefixes[route.Prefix] = &database.Prefix{
Prefix: route.Prefix,
IPVersion: route.IPVersion,
FirstSeen: route.LastUpdated,
}
m.PrefixCount++
if route.IPVersion == 4 {
m.IPv4Prefixes++
} else {
m.IPv6Prefixes++
}
}
m.RouteCount++
}
return nil
}
// DeleteLiveRouteBatch mock implementation
func (m *mockStore) DeleteLiveRouteBatch(_ []database.LiveRouteDeletion) error {
// Simple mock - just return nil
return nil
}
// DeleteLiveRoutesOfPeer mock implementation
func (m *mockStore) DeleteLiveRoutesOfPeer(_ string) (int, error) {
// Simple mock - deletes nothing
return 0, nil
}
// GetOrCreateASNBatch mock implementation
func (m *mockStore) GetOrCreateASNBatch(asns map[int]database.SeenTimes) error {
m.mu.Lock()
defer m.mu.Unlock()
for number, seen := range asns {
if _, exists := m.ASNs[number]; !exists {
m.ASNs[number] = &database.ASN{
ASN: number,
FirstSeen: seen.First,
LastSeen: seen.Last,
}
m.ASNCount++
}
}
return nil
}
// UpdatePeerBatch mock implementation
func (m *mockStore) UpdatePeerBatch(_ map[string]database.PeerUpdate) error {
// Simple mock - just return nil
return nil
}
// Vacuum mock implementation
func (m *mockStore) Vacuum(_ context.Context) error {
return nil
}
// Analyze mock implementation
func (m *mockStore) Analyze(_ context.Context) error {
return nil
}
// Checkpoint mock implementation
func (m *mockStore) Checkpoint(_ context.Context) error {
return nil
}
// Ping mock implementation
func (m *mockStore) Ping(_ context.Context) error {
return nil
}
func TestRouteWatchLiveFeed(t *testing.T) {
t.Parallel()
if testing.Short() {
t.Skip("skipping live RIPE RIS network feed test in short mode; " +
"run without -short to include it")
}
// Create mock database
mockDB := newMockStore()
defer func() {
err := mockDB.Close()
if err != nil {
t.Errorf("failed to close mock database: %v", err)
}
}()
logger := logger.New()
// Create metrics tracker
metricsTracker := metrics.New()
// Create streamer
s := streamer.New(logger, metricsTracker)
// Create test config with empty state dir (no snapshot loading)
cfg := &config.Config{
StateDir: "",
MaxRuntime: 5 * time.Second,
EnableBatchedDatabaseWrites: true,
}
// Create server
srv := server.New(mockDB, s, logger, cfg)
// Create RouteWatch with 5 second limit
deps := Dependencies{
DB: mockDB,
Streamer: s,
Server: srv,
Logger: logger,
Config: cfg,
}
rw := New(deps)
// Run with context
ctx := context.Background()
go func() {
_ = rw.Run(ctx)
}()
// Wait for the configured duration
time.Sleep(5 * time.Second)
// Force peering processing for test
if rw.peeringHandler != nil {
rw.peeringHandler.ProcessPeeringsNow()
}
// Get statistics
stats, err := mockDB.GetStats()
if err != nil {
t.Fatalf("Failed to get stats: %v", err)
}
if stats.ASNs == 0 {
t.Error("Expected to receive some ASNs from live feed")
}
t.Logf("Received %d unique ASNs in 5 seconds", stats.ASNs)
if stats.Prefixes == 0 {
t.Error("Expected to receive some prefixes from live feed")
}
t.Logf("Received %d unique prefixes (%d IPv4, %d IPv6) in 5 seconds",
stats.Prefixes, stats.IPv4Prefixes, stats.IPv6Prefixes)
if stats.Peerings == 0 {
t.Error("Expected to receive some peerings from live feed")
}
t.Logf("Recorded %d AS peering relationships in 5 seconds", stats.Peerings)
}
// gatedPrefixHandler is a PrefixHandler whose worker in the streamer waits on
// the first message until release is closed, so the messages after it stay on
// the streamer's queue.
type gatedPrefixHandler struct {
*PrefixHandler
release chan struct{}
}
func (h gatedPrefixHandler) HandleMessage(msg *ristypes.RISMessage) {
<-h.release
h.PrefixHandler.HandleMessage(msg)
}
// announcementFeed returns the lines RIS Live would send for one UPDATE per
// prefix, each announcing that prefix from testPeerIP with origin testASNA,
// followed by a KEEPALIVE. The streamer hands a KEEPALIVE to no handler, so
// once a raw handler has seen it, every update before it is queued.
func announcementFeed(prefixes []string) string {
var feed strings.Builder
for _, prefix := range prefixes {
fmt.Fprintf(&feed, `{"type":"ris_message","data":{"type":"UPDATE",`+
`"peer":"%s","peer_asn":"%d","timestamp":1700000000,"path":[%d],`+
`"announcements":[{"next_hop":"%s","prefixes":["%s"]}]}}`+"\n",
testPeerIP, testASNA, testASNA, testPeerIP, prefix)
}
fmt.Fprintf(&feed, `{"type":"ris_message","data":{"type":"KEEPALIVE",`+
`"peer":"%s","peer_asn":"%d","timestamp":1700000000}}`+"\n",
testPeerIP, testASNA)
return feed.String()
}
// TestShutdownWritesTheUpdatesQueuedInTheStreamer has a local server send
// announcements of three prefixes and keep the response open. The prefix
// handler's worker waits on the first one until the server sees the streamer
// close the connection, which only Shutdown does, so all three are still in
// the streamer when Shutdown starts. Each prefix must be in the database when
// Shutdown returns.
func TestShutdownWritesTheUpdatesQueuedInTheStreamer(t *testing.T) {
t.Parallel()
const maxWait = 10 * time.Second
prefixes := []string{"192.0.2.0/24", "198.51.100.0/24", "203.0.113.0/24"}
feed := announcementFeed(prefixes)
release := make(chan struct{})
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, r *http.Request) {
_, _ = io.WriteString(w, feed)
_ = http.NewResponseController(w).Flush()
<-r.Context().Done()
close(release)
}))
defer srv.Close()
store := newMockStore()
log := logger.New()
cfg := &config.Config{}
s := streamer.New(log, metrics.New())
s.SetURL(srv.URL)
rw := New(Dependencies{
DB: store,
Streamer: s,
Server: server.New(store, s, log, cfg),
Logger: log,
Config: cfg,
})
rw.prefixHandler = NewPrefixHandler(store, log)
s.RegisterHandler(gatedPrefixHandler{
PrefixHandler: rw.prefixHandler, release: release,
})
linesSeen := make(chan struct{}, len(prefixes)+1)
s.RegisterRawHandler(func(string) { linesSeen <- struct{}{} })
err := s.Start(t.Context())
if err != nil {
t.Fatalf("Start returned error: %v", err)
}
for range cap(linesSeen) {
select {
case <-linesSeen:
case <-time.After(maxWait):
t.Fatalf("the streamer did not read %d lines within %s",
cap(linesSeen), maxWait)
}
}
shutDown := make(chan struct{})
go func() {
rw.Shutdown(t.Context())
close(shutDown)
}()
select {
case <-shutDown:
case <-time.After(maxWait):
t.Fatalf("Shutdown did not return within %s", maxWait)
}
store.mu.Lock()
defer store.mu.Unlock()
for _, prefix := range prefixes {
if _, ok := store.Prefixes[prefix]; !ok {
t.Errorf("%s was queued in the streamer when Shutdown started "+
"but is not in the database", prefix)
}
}
}
// freePort returns a TCP port that was free a moment ago, for a server the
// test starts.
func freePort(t *testing.T) int {
t.Helper()
var listenConfig net.ListenConfig
listener, err := listenConfig.Listen(t.Context(), "tcp", ":0")
if err != nil {
t.Fatalf("failed to find a free port: %v", err)
}
defer func() { _ = listener.Close() }()
addr, ok := listener.Addr().(*net.TCPAddr)
if !ok {
t.Fatalf("listener address %v is not a TCP address", listener.Addr())
}
return addr.Port
}
// TestShutdownReturnsByItsDeadlineWhileAWriteIsBlocked starts the HTTP server
// and has the AS handler start a write that the store holds until the test
// ends, past Shutdown's deadline. Shutdown must return within its deadline
// plus a small margin, with the HTTP server stopped, and must neither
// checkpoint nor close the database: the real one would wait for the write.
// Before the fix Shutdown waited for the write.
func TestShutdownReturnsByItsDeadlineWhileAWriteIsBlocked(t *testing.T) {
t.Parallel()
const (
deadline = 100 * time.Millisecond
// margin is how long Shutdown may take after its deadline, to stop
// the HTTP server.
margin = 2 * time.Second
)
blocking := newBlockingStore()
defer blocking.releaseWrites()
store := &closeOrderStore{Store: blocking}
log := logger.New()
port := freePort(t)
cfg := &config.Config{Port: port}
s := streamer.New(log, metrics.New())
rw := New(Dependencies{
DB: store,
Streamer: s,
Server: server.New(store, s, log, cfg),
Logger: log,
Config: cfg,
})
err := rw.server.Start(t.Context())
if err != nil {
t.Fatalf("failed to start the HTTP server: %v", err)
}
rw.dbHandler = NewASHandler(store, log)
fill := ristypes.ASPath{Sequence: slices.Repeat([]int{testASNA}, asnBatchSize)}
startBlockedWrite(t, blocking, rw.dbHandler,
[]*ristypes.RISMessage{{Path: fill}})
ctx, cancel := context.WithTimeout(t.Context(), deadline)
defer cancel()
shutDown := make(chan struct{})
go func() {
rw.Shutdown(ctx)
close(shutDown)
}()
select {
case <-shutDown:
case <-time.After(deadline + margin):
t.Fatalf("Shutdown did not return within %s of its deadline while "+
"a write was blocked", margin)
}
var dialer net.Dialer
addr := net.JoinHostPort("127.0.0.1", strconv.Itoa(port))
conn, err := dialer.DialContext(t.Context(), "tcp", addr)
if err == nil {
_ = conn.Close()
t.Errorf("the HTTP server still accepts connections on %s after "+
"Shutdown returned", addr)
}
store.callsMu.Lock()
defer store.callsMu.Unlock()
if slices.Contains(store.calls, callCheckpoint) ||
slices.Contains(store.calls, callClose) {
t.Errorf("Shutdown made the calls %q past its deadline, want no %s "+
"and no %s", store.calls, callCheckpoint, callClose)
}
}
// closeOrderStore is a store that records, in order, the AS writes,
// checkpoints and closes it is given. It passes the AS writes on to Store.
type closeOrderStore struct {
database.Store
callsMu sync.Mutex
calls []string
}
// The calls closeOrderStore records.
const (
callWrite = "AS write"
callCheckpoint = "checkpoint"
callClose = "close"
)
func (s *closeOrderStore) GetOrCreateASNBatch(
asns map[int]database.SeenTimes,
) error {
s.record(callWrite)
return s.Store.GetOrCreateASNBatch(asns)
}
func (s *closeOrderStore) Checkpoint(_ context.Context) error {
s.record(callCheckpoint)
return nil
}
func (s *closeOrderStore) Close() error {
s.record(callClose)
return nil
}
func (s *closeOrderStore) record(call string) {
s.callsMu.Lock()
defer s.callsMu.Unlock()
s.calls = append(s.calls, call)
}
// TestShutdownCheckpointsAndClosesTheDatabaseLast hands the AS handler one
// message and shuts down with time to spare. The handler's last write must
// come first, then a checkpoint, then the close. Before the fix Shutdown
// neither checkpointed nor closed the database.
func TestShutdownCheckpointsAndClosesTheDatabaseLast(t *testing.T) {
t.Parallel()
const deadline = 10 * time.Second
store := &closeOrderStore{Store: newMockStore()}
log := logger.New()
cfg := &config.Config{}
s := streamer.New(log, metrics.New())
rw := New(Dependencies{
DB: store,
Streamer: s,
Server: server.New(store, s, log, cfg),
Logger: log,
Config: cfg,
})
rw.dbHandler = NewASHandler(store, log)
rw.dbHandler.HandleMessage(&ristypes.RISMessage{
Path: ristypes.ASPath{Sequence: []int{testASNA}},
})
ctx, cancel := context.WithTimeout(t.Context(), deadline)
defer cancel()
rw.Shutdown(ctx)
store.callsMu.Lock()
defer store.callsMu.Unlock()
want := []string{callWrite, callCheckpoint, callClose}
if !slices.Equal(store.calls, want) {
t.Errorf("Shutdown made the calls %q, want %q", store.calls, want)
}
}
internal/routewatch/app_integration_test.go
-500
View File
@@ -1,500 +0,0 @@
package routewatch
import (
"context"
"fmt"
"strings"
"sync"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/config"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/metrics"
"git.eeqj.de/sneak/routewatch/internal/server"
"git.eeqj.de/sneak/routewatch/internal/streamer"
"github.com/google/uuid"
)
// mockStore is a mock implementation of database.Store for testing
type mockStore struct {
mu sync.Mutex
// Counters for tracking calls
ASNCount int
PrefixCount int
PeeringCount int
RouteCount int
WithdrawalCount int
// Track unique items
ASNs map[int]*database.ASN
Prefixes map[string]*database.Prefix
Peerings map[string]bool // key is "from_to"
Routes map[string]bool // key is "prefix_origin_peer"
// Track IP versions
IPv4Prefixes int
IPv6Prefixes int
}
// newMockStore creates a new mock store
func newMockStore() *mockStore {
return &mockStore{
ASNs: make(map[int]*database.ASN),
Prefixes: make(map[string]*database.Prefix),
Peerings: make(map[string]bool),
Routes: make(map[string]bool),
}
}
// GetOrCreateASN mock implementation
func (m *mockStore) GetOrCreateASN(number int, timestamp time.Time) (*database.ASN, error) {
m.mu.Lock()
defer m.mu.Unlock()
if asn, exists := m.ASNs[number]; exists {
asn.LastSeen = timestamp
return asn, nil
}
asn := &database.ASN{
ASN: number,
FirstSeen: timestamp,
LastSeen: timestamp,
}
m.ASNs[number] = asn
m.ASNCount++
return asn, nil
}
// UpdatePrefixesBatch mock implementation
func (m *mockStore) UpdatePrefixesBatch(prefixes map[string]time.Time) error {
m.mu.Lock()
defer m.mu.Unlock()
for prefix, timestamp := range prefixes {
if p, exists := m.Prefixes[prefix]; exists {
p.LastSeen = timestamp
} else {
const (
ipVersionV4 = 4
ipVersionV6 = 6
)
ipVersion := ipVersionV4
if strings.Contains(prefix, ":") {
ipVersion = ipVersionV6
}
m.Prefixes[prefix] = &database.Prefix{
ID: uuid.New(),
Prefix: prefix,
IPVersion: ipVersion,
FirstSeen: timestamp,
LastSeen: timestamp,
}
}
}
return nil
}
// GetOrCreatePrefix mock implementation
func (m *mockStore) GetOrCreatePrefix(prefix string, timestamp time.Time) (*database.Prefix, error) {
m.mu.Lock()
defer m.mu.Unlock()
if p, exists := m.Prefixes[prefix]; exists {
p.LastSeen = timestamp
return p, nil
}
const (
ipVersionV4 = 4
ipVersionV6 = 6
)
ipVersion := ipVersionV4
if strings.Contains(prefix, ":") {
ipVersion = ipVersionV6
}
p := &database.Prefix{
ID: uuid.New(),
Prefix: prefix,
IPVersion: ipVersion,
FirstSeen: timestamp,
LastSeen: timestamp,
}
m.Prefixes[prefix] = p
m.PrefixCount++
if ipVersion == ipVersionV4 {
m.IPv4Prefixes++
} else {
m.IPv6Prefixes++
}
return p, nil
}
// RecordAnnouncement mock implementation
func (m *mockStore) RecordAnnouncement(_ *database.Announcement) error {
// Not tracking announcements in detail for now
return nil
}
// RecordPeering mock implementation
func (m *mockStore) RecordPeering(asA, asB int, _ time.Time) error {
m.mu.Lock()
defer m.mu.Unlock()
// Normalize
if asA > asB {
asA, asB = asB, asA
}
key := fmt.Sprintf("%d_%d", asA, asB)
if !m.Peerings[key] {
m.Peerings[key] = true
m.PeeringCount++
}
return nil
}
// UpdatePeer mock implementation
func (m *mockStore) UpdatePeer(peerIP string, peerASN int, messageType string, timestamp time.Time) error {
// Simple mock - just return nil
return nil
}
// Close mock implementation
func (m *mockStore) Close() error {
return nil
}
// GetStats returns statistics about the mock store
func (m *mockStore) GetStats() (database.Stats, error) {
m.mu.Lock()
defer m.mu.Unlock()
return database.Stats{
ASNs: len(m.ASNs),
Prefixes: len(m.Prefixes),
IPv4Prefixes: m.IPv4Prefixes,
IPv6Prefixes: m.IPv6Prefixes,
Peerings: m.PeeringCount,
Peers: 10, // Mock peer count
}, nil
}
// GetStatsContext returns statistics about the mock store with context support
func (m *mockStore) GetStatsContext(ctx context.Context) (database.Stats, error) {
return m.GetStats()
}
// UpsertLiveRoute mock implementation
func (m *mockStore) UpsertLiveRoute(route *database.LiveRoute) error {
// Simple mock - just return nil
return nil
}
// DeleteLiveRoute mock implementation
func (m *mockStore) DeleteLiveRoute(prefix string, originASN int, peerIP string) error {
// Simple mock - just return nil
return nil
}
// GetPrefixDistribution mock implementation
func (m *mockStore) GetPrefixDistribution() (ipv4 []database.PrefixDistribution, ipv6 []database.PrefixDistribution, err error) {
// Return empty distributions for now
return nil, nil, nil
}
// GetPrefixDistributionContext mock implementation with context support
func (m *mockStore) GetPrefixDistributionContext(ctx context.Context) (ipv4 []database.PrefixDistribution, ipv6 []database.PrefixDistribution, err error) {
return m.GetPrefixDistribution()
}
// GetLiveRouteCounts mock implementation
func (m *mockStore) GetLiveRouteCounts() (ipv4Count, ipv6Count int, err error) {
// Return mock counts
return m.RouteCount / 2, m.RouteCount / 2, nil
}
// GetLiveRouteCountsContext mock implementation with context support
func (m *mockStore) GetLiveRouteCountsContext(ctx context.Context) (ipv4Count, ipv6Count int, err error) {
return m.GetLiveRouteCounts()
}
// GetASInfoForIP mock implementation
func (m *mockStore) GetASInfoForIP(ip string) (*database.ASInfo, error) {
// Simple mock - return a test AS
now := time.Now()
return &database.ASInfo{
ASN: 15169,
Handle: "GOOGLE",
Description: "Google LLC",
Prefix: "8.8.8.0/24",
LastUpdated: now.Add(-5 * time.Minute),
Age: "5m0s",
}, nil
}
// GetASInfoForIPContext mock implementation with context support
func (m *mockStore) GetASInfoForIPContext(ctx context.Context, ip string) (*database.ASInfo, error) {
return m.GetASInfoForIP(ip)
}
// GetASDetails mock implementation
func (m *mockStore) GetASDetails(asn int) (*database.ASN, []database.LiveRoute, error) {
m.mu.Lock()
defer m.mu.Unlock()
// Check if ASN exists
if asnInfo, exists := m.ASNs[asn]; exists {
// Return empty prefixes for now
return asnInfo, []database.LiveRoute{}, nil
}
return nil, nil, database.ErrNoRoute
}
// GetASDetailsContext mock implementation with context support
func (m *mockStore) GetASDetailsContext(ctx context.Context, asn int) (*database.ASN, []database.LiveRoute, error) {
return m.GetASDetails(asn)
}
// GetPrefixDetails mock implementation
func (m *mockStore) GetPrefixDetails(prefix string) ([]database.LiveRoute, error) {
// Return empty routes for now
return []database.LiveRoute{}, nil
}
// GetPrefixDetailsContext mock implementation with context support
func (m *mockStore) GetPrefixDetailsContext(ctx context.Context, prefix string) ([]database.LiveRoute, error) {
return m.GetPrefixDetails(prefix)
}
func (m *mockStore) GetRandomPrefixesByLength(maskLength, ipVersion, limit int) ([]database.LiveRoute, error) {
// Return empty routes for now
return []database.LiveRoute{}, nil
}
// GetRandomPrefixesByLengthContext mock implementation with context support
func (m *mockStore) GetRandomPrefixesByLengthContext(ctx context.Context, maskLength, ipVersion, limit int) ([]database.LiveRoute, error) {
return m.GetRandomPrefixesByLength(maskLength, ipVersion, limit)
}
// GetASPeers mock implementation
func (m *mockStore) GetASPeers(asn int) ([]database.ASPeer, error) {
// Return empty peers for now
return []database.ASPeer{}, nil
}
// GetASPeersContext mock implementation with context support
func (m *mockStore) GetASPeersContext(ctx context.Context, asn int) ([]database.ASPeer, error) {
return m.GetASPeers(asn)
}
// GetIPInfo mock implementation
func (m *mockStore) GetIPInfo(ip string) (*database.IPInfo, error) {
return m.GetIPInfoContext(context.Background(), ip)
}
// GetIPInfoContext mock implementation with context support
func (m *mockStore) GetIPInfoContext(ctx context.Context, ip string) (*database.IPInfo, error) {
now := time.Now()
return &database.IPInfo{
IP: ip,
Netblock: "8.8.8.0/24",
MaskLength: 24,
IPVersion: 4,
NumPeers: 3,
ASN: 15169,
Handle: "GOOGLE",
Description: "Google LLC",
CountryCode: "US",
FirstSeen: now.Add(-24 * time.Hour),
LastSeen: now,
}, nil
}
// GetNextStaleASN mock implementation
func (m *mockStore) GetNextStaleASN(ctx context.Context, staleThreshold time.Duration) (int, error) {
return 0, database.ErrNoStaleASN
}
// UpdateASNWHOIS mock implementation
func (m *mockStore) UpdateASNWHOIS(ctx context.Context, update *database.ASNWHOISUpdate) error {
return nil
}
// GetWHOISStats mock implementation
func (m *mockStore) GetWHOISStats(ctx context.Context, staleThreshold time.Duration) (*database.WHOISStats, error) {
m.mu.Lock()
defer m.mu.Unlock()
return &database.WHOISStats{
TotalASNs: len(m.ASNs),
FreshASNs: 0,
StaleASNs: 0,
NeverFetched: len(m.ASNs),
}, nil
}
// UpsertLiveRouteBatch mock implementation
func (m *mockStore) UpsertLiveRouteBatch(routes []*database.LiveRoute) error {
m.mu.Lock()
defer m.mu.Unlock()
for _, route := range routes {
// Track prefix
if _, exists := m.Prefixes[route.Prefix]; !exists {
m.Prefixes[route.Prefix] = &database.Prefix{
ID: uuid.New(),
Prefix: route.Prefix,
IPVersion: route.IPVersion,
FirstSeen: route.LastUpdated,
LastSeen: route.LastUpdated,
}
m.PrefixCount++
if route.IPVersion == 4 {
m.IPv4Prefixes++
} else {
m.IPv6Prefixes++
}
}
m.RouteCount++
}
return nil
}
// DeleteLiveRouteBatch mock implementation
func (m *mockStore) DeleteLiveRouteBatch(deletions []database.LiveRouteDeletion) error {
// Simple mock - just return nil
return nil
}
// GetOrCreateASNBatch mock implementation
func (m *mockStore) GetOrCreateASNBatch(asns map[int]time.Time) error {
m.mu.Lock()
defer m.mu.Unlock()
for number, timestamp := range asns {
if _, exists := m.ASNs[number]; !exists {
m.ASNs[number] = &database.ASN{
ASN: number,
FirstSeen: timestamp,
LastSeen: timestamp,
}
m.ASNCount++
}
}
return nil
}
// UpdatePeerBatch mock implementation
func (m *mockStore) UpdatePeerBatch(peers map[string]database.PeerUpdate) error {
// Simple mock - just return nil
return nil
}
// Vacuum mock implementation
func (m *mockStore) Vacuum(ctx context.Context) error {
return nil
}
// Analyze mock implementation
func (m *mockStore) Analyze(ctx context.Context) error {
return nil
}
// Checkpoint mock implementation
func (m *mockStore) Checkpoint(ctx context.Context) error {
return nil
}
// Ping mock implementation
func (m *mockStore) Ping(ctx context.Context) error {
return nil
}
func TestRouteWatchLiveFeed(t *testing.T) {
if testing.Short() {
t.Skip("skipping live RIPE RIS network feed test in short mode; run without -short to include it")
}
// Create mock database
mockDB := newMockStore()
defer mockDB.Close()
logger := logger.New()
// Create metrics tracker
metricsTracker := metrics.New()
// Create streamer
s := streamer.New(logger, metricsTracker)
// Create test config with empty state dir (no snapshot loading)
cfg := &config.Config{
StateDir: "",
MaxRuntime: 5 * time.Second,
EnableBatchedDatabaseWrites: true,
}
// Create server
srv := server.New(mockDB, s, logger)
// Create RouteWatch with 5 second limit
deps := Dependencies{
DB: mockDB,
Streamer: s,
Server: srv,
Logger: logger,
Config: cfg,
}
rw := New(deps)
// Run with context
ctx := context.Background()
go func() {
_ = rw.Run(ctx)
}()
// Wait for the configured duration
time.Sleep(5 * time.Second)
// Force peering processing for test
if rw.peeringHandler != nil {
rw.peeringHandler.ProcessPeeringsNow()
}
// Get statistics
stats, err := mockDB.GetStats()
if err != nil {
t.Fatalf("Failed to get stats: %v", err)
}
if stats.ASNs == 0 {
t.Error("Expected to receive some ASNs from live feed")
}
t.Logf("Received %d unique ASNs in 5 seconds", stats.ASNs)
if stats.Prefixes == 0 {
t.Error("Expected to receive some prefixes from live feed")
}
t.Logf("Received %d unique prefixes (%d IPv4, %d IPv6) in 5 seconds", stats.Prefixes, stats.IPv4Prefixes, stats.IPv6Prefixes)
if stats.Peerings == 0 {
t.Error("Expected to receive some peerings from live feed")
}
t.Logf("Recorded %d AS peering relationships in 5 seconds", stats.Peerings)
}
internal/routewatch/app_test.go
+2 -2
View File
@@ -1,3 +1,3 @@
package routewatch
package routewatch_test
// Tests for routewatch package are in app_integration_test.go
// Tests for routewatch package are in app_integration_internal_test.go
internal/routewatch/ashandler.go
+127 -57
View File
@@ -1,6 +1,8 @@
package routewatch
import (
"context"
"slices"
"sync"
"time"
@@ -19,6 +21,12 @@ const (
// asnBatchSize is the number of ASN operations to batch together
asnBatchSize = 30000
// asnBatchMaxSize is the most ASN operations the batch holds while the
// flush goroutine writes the previous one: ten batches, about 10 MiB
// (README "Memory"). At this size HandleMessage waits until the flush
// goroutine takes the batch.
asnBatchMaxSize = 300000
// asnBatchTimeout is the maximum time to wait before flushing a batch
// DO NOT reduce this timeout - larger batches are more efficient
asnBatchTimeout = 2 * time.Second
@@ -32,12 +40,23 @@ type ASHandler struct {
db database.Store
logger *logger.Logger
// Batching
mu sync.Mutex
batch []asnOp
lastFlush time.Time
stopCh chan struct{}
wg sync.WaitGroup
// Batching. mu guards batch and writing only: the flush goroutine takes
// the batch under mu and writes it after releasing mu, so HandleMessage
// waits for a database write only while batch is at asnBatchMaxSize.
mu sync.Mutex
batch []asnOp
// writing is how many operations the flush goroutine took from batch
// and has not finished writing.
writing int
// batchTaken is signalled when the flush goroutine takes batch; its L is
// &mu.
batchTaken sync.Cond
batchFullCh chan struct{} // wakes the flush goroutine when batch is full
lastFlush time.Time // used only by the flush goroutine
// stopCh hands Stop's context to the flush goroutine. It holds one, so
// Stop does not wait for a write in progress to hand it over.
stopCh chan context.Context
done chan struct{} // closed when the flush goroutine returns
}
type asnOp struct {
@@ -52,15 +71,17 @@ type asnOp struct {
// background goroutine is terminated.
func NewASHandler(db database.Store, logger *logger.Logger) *ASHandler {
h := &ASHandler{
db: db,
logger: logger,
batch: make([]asnOp, 0, asnBatchSize),
lastFlush: time.Now(),
stopCh: make(chan struct{}),
db: db,
logger: logger,
batch: make([]asnOp, 0, asnBatchSize),
batchFullCh: make(chan struct{}, 1),
lastFlush: time.Now(),
stopCh: make(chan context.Context, 1),
done: make(chan struct{}),
}
h.batchTaken.L = &h.mu
// Start the flush timer goroutine
h.wg.Add(1)
go h.flushLoop()
return h
@@ -71,7 +92,7 @@ func NewASHandler(db database.Store, logger *logger.Logger) *ASHandler {
// the AS path information needed to track autonomous systems.
func (h *ASHandler) WantsMessage(messageType string) bool {
// We only care about UPDATE messages for the database
return messageType == "UPDATE"
return messageType == ristypes.MessageTypeUpdate
}
// QueueCapacity returns the recommended message queue size for this handler.
@@ -84,22 +105,25 @@ func (h *ASHandler) QueueCapacity() int {
// HandleMessage processes a RIS Live BGP message by extracting all ASNs from
// the AS path and queuing them for batch insertion into the database. The
// origin ASN (last element in the path) and all transit ASNs are recorded
// with their associated timestamps. The batch is automatically flushed when
// it reaches the configured size threshold.
// origin ASN and all transit ASNs are recorded with their associated
// timestamps. When the batch reaches the configured size threshold, the flush
// goroutine is woken to write it.
func (h *ASHandler) HandleMessage(msg *ristypes.RISMessage) {
// Use the pre-parsed timestamp
timestamp := msg.ParsedTimestamp
// Get origin ASN from path (last element)
var originASN int
if len(msg.Path) > 0 {
originASN = msg.Path[len(msg.Path)-1]
}
originASN := msg.Path.Origin()
h.mu.Lock()
defer h.mu.Unlock()
// While the batch is at its maximum size, wait until the flush goroutine
// takes it. Meanwhile the handler takes nothing off its queue, which
// fills, and the streamer drops messages for it.
for len(h.batch) >= asnBatchMaxSize {
h.batchTaken.Wait()
}
// Queue origin ASN operation
if originASN > 0 {
h.batch = append(h.batch, asnOp{
@@ -108,8 +132,8 @@ func (h *ASHandler) HandleMessage(msg *ristypes.RISMessage) {
})
}
// Also track all ASNs in the path
for _, asn := range msg.Path {
// Also track all ASNs in the path, the members of an AS set included
for _, asn := range slices.Concat(msg.Path.Sequence, msg.Path.Set) {
if asn > 0 {
h.batch = append(h.batch, asnOp{
number: asn,
@@ -118,67 +142,113 @@ func (h *ASHandler) HandleMessage(msg *ristypes.RISMessage) {
}
}
// Check if we need to flush
// A full batch wakes the flush goroutine. If it is writing, the wake-up
// waits in batchFullCh and the batch keeps growing until the write ends
// or the batch reaches asnBatchMaxSize.
if len(h.batch) >= asnBatchSize {
h.flushBatchLocked()
select {
case h.batchFullCh <- struct{}{}:
default: // already woken
}
}
}
// flushLoop runs in a goroutine and periodically flushes batches
// Stop stops the flush goroutine, which writes the operations the handler holds
// before it returns, and waits for it until ctx ends. Once ctx has ended the
// flush goroutine skips that last write, and a write still running is lost
// when the process exits, so Stop logs how many operations are not written.
func (h *ASHandler) Stop(ctx context.Context) {
h.stopCh <- ctx
select {
case <-h.done:
case <-ctx.Done():
}
h.mu.Lock()
unwritten := h.writing + len(h.batch)
h.mu.Unlock()
if unwritten > 0 {
h.logger.Error("Stop deadline passed before the AS handler wrote "+
"its updates", "updates", unwritten)
}
}
// flushLoop runs in a goroutine, the only one that writes the batch: it
// flushes when the batch is full, periodically, and once more on Stop.
func (h *ASHandler) flushLoop() {
defer h.wg.Done()
defer close(h.done)
ticker := time.NewTicker(asnBatchTimeout)
defer ticker.Stop()
for {
select {
case <-ticker.C:
h.mu.Lock()
if time.Since(h.lastFlush) >= asnBatchTimeout {
h.flushBatchLocked()
h.flushBatch()
}
case <-h.batchFullCh:
h.flushBatch()
case ctx := <-h.stopCh:
// Final flush, unless the stop deadline has passed
if ctx.Err() == nil {
h.flushBatch()
}
h.mu.Unlock()
case <-h.stopCh:
// Final flush
h.mu.Lock()
h.flushBatchLocked()
h.mu.Unlock()
return
}
}
}
// flushBatchLocked flushes the ASN batch to the database (must be called with mutex held)
func (h *ASHandler) flushBatchLocked() {
// takeBatch returns the pending operations and leaves an empty batch in their
// place for HandleMessage to fill. They count in writing until flushBatch has
// written them.
func (h *ASHandler) takeBatch() []asnOp {
h.mu.Lock()
defer h.mu.Unlock()
if len(h.batch) == 0 {
return nil
}
batch := h.batch
h.writing = len(batch)
h.batch = make([]asnOp, 0, asnBatchSize)
h.batchTaken.Broadcast()
return batch
}
// flushBatch takes the pending batch and writes it to the database, without
// holding the mutex during the write.
func (h *ASHandler) flushBatch() {
batch := h.takeBatch()
if len(batch) == 0 {
return
}
// Process ASNs first (deduped)
asnMap := make(map[int]time.Time)
for _, op := range h.batch {
if existing, ok := asnMap[op.number]; !ok || op.timestamp.After(existing) {
asnMap[op.number] = op.timestamp
}
// The earliest and the latest time the batch saw each AS. The batch holds
// several updates for an AS, from many peers, so a new AS is stored first
// seen at the earliest of them and last seen at the latest.
asns := make(map[int]database.SeenTimes)
for _, op := range batch {
seen := asns[op.number]
seen.Add(op.timestamp)
asns[op.number] = seen
}
// Process all ASNs in a single batch transaction
if err := h.db.GetOrCreateASNBatch(asnMap); err != nil {
h.logger.Error("Failed to process ASN batch", "error", err, "count", len(asnMap))
err := h.db.GetOrCreateASNBatch(asns)
if err != nil {
h.logger.Error("Failed to process ASN batch",
"error", err, "count", len(asns))
}
// Clear batch
h.batch = h.batch[:0]
h.mu.Lock()
h.writing = 0
h.mu.Unlock()
h.lastFlush = time.Now()
}
// Stop gracefully shuts down the ASHandler by signaling the background flush
// goroutine to terminate and waiting for it to complete. Any pending ASN
// operations in the current batch are flushed to the database before Stop
// returns. This method should be called during application shutdown to ensure
// no data is lost.
func (h *ASHandler) Stop() {
close(h.stopCh)
h.wg.Wait()
}
internal/routewatch/ashandler_internal_test.go
+44
View File
@@ -0,0 +1,44 @@
package routewatch
import (
"testing"
"git.eeqj.de/sneak/routewatch/internal/logger"
)
// TestASHandlerStoresTheEarliestAndLatestTimesOfANewAS handles two updates
// whose path carries an AS the database does not hold yet, the second 5
// seconds after the first, in one batch into a real database. The AS is stored
// first seen at the time of the first update and last seen at the time of the
// second.
func TestASHandlerStoresTheEarliestAndLatestTimesOfANewAS(t *testing.T) {
t.Parallel()
db := newTestDatabase(t)
// Built without NewASHandler's flush goroutine, so the test flushes itself.
h := &ASHandler{db: db, logger: logger.New()}
first := peerAnnouncement(testPeerIP, testV4Prefix)
second := peerAnnouncement(testPeerIP, testV4Prefix)
second.ParsedTimestamp = first.ParsedTimestamp.Add(secondUpdateAfter)
h.HandleMessage(first)
h.HandleMessage(second)
h.flushBatch()
asn, _, err := db.GetASDetails(testASNA)
if err != nil {
t.Fatalf("GetASDetails(%d): %v", testASNA, err)
}
if !asn.FirstSeen.Equal(first.ParsedTimestamp) {
t.Errorf("first_seen of AS%d = %s, want %s", testASNA,
asn.FirstSeen, first.ParsedTimestamp)
}
if !asn.LastSeen.Equal(second.ParsedTimestamp) {
t.Errorf("last_seen of AS%d = %s, want %s", testASNA,
asn.LastSeen, second.ParsedTimestamp)
}
}
internal/routewatch/asnfetcher.go
+50 -32
View File
@@ -1,8 +1,8 @@
// Package routewatch contains the ASN WHOIS fetcher for background updates.
package routewatch
import (
"context"
"errors"
"log/slog"
"sync"
"time"
@@ -43,7 +43,7 @@ type ASNFetcher struct {
logger *slog.Logger
immediateQueue chan int
stopCh chan struct{}
wg sync.WaitGroup
done chan struct{} // closed when the fetcher goroutine returns
// fetchMu ensures only one fetch runs at a time
fetchMu sync.Mutex
@@ -67,16 +67,18 @@ func NewASNFetcher(db database.Store, logger *slog.Logger) *ASNFetcher {
logger: logger.With("component", "asn_fetcher"),
immediateQueue: make(chan int, immediateQueueSize),
stopCh: make(chan struct{}),
done: make(chan struct{}),
currentInterval: baseInterval,
successTimes: make([]time.Time, 0),
errorTimes: make([]time.Time, 0),
}
}
// Start begins the background ASN fetcher goroutine.
func (f *ASNFetcher) Start() {
f.wg.Add(1)
go f.run()
// Start begins the background ASN fetcher goroutine. It runs until Stop:
// the end of ctx does not stop it.
func (f *ASNFetcher) Start(ctx context.Context) {
go f.run(context.WithoutCancel(ctx))
f.logger.Info("ASN fetcher started",
"base_interval", baseInterval,
"min_interval", minInterval,
@@ -84,11 +86,17 @@ func (f *ASNFetcher) Start() {
)
}
// Stop gracefully shuts down the fetcher.
func (f *ASNFetcher) Stop() {
// Stop stops the fetcher and waits, until ctx ends, for a lookup in progress to
// finish.
func (f *ASNFetcher) Stop(ctx context.Context) {
close(f.stopCh)
f.wg.Wait()
f.logger.Info("ASN fetcher stopped")
select {
case <-f.done:
f.logger.Info("ASN fetcher stopped")
case <-ctx.Done():
f.logger.Warn("Stop deadline passed before the ASN fetcher stopped")
}
}
// QueueImmediate queues an ASN for immediate WHOIS lookup.
@@ -151,10 +159,7 @@ func (f *ASNFetcher) recordSuccess() {
f.consecutiveFails = 0
// Decrease interval by half, but not below minimum
newInterval := f.currentInterval / backoffMultiplier
if newInterval < minInterval {
newInterval = minInterval
}
newInterval := max(f.currentInterval/backoffMultiplier, minInterval)
if newInterval != f.currentInterval {
f.logger.Debug("Decreased fetch interval",
@@ -177,10 +182,7 @@ func (f *ASNFetcher) recordFailure() {
f.consecutiveFails++
// Exponential backoff: multiply by 2, capped at max
newInterval := f.currentInterval * backoffMultiplier
if newInterval > maxInterval {
newInterval = maxInterval
}
newInterval := min(f.currentInterval*backoffMultiplier, maxInterval)
if newInterval != f.currentInterval {
f.logger.Debug("Increased fetch interval due to failure",
@@ -199,8 +201,8 @@ func (f *ASNFetcher) recordFailure() {
}
// run is the main background loop.
func (f *ASNFetcher) run() {
defer f.wg.Done()
func (f *ASNFetcher) run(ctx context.Context) {
defer close(f.done)
timer := time.NewTimer(f.getInterval())
defer timer.Stop()
@@ -212,13 +214,13 @@ func (f *ASNFetcher) run() {
case asn := <-f.immediateQueue:
// Process immediate request (respects lock)
f.tryFetch(asn)
f.tryFetch(ctx, asn)
// Reset timer after immediate fetch
timer.Reset(f.getInterval())
case <-timer.C:
// Background fetch of stale/missing ASN
f.fetchNextStale()
f.fetchNextStale(ctx)
// Reset timer with potentially updated interval
timer.Reset(f.getInterval())
}
@@ -227,7 +229,7 @@ func (f *ASNFetcher) run() {
// tryFetch attempts to fetch and update an ASN, respecting the fetch lock.
// Returns true if fetch was successful.
func (f *ASNFetcher) tryFetch(asn int) bool {
func (f *ASNFetcher) tryFetch(ctx context.Context, asn int) bool {
// Try to acquire lock, skip if another fetch is running
if !f.fetchMu.TryLock() {
f.logger.Debug("Skipping fetch, another fetch in progress", "asn", asn)
@@ -236,11 +238,11 @@ func (f *ASNFetcher) tryFetch(asn int) bool {
}
defer f.fetchMu.Unlock()
return f.fetchAndUpdate(asn)
return f.fetchAndUpdate(ctx, asn)
}
// fetchNextStale finds and fetches the next ASN needing WHOIS data.
func (f *ASNFetcher) fetchNextStale() {
func (f *ASNFetcher) fetchNextStale(ctx context.Context) {
// Try to acquire lock, skip if another fetch is running
if !f.fetchMu.TryLock() {
f.logger.Debug("Skipping stale fetch, another fetch in progress")
@@ -249,12 +251,12 @@ func (f *ASNFetcher) fetchNextStale() {
}
defer f.fetchMu.Unlock()
ctx, cancel := context.WithTimeout(context.Background(), time.Minute)
lookupCtx, cancel := context.WithTimeout(ctx, time.Minute)
defer cancel()
asn, err := f.db.GetNextStaleASN(ctx, whoisStaleThreshold)
asn, err := f.db.GetNextStaleASN(lookupCtx, whoisStaleThreshold)
if err != nil {
if err != database.ErrNoStaleASN {
if !errors.Is(err, database.ErrNoStaleASN) {
f.logger.Error("Failed to get stale ASN", "error", err)
f.recordFailure()
}
@@ -263,18 +265,34 @@ func (f *ASNFetcher) fetchNextStale() {
return
}
f.fetchAndUpdate(asn)
f.fetchAndUpdate(ctx, asn)
}
// fetchAndUpdate performs a WHOIS lookup and updates the database.
// Returns true if successful.
func (f *ASNFetcher) fetchAndUpdate(asn int) bool {
ctx, cancel := context.WithTimeout(context.Background(), time.Minute)
// Returns true if successful. A failed lookup leaves the AS's row as it was,
// so the AS stays stale. An AS IANA refers to no RIR is stored with no
// WHOIS data, so it is not looked up again until it is stale; that counts
// as neither a success nor a failure.
func (f *ASNFetcher) fetchAndUpdate(ctx context.Context, asn int) bool {
ctx, cancel := context.WithTimeout(ctx, time.Minute)
defer cancel()
f.logger.Info("Fetching WHOIS data", "asn", asn)
info, err := f.whoisClient.LookupASN(ctx, asn)
if errors.Is(err, whois.ErrNoReferral) {
f.logger.Info("IANA refers the AS to no RIR", "asn", asn)
err = f.db.UpdateASNWHOIS(ctx, &database.ASNWHOISUpdate{ASN: asn})
if err != nil {
f.logger.Error("Failed to update ASN WHOIS data",
"asn", asn, "error", err)
f.recordFailure()
}
return false
}
if err != nil {
f.logger.Error("WHOIS lookup failed", "asn", asn, "error", err)
f.recordFailure()
internal/routewatch/asnfetcher_internal_test.go
+205
View File
@@ -0,0 +1,205 @@
package routewatch
import (
"bufio"
"context"
"errors"
"io"
"net"
"sync"
"syscall"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
)
// The WHOIS servers a lookup asks: IANA first, then the RIR IANA refers it
// to.
const (
ianaWHOISServer = "whois.iana.org:43"
ripeWHOISServer = "whois.ripe.net:43"
)
// ianaRefersToRIPE is IANA's answer for an AS in a block it gave to RIPE.
const ianaRefersToRIPE = `refer: whois.ripe.net
as-block: 64496-64511
organisation: Assigned by RIPE NCC
changed: 2002-09-01
source: IANA
`
// ripeAccessDenied is RIPE's answer to a host it no longer serves.
const ripeAccessDenied = `% This is the RIPE Database query service.
%ERROR:201: access denied for 192.0.2.1
`
// ianaPrivateUseBlock is IANA's answer for a private-use AS: its block and
// no referral.
const ianaPrivateUseBlock = `as-block: 64512-65534
organisation: Reserved for Private Use
source: IANA
`
// newFetcherWithWHOISAnswers returns an ASN fetcher on a new database that
// holds testASNA, never looked up. The fetcher's WHOIS client connects to a
// fake server for each address in answers, which reads the query and sends
// answers[address]; connecting to any other address is refused.
func newFetcherWithWHOISAnswers(
t *testing.T, answers map[string]string,
) (*ASNFetcher, *database.Database) {
t.Helper()
db := newTestDatabase(t)
_, err := db.GetOrCreateASN(testASNA, time.Now())
if err != nil {
t.Fatalf("GetOrCreateASN: %v", err)
}
var servers sync.WaitGroup
t.Cleanup(servers.Wait)
f := NewASNFetcher(db, logger.New().Logger)
f.whoisClient.DialContext = func(
_ context.Context, _, address string,
) (net.Conn, error) {
answer, ok := answers[address]
if !ok {
return nil, syscall.ECONNREFUSED
}
client, server := net.Pipe()
servers.Go(func() {
defer func() { _ = server.Close() }()
_, _ = bufio.NewReader(server).ReadString('\n')
_, _ = io.WriteString(server, answer)
})
return client, nil
}
return f, db
}
// TestFetchAndUpdateKeepsTheRowWhenTheLookupFails looks up an AS IANA refers
// to RIPE when RIPE's server refuses the connection, and when it answers only
// with an error. The AS must stay never looked up, and the fetcher must back
// off.
func TestFetchAndUpdateKeepsTheRowWhenTheLookupFails(t *testing.T) {
t.Parallel()
for name, answers := range map[string]map[string]string{
"connection refused": {ianaWHOISServer: ianaRefersToRIPE},
"only an error": {
ianaWHOISServer: ianaRefersToRIPE,
ripeWHOISServer: ripeAccessDenied,
},
} {
t.Run(name, func(t *testing.T) {
t.Parallel()
f, db := newFetcherWithWHOISAnswers(t, answers)
if f.fetchAndUpdate(t.Context(), testASNA) {
t.Error("fetchAndUpdate reported success")
}
stats, err := db.GetWHOISStats(t.Context(), whoisStaleThreshold)
if err != nil {
t.Fatalf("GetWHOISStats: %v", err)
}
if stats.NeverFetched != 1 {
t.Errorf("never looked up = %d, want 1: the failed lookup "+
"was stored", stats.NeverFetched)
}
got := f.GetStats()
if got.CurrentInterval != baseInterval*backoffMultiplier ||
got.ConsecutiveFails != 1 {
t.Errorf("interval %v after %d failures, want %v after 1",
got.CurrentInterval, got.ConsecutiveFails,
baseInterval*backoffMultiplier)
}
})
}
}
// TestFetchAndUpdateMarksAnASWithoutReferral looks up a private-use AS, which
// IANA refers to no RIR. The AS must not be looked up again until it is
// stale, and the lookup counts as neither a success nor a failure.
func TestFetchAndUpdateMarksAnASWithoutReferral(t *testing.T) {
t.Parallel()
f, db := newFetcherWithWHOISAnswers(t,
map[string]string{ianaWHOISServer: ianaPrivateUseBlock})
f.fetchAndUpdate(t.Context(), testASNA)
asn, err := db.GetNextStaleASN(t.Context(), whoisStaleThreshold)
if !errors.Is(err, database.ErrNoStaleASN) {
t.Errorf("GetNextStaleASN = %d, %v; want ErrNoStaleASN", asn, err)
}
got := f.GetStats()
if got.CurrentInterval != baseInterval || got.SuccessesLastHour != 0 ||
got.ErrorsLastHour != 0 {
t.Errorf("interval %v, %d successes, %d errors; want %v, 0, 0",
got.CurrentInterval, got.SuccessesLastHour, got.ErrorsLastHour,
baseInterval)
}
}
// TestASNFetcherStopReturnsByItsDeadline starts a fetcher, has it look up an
// AS whose WHOIS server does not answer until the test ends, and stops it with
// a short deadline while that lookup is in progress. Stop must return all the
// same instead of waiting for the lookup.
func TestASNFetcherStopReturnsByItsDeadline(t *testing.T) {
t.Parallel()
const stopDeadline = 100 * time.Millisecond
lookupStarted := make(chan struct{})
release := make(chan struct{})
f := NewASNFetcher(newMockStore(), logger.New().Logger)
f.whoisClient.DialContext = func(
context.Context, string, string,
) (net.Conn, error) {
close(lookupStarted)
<-release
return nil, syscall.ECONNREFUSED
}
f.Start(t.Context())
f.QueueImmediate(testASNA)
select {
case <-lookupStarted:
case <-time.After(handleMessageTimeout):
t.Fatal("the fetcher did not start the lookup")
}
ctx, cancel := context.WithTimeout(t.Context(), stopDeadline)
defer cancel()
if !returnsInTime(func() { f.Stop(ctx) }) {
t.Fatalf("Stop with a deadline of %s did not return while a lookup "+
"was in progress", stopDeadline)
}
close(release)
select {
case <-f.done:
case <-time.After(handleMessageTimeout):
t.Fatal("the fetcher goroutine did not return after the lookup ended")
}
}
internal/routewatch/batchwrite_internal_test.go
+645
View File
@@ -0,0 +1,645 @@
package routewatch
import (
"bytes"
"context"
"encoding/json"
"log/slog"
"slices"
"sync"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
)
const (
// handleMessageTimeout is how long a test waits for HandleMessage to
// return, or for a write to start, before it fails.
handleMessageTimeout = 2 * time.Second
// stillWaitingAfter is how long a test watches a HandleMessage that must
// wait. One that does not wait returns well within it.
stillWaitingAfter = 100 * time.Millisecond
)
// blockingStore is a mockStore whose batch writes for the AS, peer and prefix
// handlers, delete of a peer's routes and peering write wait until the test
// calls releaseWrites. writeStarted is closed when the first of them starts. It
// records what each batch write was given.
type blockingStore struct {
*mockStore
release chan struct{}
releaseOnce sync.Once
writeStarted chan struct{}
startOnce sync.Once
mu sync.Mutex
prefixes map[string]bool
asns map[int]bool
peers map[string]bool
}
func newBlockingStore() *blockingStore {
return &blockingStore{
mockStore: newMockStore(),
release: make(chan struct{}),
writeStarted: make(chan struct{}),
prefixes: make(map[string]bool),
asns: make(map[int]bool),
peers: make(map[string]bool),
}
}
func (s *blockingStore) UpsertLiveRouteBatch(routes []*database.LiveRoute) error {
s.waitForRelease()
s.mu.Lock()
defer s.mu.Unlock()
for _, route := range routes {
s.prefixes[route.Prefix] = true
}
return nil
}
func (s *blockingStore) GetOrCreateASNBatch(asns map[int]database.SeenTimes) error {
s.waitForRelease()
s.mu.Lock()
defer s.mu.Unlock()
for asn := range asns {
s.asns[asn] = true
}
return nil
}
func (s *blockingStore) UpdatePeerBatch(peers map[string]database.PeerUpdate) error {
s.waitForRelease()
s.mu.Lock()
defer s.mu.Unlock()
for peerIP := range peers {
s.peers[peerIP] = true
}
return nil
}
func (s *blockingStore) DeleteLiveRoutesOfPeer(peerIP string) (int, error) {
s.waitForRelease()
return s.mockStore.DeleteLiveRoutesOfPeer(peerIP)
}
func (s *blockingStore) RecordPeering(
asA, asB int, seen database.SeenTimes,
) error {
s.waitForRelease()
return s.mockStore.RecordPeering(asA, asB, seen)
}
func (s *blockingStore) waitForRelease() {
s.startOnce.Do(func() { close(s.writeStarted) })
<-s.release
}
// releaseWrites lets the held writes, and all later ones, finish. A second call
// does nothing.
func (s *blockingStore) releaseWrites() {
s.releaseOnce.Do(func() { close(s.release) })
}
// stopLog holds what a handler's Stop logs it did not write.
//
//nolint:tagliatelle // field names are the keys the handlers log
type stopLog struct {
Msg string `json:"msg"`
Updates int `json:"updates"`
PeersDown int `json:"peers_down"`
Paths int `json:"paths"`
}
// newErrorLogger returns a logger that writes only errors, as JSON lines, to
// logs.
func newErrorLogger(logs *bytes.Buffer) *logger.Logger {
return &logger.Logger{Logger: slog.New(slog.NewJSONHandler(logs,
&slog.HandlerOptions{Level: slog.LevelError}))}
}
// lastLogged decodes the last line in logs, which is the one Stop logs: a
// handler logs nothing after it.
func lastLogged(t *testing.T, logs *bytes.Buffer) stopLog {
t.Helper()
lines := bytes.Split(bytes.TrimSpace(logs.Bytes()), []byte("\n"))
var logged stopLog
err := json.Unmarshal(lines[len(lines)-1], &logged)
if err != nil {
t.Fatalf("want the error Stop logs, got %q: %v", logs.String(), err)
}
return logged
}
// returnsInTime reports whether f returns within handleMessageTimeout.
func returnsInTime(f func()) bool {
done := make(chan struct{})
go func() {
f()
close(done)
}()
select {
case <-done:
return true
case <-time.After(handleMessageTimeout):
return false
}
}
// startBlockedWrite hands h the messages in fill, which fill its batch so that
// the handler writes it to store, and waits until that write starts. store
// holds it until the test calls releaseWrites.
func startBlockedWrite(
t *testing.T,
store *blockingStore,
h interface {
HandleMessage(msg *ristypes.RISMessage)
},
fill []*ristypes.RISMessage,
) {
t.Helper()
if !returnsInTime(func() {
for _, msg := range fill {
h.HandleMessage(msg)
}
}) {
t.Fatal("HandleMessage did not return after filling the batch: " +
"it waits for the database write")
}
select {
case <-store.writeStarted:
case <-time.After(handleMessageTimeout):
t.Fatal("the handler did not write its full batch")
}
}
// handleWhileAWriteIsBlocked starts a blocked write with the messages in fill
// and hands h the message extra while store holds the write. HandleMessage must
// return each time. Then it lets the writes finish and stops h, which writes
// what it holds.
func handleWhileAWriteIsBlocked(
t *testing.T,
store *blockingStore,
h interface {
HandleMessage(msg *ristypes.RISMessage)
Stop(ctx context.Context)
},
fill []*ristypes.RISMessage,
extra *ristypes.RISMessage,
) {
t.Helper()
defer h.Stop(t.Context())
defer store.releaseWrites()
startBlockedWrite(t, store, h, fill)
if !returnsInTime(func() { h.HandleMessage(extra) }) {
t.Fatal("HandleMessage did not return while a database write was blocked")
}
}
// handleAtMaxBatchSize starts a blocked write with the messages in fill, then,
// while store holds the write, hands h the messages in toMax, which bring its
// next batch to its maximum size, and the message extra. HandleMessage for
// extra must wait until the test lets the write finish, and return after. Then
// it stops h, which writes what it holds.
func handleAtMaxBatchSize(
t *testing.T,
store *blockingStore,
h interface {
HandleMessage(msg *ristypes.RISMessage)
Stop(ctx context.Context)
},
fill, toMax []*ristypes.RISMessage,
extra *ristypes.RISMessage,
) {
t.Helper()
defer h.Stop(t.Context())
defer store.releaseWrites()
startBlockedWrite(t, store, h, fill)
for _, msg := range toMax {
h.HandleMessage(msg)
}
returned := make(chan struct{})
go func() {
h.HandleMessage(extra)
close(returned)
}()
select {
case <-returned:
t.Fatal("HandleMessage returned while the batch was at its maximum " +
"size and a write was blocked")
case <-time.After(stillWaitingAfter):
}
store.releaseWrites()
select {
case <-returned:
case <-time.After(handleMessageTimeout):
t.Fatal("HandleMessage did not return after the database write finished")
}
}
// stopWhileAWriteIsBlocked starts a blocked write with the messages in fill,
// hands h the message extra, and stops h with a deadline that passes while
// store holds the write. Stop must return all the same. Then it lets the write
// finish and waits until done, closed when h's flush goroutine returns. Past
// the deadline that goroutine skips its last write, so extra is not written,
// and Stop logs as not written both extra and what the held write took.
func stopWhileAWriteIsBlocked(
t *testing.T,
store *blockingStore,
h interface {
HandleMessage(msg *ristypes.RISMessage)
Stop(ctx context.Context)
},
done <-chan struct{},
fill []*ristypes.RISMessage,
extra *ristypes.RISMessage,
) {
t.Helper()
const stopDeadline = 100 * time.Millisecond
defer store.releaseWrites()
startBlockedWrite(t, store, h, fill)
h.HandleMessage(extra)
ctx, cancel := context.WithTimeout(t.Context(), stopDeadline)
defer cancel()
if !returnsInTime(func() { h.Stop(ctx) }) {
t.Fatalf("Stop with a deadline of %s did not return while a write "+
"was blocked", stopDeadline)
}
store.releaseWrites()
select {
case <-done:
case <-time.After(handleMessageTimeout):
t.Fatal("the flush goroutine did not return after the write finished")
}
}
func TestPrefixHandlerSkipsItsLastWriteAfterTheStopDeadline(t *testing.T) {
t.Parallel()
const extraPrefix = "198.51.100.0/24"
var logs bytes.Buffer
store := newBlockingStore()
h := NewPrefixHandler(store, newErrorLogger(&logs))
fill := slices.Repeat([]string{testV4Prefix}, prefixBatchSize)
// The peer's session goes down first, so the write held is the delete of
// its routes, and Stop must log that peer down too.
stopWhileAWriteIsBlocked(t, store, h, h.done,
[]*ristypes.RISMessage{sessionDown(), prefixAnnouncement(fill)},
prefixAnnouncement([]string{extraPrefix}))
store.mu.Lock()
defer store.mu.Unlock()
if store.prefixes[extraPrefix] {
t.Errorf("%s was written after the stop deadline", extraPrefix)
}
logged := lastLogged(t, &logs)
if logged.Updates != prefixBatchSize+1 || logged.PeersDown != 1 {
t.Errorf("Stop logged %d updates and %d peer downs not written, "+
"want %d and 1", logged.Updates, logged.PeersDown, prefixBatchSize+1)
}
}
func TestASHandlerSkipsItsLastWriteAfterTheStopDeadline(t *testing.T) {
t.Parallel()
// Each message gives an update for its origin AS and one for each AS
// in its path: fill asnBatchSize+1, extra 2.
const wantUpdates = asnBatchSize + 1 + 2
var logs bytes.Buffer
store := newBlockingStore()
h := NewASHandler(store, newErrorLogger(&logs))
fill := ristypes.ASPath{Sequence: slices.Repeat([]int{testASNA}, asnBatchSize)}
stopWhileAWriteIsBlocked(t, store, h, h.done,
[]*ristypes.RISMessage{{Path: fill}},
&ristypes.RISMessage{Path: ristypes.ASPath{Sequence: []int{testASNB}}})
store.mu.Lock()
defer store.mu.Unlock()
if store.asns[testASNB] {
t.Errorf("AS%d was written after the stop deadline", testASNB)
}
if logged := lastLogged(t, &logs); logged.Updates != wantUpdates {
t.Errorf("Stop logged %d updates not written, want %d",
logged.Updates, wantUpdates)
}
}
func TestPeerHandlerSkipsItsLastWriteAfterTheStopDeadline(t *testing.T) {
t.Parallel()
const otherPeerIP = "2001:db8:ffff::2"
var logs bytes.Buffer
store := newBlockingStore()
h := NewPeerHandler(store, newErrorLogger(&logs))
stopWhileAWriteIsBlocked(t, store, h, h.done,
slices.Repeat([]*ristypes.RISMessage{{Peer: testPeerIP}}, peerBatchSize),
&ristypes.RISMessage{Peer: otherPeerIP})
store.mu.Lock()
defer store.mu.Unlock()
if store.peers[otherPeerIP] {
t.Errorf("peer %s was written after the stop deadline", otherPeerIP)
}
if logged := lastLogged(t, &logs); logged.Updates != peerBatchSize+1 {
t.Errorf("Stop logged %d updates not written, want %d",
logged.Updates, peerBatchSize+1)
}
}
// TestPeeringHandlerStopCountsTheWriteInProgress hands a running peering
// handler one AS path and stops it with a deadline that passes while store
// holds Stop's last write, of that path's peering. Stop must log that path as
// not written.
func TestPeeringHandlerStopCountsTheWriteInProgress(t *testing.T) {
t.Parallel()
const stopDeadline = 100 * time.Millisecond
var logs bytes.Buffer
store := newBlockingStore()
defer store.releaseWrites()
h := NewPeeringHandler(store, newErrorLogger(&logs))
handle(h, time.Now().UTC(), testASNA, testASNB)
ctx, cancel := context.WithTimeout(t.Context(), stopDeadline)
defer cancel()
if !returnsInTime(func() { h.Stop(ctx) }) {
t.Fatalf("Stop with a deadline of %s did not return while a write "+
"was blocked", stopDeadline)
}
store.releaseWrites()
select {
case <-h.done:
case <-time.After(handleMessageTimeout):
t.Fatal("the processing goroutine did not return after the write " +
"finished")
}
if logged := lastLogged(t, &logs); logged.Paths != 1 {
t.Errorf("Stop logged %d paths not written, want 1", logged.Paths)
}
}
// prefixAnnouncement returns a message in which testPeerIP announces prefixes
// with origin AS testASNA.
func prefixAnnouncement(prefixes []string) *ristypes.RISMessage {
return &ristypes.RISMessage{
Peer: testPeerIP,
Path: ristypes.ASPath{Sequence: []int{testASNA}},
Announcements: []ristypes.RISAnnouncement{
{NextHop: testPeerIP, Prefixes: prefixes},
},
}
}
func TestPrefixHandlerHandlesMessagesWhileAWriteIsBlocked(t *testing.T) {
t.Parallel()
const extraPrefix = "198.51.100.0/24"
store := newBlockingStore()
h := NewPrefixHandler(store, logger.New())
fill := slices.Repeat([]string{testV4Prefix}, prefixBatchSize)
handleWhileAWriteIsBlocked(t, store, h,
[]*ristypes.RISMessage{prefixAnnouncement(fill)},
prefixAnnouncement([]string{extraPrefix}))
store.mu.Lock()
defer store.mu.Unlock()
if !store.prefixes[extraPrefix] {
t.Errorf("Stop did not write %s, handled while a write was blocked",
extraPrefix)
}
}
func TestASHandlerHandlesMessagesWhileAWriteIsBlocked(t *testing.T) {
t.Parallel()
store := newBlockingStore()
h := NewASHandler(store, logger.New())
fill := ristypes.ASPath{Sequence: slices.Repeat([]int{testASNA}, asnBatchSize)}
handleWhileAWriteIsBlocked(t, store, h,
[]*ristypes.RISMessage{{Path: fill}},
&ristypes.RISMessage{Path: ristypes.ASPath{Sequence: []int{testASNB}}})
store.mu.Lock()
defer store.mu.Unlock()
if !store.asns[testASNB] {
t.Errorf("Stop did not write AS%d, handled while a write was blocked",
testASNB)
}
}
func TestPeerHandlerHandlesMessagesWhileAWriteIsBlocked(t *testing.T) {
t.Parallel()
const otherPeerIP = "2001:db8:ffff::2"
store := newBlockingStore()
h := NewPeerHandler(store, logger.New())
handleWhileAWriteIsBlocked(t, store, h,
slices.Repeat([]*ristypes.RISMessage{{Peer: testPeerIP}}, peerBatchSize),
&ristypes.RISMessage{Peer: otherPeerIP})
store.mu.Lock()
defer store.mu.Unlock()
if !store.peers[otherPeerIP] {
t.Errorf("Stop did not write peer %s, handled while a write was blocked",
otherPeerIP)
}
}
func TestPrefixHandlerWaitsAtTheMaximumBatchSize(t *testing.T) {
t.Parallel()
const extraPrefix = "198.51.100.0/24"
store := newBlockingStore()
h := NewPrefixHandler(store, logger.New())
fill := slices.Repeat([]string{testV4Prefix}, prefixBatchSize)
toMax := slices.Repeat([]string{testV4Prefix}, prefixBatchMaxSize)
handleAtMaxBatchSize(t, store, h,
[]*ristypes.RISMessage{prefixAnnouncement(fill)},
[]*ristypes.RISMessage{prefixAnnouncement(toMax)},
prefixAnnouncement([]string{extraPrefix}))
store.mu.Lock()
defer store.mu.Unlock()
if !store.prefixes[extraPrefix] {
t.Errorf("Stop did not write %s, handled after the batch reached its "+
"maximum size", extraPrefix)
}
}
func TestASHandlerWaitsAtTheMaximumBatchSize(t *testing.T) {
t.Parallel()
store := newBlockingStore()
h := NewASHandler(store, logger.New())
fill := ristypes.ASPath{Sequence: slices.Repeat([]int{testASNA}, asnBatchSize)}
toMax := ristypes.ASPath{Sequence: slices.Repeat([]int{testASNA}, asnBatchMaxSize)}
handleAtMaxBatchSize(t, store, h,
[]*ristypes.RISMessage{{Path: fill}},
[]*ristypes.RISMessage{{Path: toMax}},
&ristypes.RISMessage{Path: ristypes.ASPath{Sequence: []int{testASNB}}})
store.mu.Lock()
defer store.mu.Unlock()
if !store.asns[testASNB] {
t.Errorf("Stop did not write AS%d, handled after the batch reached its "+
"maximum size", testASNB)
}
}
func TestPeerHandlerWaitsAtTheMaximumBatchSize(t *testing.T) {
t.Parallel()
const otherPeerIP = "2001:db8:ffff::2"
store := newBlockingStore()
h := NewPeerHandler(store, logger.New())
message := []*ristypes.RISMessage{{Peer: testPeerIP}}
handleAtMaxBatchSize(t, store, h,
slices.Repeat(message, peerBatchSize),
slices.Repeat(message, peerBatchMaxSize),
&ristypes.RISMessage{Peer: otherPeerIP})
store.mu.Lock()
defer store.mu.Unlock()
if !store.peers[otherPeerIP] {
t.Errorf("Stop did not write peer %s, handled after the batch reached "+
"its maximum size", otherPeerIP)
}
}
// TestHandlersWakeTheFlushGoroutineWhenTheBatchIsFull hands each handler, built
// without its flush goroutine so that nothing takes a wake-up, one update short
// of a full batch, and a second one exactly a full batch. Only the full batch
// leaves a wake-up in batchFullCh.
func TestHandlersWakeTheFlushGoroutineWhenTheBatchIsFull(t *testing.T) {
t.Parallel()
tests := []struct {
name string
batchSize int
// handle hands a new handler n updates and returns its batchFullCh.
handle func(n int) chan struct{}
}{
{
name: "prefix",
batchSize: prefixBatchSize,
handle: func(n int) chan struct{} {
h := &PrefixHandler{batchFullCh: make(chan struct{}, 1)}
h.HandleMessage(prefixAnnouncement(
slices.Repeat([]string{testV4Prefix}, n)))
return h.batchFullCh
},
},
{
name: "AS",
batchSize: asnBatchSize,
handle: func(n int) chan struct{} {
h := &ASHandler{batchFullCh: make(chan struct{}, 1)}
// The origin AS and each AS in the path are one update each.
h.HandleMessage(&ristypes.RISMessage{
Path: ristypes.ASPath{Sequence: slices.Repeat([]int{testASNA}, n-1)},
})
return h.batchFullCh
},
},
{
name: "peer",
batchSize: peerBatchSize,
handle: func(n int) chan struct{} {
h := &PeerHandler{batchFullCh: make(chan struct{}, 1)}
for range n {
h.HandleMessage(&ristypes.RISMessage{Peer: testPeerIP})
}
return h.batchFullCh
},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
t.Parallel()
got := len(tt.handle(tt.batchSize - 1))
if got != 0 {
t.Errorf("one update short of a full batch left %d wake-ups, "+
"want none", got)
}
got = len(tt.handle(tt.batchSize))
if got != 1 {
t.Errorf("a full batch left %d wake-ups, want 1", got)
}
})
}
}
internal/routewatch/cli.go
+36 -26
View File
@@ -3,10 +3,8 @@ package routewatch
import (
"context"
"os"
"os/signal"
"runtime"
"strings"
"syscall"
"time"
"git.eeqj.de/sneak/routewatch/internal/logger"
@@ -14,12 +12,20 @@ import (
)
const (
// shutdownTimeout is the maximum time allowed for graceful shutdown
shutdownTimeout = 60 * time.Second
// shutdownTimeout is how long Shutdown has after a stop signal. upaas,
// like docker stop by default, kills the container 10 seconds after the
// stop signal (README "Running under upaas").
shutdownTimeout = 8 * time.Second
// stopTimeout is fx's limit on the stop hook, past which fx exits with
// status 1: shutdownTimeout and time for Shutdown to close the HTTP
// server after its deadline.
stopTimeout = 9 * time.Second
// debugInterval is how often to log debug stats
debugInterval = 60 * time.Second
// bytesPerMB is bytes per megabyte
bytesPerMB = 1024 * 1024
// failureExitCode is the exit status when RouteWatch stops on an error
failureExitCode = 1
)
// logDebugStats logs goroutine count and memory usage
@@ -54,43 +60,47 @@ func logDebugStats(logger *logger.Logger) {
}
// CLIEntry is the main entry point for the routewatch command-line interface.
// It initializes the application using the fx dependency injection framework,
// sets up signal handling for graceful shutdown, and starts the RouteWatch service.
// This function blocks until the application receives a shutdown signal or encounters
// a fatal error.
// It initializes the application using the fx dependency injection framework
// and starts the RouteWatch service. It blocks until fx receives SIGINT or
// SIGTERM or RouteWatch stops on an error; fx then runs the stop hook and sets
// the exit code.
func CLIEntry() {
app := fx.New(
getModule(),
fx.StopTimeout(shutdownTimeout), // Allow 60 seconds for graceful shutdown
fx.Invoke(func(lc fx.Lifecycle, rw *RouteWatch, logger *logger.Logger, shutdowner fx.Shutdowner) {
fx.Options(getModule()...),
fx.StopTimeout(stopTimeout),
fx.Invoke(func(
lc fx.Lifecycle,
rw *RouteWatch,
logger *logger.Logger,
shutdowner fx.Shutdowner,
) {
lc.Append(fx.Hook{
OnStart: func(ctx context.Context) error {
// Start debug stats logging
go logDebugStats(logger)
// Handle shutdown signals
sigCh := make(chan os.Signal, 1)
signal.Notify(sigCh, os.Interrupt, syscall.SIGTERM)
go func() {
<-sigCh
logger.Info("Received shutdown signal")
if err := shutdowner.Shutdown(); err != nil {
logger.Error("Failed to shutdown gracefully", "error", err)
}
}()
go func() {
if err := rw.Run(ctx); err != nil {
err := rw.Run(ctx)
if err != nil {
logger.Error("RouteWatch error", "error", err)
err = shutdowner.Shutdown(fx.ExitCode(failureExitCode))
if err != nil {
logger.Error("Failed to shutdown gracefully",
"error", err)
}
}
}()
return nil
},
OnStop: func(_ context.Context) error {
OnStop: func(ctx context.Context) error {
logger.Info("Shutting down RouteWatch")
rw.Shutdown()
ctx, cancel := context.WithTimeout(ctx, shutdownTimeout)
defer cancel()
rw.Shutdown(ctx)
return nil
},
internal/routewatch/dbmaintainer.go
+87 -68
View File
@@ -1,8 +1,8 @@
// Package routewatch contains the database maintainer for background maintenance tasks.
package routewatch
import (
"context"
"errors"
"log/slog"
"sync"
"time"
@@ -35,12 +35,18 @@ const (
analyzeTimeout = 5 * time.Minute
)
// checkpointIncompleteMessage is logged, at debug level like a completed
// checkpoint, when a checkpoint leaves part of the WAL for a later one: while
// reads overlap, most do. The database warns when the WAL file stays over its
// size limit.
const checkpointIncompleteMessage = "WAL checkpoint incomplete"
// DBMaintainer handles background database maintenance tasks.
type DBMaintainer struct {
db database.Store
logger *slog.Logger
stopCh chan struct{}
wg sync.WaitGroup
done chan struct{} // closed when the maintenance goroutine returns
// Stats tracking
statsMu sync.Mutex
@@ -61,13 +67,15 @@ func NewDBMaintainer(db database.Store, logger *slog.Logger) *DBMaintainer {
db: db,
logger: logger.With("component", "db_maintainer"),
stopCh: make(chan struct{}),
done: make(chan struct{}),
}
}
// Start begins the background maintenance goroutine.
func (m *DBMaintainer) Start() {
m.wg.Add(1)
go m.run()
// Start begins the background maintenance goroutine. It runs until Stop:
// the end of ctx does not stop it.
func (m *DBMaintainer) Start(ctx context.Context) {
go m.run(context.WithoutCancel(ctx))
m.logger.Info("Database maintainer started",
"checkpoint_interval", checkpointInterval,
"vacuum_interval", vacuumInterval,
@@ -75,23 +83,32 @@ func (m *DBMaintainer) Start() {
)
}
// Stop gracefully shuts down the maintainer.
func (m *DBMaintainer) Stop() {
// Stop stops the maintainer and waits, until ctx ends, for a maintenance task
// in progress to finish.
func (m *DBMaintainer) Stop(ctx context.Context) {
close(m.stopCh)
m.wg.Wait()
m.logger.Info("Database maintainer stopped")
select {
case <-m.done:
m.logger.Info("Database maintainer stopped")
case <-ctx.Done():
m.logger.Warn("Stop deadline passed before the database maintainer " +
"stopped")
}
}
// run is the main background loop.
func (m *DBMaintainer) run() {
defer m.wg.Done()
func (m *DBMaintainer) run(ctx context.Context) {
defer close(m.done)
// Use different timers for each task
checkpointTimer := time.NewTimer(checkpointInterval)
vacuumTimer := time.NewTimer(vacuumInterval)
analyzeTimer := time.NewTimer(analyzeInterval)
defer checkpointTimer.Stop()
vacuumTimer := time.NewTimer(vacuumInterval)
defer vacuumTimer.Stop()
analyzeTimer := time.NewTimer(analyzeInterval)
defer analyzeTimer.Stop()
for {
@@ -100,90 +117,92 @@ func (m *DBMaintainer) run() {
return
case <-checkpointTimer.C:
m.runCheckpoint()
m.runCheckpoint(ctx)
checkpointTimer.Reset(checkpointInterval)
case <-vacuumTimer.C:
m.runVacuum()
m.runVacuum(ctx)
vacuumTimer.Reset(vacuumInterval)
case <-analyzeTimer.C:
m.runAnalyze()
m.runAnalyze(ctx)
analyzeTimer.Reset(analyzeInterval)
}
}
}
// runCheckpoint performs a WAL checkpoint to keep the WAL file small.
func (m *DBMaintainer) runCheckpoint() {
ctx, cancel := context.WithTimeout(context.Background(), checkpointTimeout)
defer cancel()
func (m *DBMaintainer) runCheckpoint(ctx context.Context) {
duration, err := runWithTimeout(ctx, checkpointTimeout, m.db.Checkpoint)
m.recordResult(&m.lastCheckpoint, &m.lastCheckpointError, &m.checkpointCount, err)
startTime := time.Now()
err := m.db.Checkpoint(ctx)
m.statsMu.Lock()
m.lastCheckpoint = time.Now()
m.lastCheckpointError = err
if err == nil {
m.checkpointCount++
}
m.statsMu.Unlock()
if err != nil {
m.logger.Error("WAL checkpoint failed", "error", err, "duration", time.Since(startTime))
} else {
m.logger.Debug("WAL checkpoint completed", "duration", time.Since(startTime))
switch {
case errors.Is(err, database.ErrCheckpointIncomplete):
m.logger.Debug(checkpointIncompleteMessage,
"error", err, "duration", duration)
case err != nil:
m.logger.Error("WAL checkpoint failed", "error", err, "duration", duration)
default:
m.logger.Debug("WAL checkpoint completed", "duration", duration)
}
}
// runVacuum performs an incremental vacuum operation on the database.
func (m *DBMaintainer) runVacuum() {
ctx, cancel := context.WithTimeout(context.Background(), vacuumTimeout)
defer cancel()
func (m *DBMaintainer) runVacuum(ctx context.Context) {
m.logger.Debug("Running incremental vacuum")
startTime := time.Now()
err := m.db.Vacuum(ctx)
m.statsMu.Lock()
m.lastVacuum = time.Now()
m.lastVacuumError = err
if err == nil {
m.vacuumCount++
}
m.statsMu.Unlock()
duration, err := runWithTimeout(ctx, vacuumTimeout, m.db.Vacuum)
m.recordResult(&m.lastVacuum, &m.lastVacuumError, &m.vacuumCount, err)
if err != nil {
m.logger.Error("Incremental vacuum failed", "error", err, "duration", time.Since(startTime))
m.logger.Error("Incremental vacuum failed", "error", err, "duration", duration)
} else {
m.logger.Debug("Incremental vacuum completed", "duration", time.Since(startTime))
m.logger.Debug("Incremental vacuum completed", "duration", duration)
}
}
// runAnalyze performs an ANALYZE operation on the database.
func (m *DBMaintainer) runAnalyze() {
ctx, cancel := context.WithTimeout(context.Background(), analyzeTimeout)
defer cancel()
func (m *DBMaintainer) runAnalyze(ctx context.Context) {
m.logger.Info("Starting database ANALYZE")
startTime := time.Now()
err := m.db.Analyze(ctx)
m.statsMu.Lock()
m.lastAnalyze = time.Now()
m.lastAnalyzeError = err
if err == nil {
m.analyzeCount++
}
m.statsMu.Unlock()
duration, err := runWithTimeout(ctx, analyzeTimeout, m.db.Analyze)
m.recordResult(&m.lastAnalyze, &m.lastAnalyzeError, &m.analyzeCount, err)
if err != nil {
m.logger.Error("ANALYZE failed", "error", err, "duration", time.Since(startTime))
m.logger.Error("ANALYZE failed", "error", err, "duration", duration)
} else {
m.logger.Info("ANALYZE completed", "duration", time.Since(startTime))
m.logger.Info("ANALYZE completed", "duration", duration)
}
}
// runWithTimeout runs operation with a context that ends after timeout and
// returns how long it took and its error.
func runWithTimeout(
ctx context.Context,
timeout time.Duration,
operation func(context.Context) error,
) (time.Duration, error) {
ctx, cancel := context.WithTimeout(ctx, timeout)
defer cancel()
startTime := time.Now()
err := operation(ctx)
return time.Since(startTime), err
}
// recordResult stores when a maintenance task last ran and its error, and
// counts it when it succeeded.
func (m *DBMaintainer) recordResult(
lastRun *time.Time, lastErr *error, successCount *int, err error,
) {
m.statsMu.Lock()
defer m.statsMu.Unlock()
*lastRun = time.Now()
*lastErr = err
if err == nil {
*successCount++
}
}
internal/routewatch/dbmaintainer_internal_test.go
+120
View File
@@ -0,0 +1,120 @@
package routewatch
import (
"bytes"
"context"
"encoding/json"
"log/slog"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/database"
)
// incompleteCheckpointStore is a store whose every checkpoint leaves part of
// the WAL uncopied.
type incompleteCheckpointStore struct{ *mockStore }
// Checkpoint reports the checkpoint incomplete.
func (incompleteCheckpointStore) Checkpoint(context.Context) error {
return database.ErrCheckpointIncomplete
}
// TestIncompleteCheckpointNotLoggedAsCompleted runs a checkpoint that leaves
// part of the WAL uncopied and checks that the maintainer logs it as
// incomplete, not as completed, and does not count it as a checkpoint that
// completed.
func TestIncompleteCheckpointNotLoggedAsCompleted(t *testing.T) {
t.Parallel()
var logs bytes.Buffer
logger := slog.New(slog.NewJSONHandler(&logs,
&slog.HandlerOptions{Level: slog.LevelDebug}))
m := NewDBMaintainer(incompleteCheckpointStore{newMockStore()}, logger)
m.runCheckpoint(t.Context())
var logged struct {
Msg string `json:"msg"`
}
err := json.Unmarshal(logs.Bytes(), &logged)
if err != nil {
t.Fatalf("want one log message, got %q: %v", logs.String(), err)
}
if logged.Msg != checkpointIncompleteMessage {
t.Errorf("logged %q, want %q", logged.Msg, checkpointIncompleteMessage)
}
if m.checkpointCount != 0 {
t.Errorf("checkpointCount = %d after an incomplete checkpoint, want 0",
m.checkpointCount)
}
}
// heldCheckpointStore is a store whose checkpoint closes started and then
// waits until release is closed.
type heldCheckpointStore struct {
*mockStore
started chan struct{}
release chan struct{}
}
// Checkpoint waits until release is closed.
func (s heldCheckpointStore) Checkpoint(context.Context) error {
close(s.started)
<-s.release
return nil
}
// TestDBMaintainerStopReturnsByItsDeadline stops the maintainer with a short
// deadline while its goroutine runs a checkpoint that the store holds until
// the test ends. Stop must return all the same instead of waiting for the
// checkpoint.
func TestDBMaintainerStopReturnsByItsDeadline(t *testing.T) {
t.Parallel()
const stopDeadline = 100 * time.Millisecond
store := heldCheckpointStore{
mockStore: newMockStore(),
started: make(chan struct{}),
release: make(chan struct{}),
}
m := NewDBMaintainer(store, slog.Default())
// Start would run the first checkpoint only after checkpointInterval.
// This goroutine stands for the one Start runs and checkpoints at once.
go func() {
defer close(m.done)
m.runCheckpoint(t.Context())
}()
select {
case <-store.started:
case <-time.After(handleMessageTimeout):
t.Fatal("the maintainer did not start the checkpoint")
}
ctx, cancel := context.WithTimeout(t.Context(), stopDeadline)
defer cancel()
if !returnsInTime(func() { m.Stop(ctx) }) {
t.Fatalf("Stop with a deadline of %s did not return while a "+
"checkpoint was in progress", stopDeadline)
}
close(store.release)
select {
case <-m.done:
case <-time.After(handleMessageTimeout):
t.Fatal("the maintenance goroutine did not return after the " +
"checkpoint ended")
}
}
internal/routewatch/handler.go
+12 -14
View File
@@ -1,6 +1,8 @@
package routewatch
import (
"slices"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
)
@@ -15,10 +17,11 @@ type SimpleHandler struct {
callback func(*ristypes.RISMessage)
}
// NewSimpleHandler creates a new SimpleHandler that accepts specific message types.
// The messageTypes parameter specifies which RIS message types this handler will process.
// If messageTypes is empty, the handler will accept all message types.
// The callback function is invoked for each message that passes the type filter.
// NewSimpleHandler creates a new SimpleHandler that accepts specific message
// types. The messageTypes parameter specifies which RIS message types this
// handler will process. If messageTypes is empty, the handler will accept all
// message types. The callback function is invoked for each message that
// passes the type filter.
func NewSimpleHandler(
logger *logger.Logger,
messageTypes []string,
@@ -31,22 +34,17 @@ func NewSimpleHandler(
}
}
// WantsMessage returns true if this handler wants to process messages of the given type.
// It checks whether messageType is in the handler's configured list of accepted types.
// If no specific types were configured (empty messageTypes slice), it returns true for all types.
// WantsMessage returns true if this handler wants to process messages of the
// given type. It checks whether messageType is in the handler's configured
// list of accepted types. If no specific types were configured (empty
// messageTypes slice), it returns true for all types.
func (h *SimpleHandler) WantsMessage(messageType string) bool {
// If no specific types are set, accept all messages
if len(h.messageTypes) == 0 {
return true
}
for _, t := range h.messageTypes {
if t == messageType {
return true
}
}
return false
return slices.Contains(h.messageTypes, messageType)
}
// HandleMessage processes a RIS message by invoking the configured callback function.
internal/routewatch/peerhandler.go
+139 -62
View File
@@ -1,9 +1,11 @@
package routewatch
// peerhandler.go provides batched peer tracking functionality for BGP route monitoring.
// It tracks BGP peers from all incoming RIS messages and maintains peer state in the database.
// peerhandler.go provides batched peer tracking functionality for BGP route
// monitoring. It tracks BGP peers from incoming UPDATE messages and maintains
// peer state in the database.
import (
"context"
"strconv"
"sync"
"time"
@@ -23,24 +25,42 @@ const (
// peerBatchSize is the number of peer updates to batch together
peerBatchSize = 10000
// peerBatchMaxSize is the most peer updates the batch holds while the
// flush goroutine writes the previous one: ten batches, about 10 MiB
// (README "Memory"). At this size HandleMessage waits until the flush
// goroutine takes the batch.
peerBatchMaxSize = 100000
// peerBatchTimeout is the maximum time to wait before flushing a batch
peerBatchTimeout = 2 * time.Second
)
// PeerHandler tracks BGP peers from all message types using batched operations.
// PeerHandler tracks BGP peers from UPDATE messages using batched operations.
// It maintains a queue of peer updates and periodically flushes them to the database
// in batches to improve performance. The handler deduplicates peer updates within
// each batch, keeping only the most recent update for each peer IP address.
// in batches to improve performance. The handler writes one row per peer IP
// address for each batch, with the earliest and the latest time of its updates.
type PeerHandler struct {
db database.Store
logger *logger.Logger
// Batching
// Batching. mu guards peerBatch and writing only: the flush goroutine
// takes the batch under mu and writes it after releasing mu, so
// HandleMessage waits for a database write only while peerBatch is at
// peerBatchMaxSize.
mu sync.Mutex
peerBatch []peerUpdate
lastFlush time.Time
stopCh chan struct{}
wg sync.WaitGroup
// writing is how many updates the flush goroutine took from peerBatch
// and has not finished writing.
writing int
// batchTaken is signalled when the flush goroutine takes peerBatch; its L
// is &mu.
batchTaken sync.Cond
batchFullCh chan struct{} // wakes the flush goroutine when peerBatch is full
lastFlush time.Time // used only by the flush goroutine
// stopCh hands Stop's context to the flush goroutine. It holds one, so
// Stop does not wait for a write in progress to hand it over.
stopCh chan context.Context
done chan struct{} // closed when the flush goroutine returns
}
type peerUpdate struct {
@@ -56,25 +76,27 @@ type peerUpdate struct {
// should be stopped by calling Stop when it is no longer needed.
func NewPeerHandler(db database.Store, logger *logger.Logger) *PeerHandler {
h := &PeerHandler{
db: db,
logger: logger,
peerBatch: make([]peerUpdate, 0, peerBatchSize),
lastFlush: time.Now(),
stopCh: make(chan struct{}),
db: db,
logger: logger,
peerBatch: make([]peerUpdate, 0, peerBatchSize),
batchFullCh: make(chan struct{}, 1),
lastFlush: time.Now(),
stopCh: make(chan context.Context, 1),
done: make(chan struct{}),
}
h.batchTaken.L = &h.mu
// Start the flush timer goroutine
h.wg.Add(1)
go h.flushLoop()
return h
}
// WantsMessage returns true for all message types since peer information
// is extracted from every RIS message regardless of type. This satisfies
// the MessageHandler interface.
func (h *PeerHandler) WantsMessage(_ string) bool {
return true
// WantsMessage returns true for UPDATE messages only. A peer whose session
// never comes up sends STATE messages and nothing else, so a STATE message is
// not a sign of a peer that carries routes.
func (h *PeerHandler) WantsMessage(messageType string) bool {
return messageType == ristypes.MessageTypeUpdate
}
// QueueCapacity returns the desired queue capacity for this handler.
@@ -87,13 +109,15 @@ func (h *PeerHandler) QueueCapacity() int {
// HandleMessage processes a RIS message to track peer information.
// It extracts the peer IP address and ASN from the message and adds
// the update to an internal batch. When the batch reaches peerBatchSize
// or the batch timeout expires, the batch is flushed to the database.
// the update to an internal batch. The flush goroutine writes the batch to the
// database when it reaches peerBatchSize or the batch timeout expires.
func (h *PeerHandler) HandleMessage(msg *ristypes.RISMessage) {
// Parse peer ASN from string
peerASN := 0
if msg.PeerASN != "" {
if asn, err := strconv.Atoi(msg.PeerASN); err == nil {
asn, err := strconv.Atoi(msg.PeerASN)
if err == nil {
peerASN = asn
}
}
@@ -101,6 +125,13 @@ func (h *PeerHandler) HandleMessage(msg *ristypes.RISMessage) {
h.mu.Lock()
defer h.mu.Unlock()
// While the batch is at its maximum size, wait until the flush goroutine
// takes it. Meanwhile the handler takes nothing off its queue, which
// fills, and the streamer drops messages for it.
for len(h.peerBatch) >= peerBatchMaxSize {
h.batchTaken.Wait()
}
// Add to batch
h.peerBatch = append(h.peerBatch, peerUpdate{
peerIP: msg.Peer,
@@ -109,74 +140,120 @@ func (h *PeerHandler) HandleMessage(msg *ristypes.RISMessage) {
timestamp: msg.ParsedTimestamp,
})
// Check if we need to flush
// A full batch wakes the flush goroutine. If it is writing, the wake-up
// waits in batchFullCh and the batch keeps growing until the write ends
// or the batch reaches peerBatchMaxSize.
if len(h.peerBatch) >= peerBatchSize {
h.flushBatchLocked()
select {
case h.batchFullCh <- struct{}{}:
default: // already woken
}
}
}
// flushLoop runs in a goroutine and periodically flushes batches
// Stop stops the flush goroutine, which writes the updates the handler holds
// before it returns, and waits for it until ctx ends. Once ctx has ended the
// flush goroutine skips that last write, and a write still running is lost
// when the process exits, so Stop logs how many updates are not written.
func (h *PeerHandler) Stop(ctx context.Context) {
h.stopCh <- ctx
select {
case <-h.done:
case <-ctx.Done():
}
h.mu.Lock()
unwritten := h.writing + len(h.peerBatch)
h.mu.Unlock()
if unwritten > 0 {
h.logger.Error("Stop deadline passed before the peer handler wrote "+
"its updates", "updates", unwritten)
}
}
// flushLoop runs in a goroutine, the only one that writes the batch: it
// flushes when the batch is full, periodically, and once more on Stop.
func (h *PeerHandler) flushLoop() {
defer h.wg.Done()
defer close(h.done)
ticker := time.NewTicker(peerBatchTimeout)
defer ticker.Stop()
for {
select {
case <-ticker.C:
h.mu.Lock()
if time.Since(h.lastFlush) >= peerBatchTimeout {
h.flushBatchLocked()
h.flushBatch()
}
case <-h.batchFullCh:
h.flushBatch()
case ctx := <-h.stopCh:
// Final flush, unless the stop deadline has passed
if ctx.Err() == nil {
h.flushBatch()
}
h.mu.Unlock()
case <-h.stopCh:
// Final flush
h.mu.Lock()
h.flushBatchLocked()
h.mu.Unlock()
return
}
}
}
// flushBatchLocked flushes the peer batch to the database (must be called with mutex held)
func (h *PeerHandler) flushBatchLocked() {
// takeBatch returns the pending updates and leaves an empty batch in their
// place for HandleMessage to fill. They count in writing until flushBatch has
// written them.
func (h *PeerHandler) takeBatch() []peerUpdate {
h.mu.Lock()
defer h.mu.Unlock()
if len(h.peerBatch) == 0 {
return nil
}
batch := h.peerBatch
h.writing = len(batch)
h.peerBatch = make([]peerUpdate, 0, peerBatchSize)
h.batchTaken.Broadcast()
return batch
}
// flushBatch takes the pending batch and writes it to the database, without
// holding the mutex during the write.
func (h *PeerHandler) flushBatch() {
batch := h.takeBatch()
if len(batch) == 0 {
return
}
// Deduplicate by peer IP, keeping the latest update for each peer
peerMap := make(map[string]peerUpdate)
for _, update := range h.peerBatch {
if existing, ok := peerMap[update.peerIP]; !ok || update.timestamp.After(existing.timestamp) {
peerMap[update.peerIP] = update
// The earliest and the latest time the batch saw each peer, with the peer
// AS and message type of its latest update. The batch holds several
// updates for a peer, so a new peer is stored first seen at the earliest of
// them and last seen at the latest.
peers := make(map[string]database.PeerUpdate)
for _, update := range batch {
peer, ok := peers[update.peerIP]
if !ok || update.timestamp.After(peer.Seen.Last) {
peer.PeerIP = update.peerIP
peer.PeerASN = update.peerASN
peer.MessageType = update.messageType
}
}
// Convert to database format
dbPeerMap := make(map[string]database.PeerUpdate)
for peerIP, update := range peerMap {
dbPeerMap[peerIP] = database.PeerUpdate{
PeerIP: update.peerIP,
PeerASN: update.peerASN,
MessageType: update.messageType,
Timestamp: update.timestamp,
}
peer.Seen.Add(update.timestamp)
peers[update.peerIP] = peer
}
// Process all peers in a single batch transaction
if err := h.db.UpdatePeerBatch(dbPeerMap); err != nil {
h.logger.Error("Failed to process peer batch", "error", err, "count", len(dbPeerMap))
err := h.db.UpdatePeerBatch(peers)
if err != nil {
h.logger.Error("Failed to process peer batch",
"error", err, "count", len(peers))
}
// Clear batch
h.peerBatch = h.peerBatch[:0]
h.mu.Lock()
h.writing = 0
h.mu.Unlock()
h.lastFlush = time.Now()
}
// Stop gracefully stops the handler and flushes remaining batches
func (h *PeerHandler) Stop() {
close(h.stopCh)
h.wg.Wait()
}
internal/routewatch/peerhandler_internal_test.go
+59
View File
@@ -0,0 +1,59 @@
package routewatch
import (
"testing"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
)
// peerBatchStore keeps the peers of the last UpdatePeerBatch call. The
// database has no read of the peer table, so the peer handler's test checks
// what the handler writes.
type peerBatchStore struct {
*mockStore
peers map[string]database.PeerUpdate
}
func (s *peerBatchStore) UpdatePeerBatch(
peers map[string]database.PeerUpdate,
) error {
s.peers = peers
return nil
}
// TestPeerHandlerWritesTheEarliestAndLatestTimesOfANewPeer handles two updates
// from one peer, the second 5 seconds after the first, in one batch. The peer
// handler writes the peer first seen at the time of the first update and last
// seen at the time of the second.
func TestPeerHandlerWritesTheEarliestAndLatestTimesOfANewPeer(t *testing.T) {
t.Parallel()
store := &peerBatchStore{mockStore: newMockStore()}
// Built without NewPeerHandler's flush goroutine, so the test flushes
// itself.
h := &PeerHandler{db: store, logger: logger.New()}
first := peerAnnouncement(testPeerIP, testV4Prefix)
second := peerAnnouncement(testPeerIP, testV4Prefix)
second.ParsedTimestamp = first.ParsedTimestamp.Add(secondUpdateAfter)
h.HandleMessage(first)
h.HandleMessage(second)
h.flushBatch()
seen := store.peers[testPeerIP].Seen
if !seen.First.Equal(first.ParsedTimestamp) {
t.Errorf("first_seen of peer %s = %s, want %s", testPeerIP,
seen.First, first.ParsedTimestamp)
}
if !seen.Last.Equal(second.ParsedTimestamp) {
t.Errorf("last_seen of peer %s = %s, want %s", testPeerIP,
seen.Last, second.ParsedTimestamp)
}
}
internal/routewatch/peeringhandler.go
+102 -53
View File
@@ -1,6 +1,7 @@
package routewatch
import (
"context"
"encoding/json"
"sync"
"time"
@@ -41,12 +42,19 @@ type PeeringHandler struct {
db database.Store
logger *logger.Logger
// In-memory AS path tracking
// In-memory AS path tracking: each path, JSON-encoded, with the earliest
// and the latest time an update carried it.
mu sync.Mutex
asPaths map[string]time.Time // key is JSON-encoded AS path
droppedPaths int // paths dropped because the map was full
asPaths map[string]database.SeenTimes
droppedPaths int // paths dropped because the map was full
// writing is how many paths processPeerings took from asPaths and has
// not finished writing the peerings of.
writing int
stopCh chan struct{}
// stopCh hands Stop's context to the processing goroutine. It holds one,
// so Stop does not wait for a write in progress to hand it over.
stopCh chan context.Context
done chan struct{} // closed when the processing goroutine returns
}
// NewPeeringHandler creates and initializes a new PeeringHandler with the
@@ -57,8 +65,9 @@ func NewPeeringHandler(db database.Store, logger *logger.Logger) *PeeringHandler
h := &PeeringHandler{
db: db,
logger: logger,
asPaths: make(map[string]time.Time),
stopCh: make(chan struct{}),
asPaths: make(map[string]database.SeenTimes),
stopCh: make(chan context.Context, 1),
done: make(chan struct{}),
}
// Start the periodic processing goroutine
@@ -72,7 +81,7 @@ func NewPeeringHandler(db database.Store, logger *logger.Logger) *PeeringHandler
// the AS path information needed to extract peering relationships.
func (h *PeeringHandler) WantsMessage(messageType string) bool {
// We only care about UPDATE messages that have AS paths
return messageType == "UPDATE"
return messageType == ristypes.MessageTypeUpdate
}
// QueueCapacity returns the buffer size for the handler's message queue.
@@ -87,51 +96,41 @@ func (h *PeeringHandler) QueueCapacity() int {
// than minPathLengthForPeering are ignored as they cannot contain valid
// peering information.
func (h *PeeringHandler) HandleMessage(msg *ristypes.RISMessage) {
// Peerings are between neighbours in the path's sequence only: the
// members of an AS set that ends the path are not neighbours of each
// other or of the AS before them.
sequence := msg.Path.Sequence
// Skip if no AS path or only one AS
if len(msg.Path) < minPathLengthForPeering {
if len(sequence) < minPathLengthForPeering {
return
}
timestamp := msg.ParsedTimestamp
// Encode AS path as JSON for use as map key
pathJSON, err := json.Marshal(msg.Path)
pathJSON, err := json.Marshal(sequence)
if err != nil {
h.logger.Error("Failed to encode AS path", "error", err)
return
}
key := string(pathJSON)
h.mu.Lock()
if _, exists := h.asPaths[key]; exists {
// Already tracked: refresh its timestamp.
h.asPaths[key] = timestamp
} else if len(h.asPaths) >= maxTrackedPaths {
defer h.mu.Unlock()
seen, exists := h.asPaths[key]
if !exists && len(h.asPaths) >= maxTrackedPaths {
// Map is full; drop this new path and count it.
h.droppedPaths++
} else {
h.asPaths[key] = timestamp
return
}
h.mu.Unlock()
}
// processLoop runs periodically to process AS paths into peerings
func (h *PeeringHandler) processLoop() {
ticker := time.NewTicker(peeringProcessInterval)
defer ticker.Stop()
for {
select {
case <-ticker.C:
h.processPeerings()
case <-h.stopCh:
// Final processing
h.processPeerings()
return
}
}
seen.Add(timestamp)
h.asPaths[key] = seen
}
// ProcessPeeringsNow triggers immediate processing of all accumulated AS
@@ -141,6 +140,52 @@ func (h *PeeringHandler) ProcessPeeringsNow() {
h.processPeerings()
}
// Stop stops the processing goroutine, which writes the peerings of the AS
// paths the handler holds before it returns, and waits for it until ctx ends.
// Once ctx has ended the processing goroutine skips that last write, and a
// write still running is lost when the process exits, so Stop logs how many
// paths' peerings are not written.
func (h *PeeringHandler) Stop(ctx context.Context) {
h.stopCh <- ctx
select {
case <-h.done:
case <-ctx.Done():
}
h.mu.Lock()
unwritten := h.writing + len(h.asPaths)
h.mu.Unlock()
if unwritten > 0 {
h.logger.Error("Stop deadline passed before the peering handler wrote "+
"its peerings", "paths", unwritten)
}
}
// processLoop runs periodically to process AS paths into peerings, and once
// more on Stop.
func (h *PeeringHandler) processLoop() {
defer close(h.done)
ticker := time.NewTicker(peeringProcessInterval)
defer ticker.Stop()
for {
select {
case <-ticker.C:
h.processPeerings()
case ctx := <-h.stopCh:
// Final processing, unless the stop deadline has passed
if ctx.Err() == nil {
h.processPeerings()
}
return
}
}
}
// processPeerings extracts peerings from AS paths and writes to database
func (h *PeeringHandler) processPeerings() {
// Take the accumulated paths and replace the map with a fresh empty one
@@ -148,7 +193,8 @@ func (h *PeeringHandler) processPeerings() {
// released, so the map never grows past a single interval's traffic.
h.mu.Lock()
paths := h.asPaths
h.asPaths = make(map[string]time.Time)
h.asPaths = make(map[string]database.SeenTimes)
h.writing = len(paths)
dropped := h.droppedPaths
h.mu.Unlock()
@@ -160,11 +206,18 @@ func (h *PeeringHandler) processPeerings() {
type peeringKey struct {
low, high int
}
peerings := make(map[peeringKey]time.Time)
for pathJSON, timestamp := range paths {
// The earliest and the latest time an update carried each peering. Several
// paths, each carried at several times, can hold a peering, so a new
// peering is stored first seen at the earliest of them and last seen at the
// latest.
peerings := make(map[peeringKey]database.SeenTimes)
for pathJSON, pathSeen := range paths {
var path []int
if err := json.Unmarshal([]byte(pathJSON), &path); err != nil {
err := json.Unmarshal([]byte(pathJSON), &path)
if err != nil {
h.logger.Error("Failed to decode AS path", "error", err)
continue
@@ -187,18 +240,20 @@ func (h *PeeringHandler) processPeerings() {
}
key := peeringKey{low: low, high: high}
// Update timestamp if this is newer
if existing, ok := peerings[key]; !ok || timestamp.After(existing) {
peerings[key] = timestamp
}
seen := peerings[key]
seen.Add(pathSeen.First)
seen.Add(pathSeen.Last)
peerings[key] = seen
}
}
// Record peerings in database
start := time.Now()
successCount := 0
for key, ts := range peerings {
err := h.db.RecordPeering(key.low, key.high, ts)
for key, seen := range peerings {
err := h.db.RecordPeering(key.low, key.high, seen)
if err != nil {
h.logger.Error("Failed to record peering",
"as_a", key.low,
@@ -210,6 +265,10 @@ func (h *PeeringHandler) processPeerings() {
}
}
h.mu.Lock()
h.writing = 0
h.mu.Unlock()
h.logger.Info("Processed AS peerings",
"paths", len(paths),
"unique_peerings", len(peerings),
@@ -218,13 +277,3 @@ func (h *PeeringHandler) processPeerings() {
"duration", time.Since(start),
)
}
// Stop gracefully shuts down the handler by signaling the background
// goroutine to stop and performing a final synchronous processing of
// any remaining AS paths. This ensures no peering data is lost during
// shutdown.
func (h *PeeringHandler) Stop() {
close(h.stopCh)
// Process any remaining peerings synchronously
h.processPeerings()
}
internal/routewatch/peeringhandler_internal_test.go
+307
View File
@@ -0,0 +1,307 @@
package routewatch
import (
"encoding/json"
"fmt"
"maps"
"slices"
"strconv"
"sync"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
)
const (
testASNA = 64500
testASNB = 64501
testASNC = 64502
testASND = 64503
)
// recordingStore wraps mockStore to count every RecordPeering call, so a
// test can tell how many times a peering was written across separate runs.
type recordingStore struct {
*mockStore
mu sync.Mutex
calls int
}
func (r *recordingStore) RecordPeering(
asA, asB int, seen database.SeenTimes,
) error {
r.mu.Lock()
r.calls++
r.mu.Unlock()
return r.mockStore.RecordPeering(asA, asB, seen)
}
func (r *recordingStore) callCount() int {
r.mu.Lock()
defer r.mu.Unlock()
return r.calls
}
// newTestHandler builds a PeeringHandler without starting the periodic
// processing goroutine, so tests drive processing explicitly.
func newTestHandler(db database.Store) *PeeringHandler {
return &PeeringHandler{
db: db,
logger: logger.New(),
asPaths: make(map[string]database.SeenTimes),
}
}
// snapshot returns how many paths the handler tracks and how many it dropped.
func snapshot(h *PeeringHandler) (int, int) {
h.mu.Lock()
defer h.mu.Unlock()
return len(h.asPaths), h.droppedPaths
}
func pathKey(t *testing.T, asns ...int) string {
t.Helper()
b, err := json.Marshal(asns)
if err != nil {
t.Fatalf("failed to marshal path: %v", err)
}
return string(b)
}
func handle(h *PeeringHandler, ts time.Time, asns ...int) {
h.HandleMessage(&ristypes.RISMessage{
Path: ristypes.ASPath{Sequence: asns},
ParsedTimestamp: ts,
})
}
// decodePath decodes an AS path from the JSON the feed writes for it, in which
// an AS set is an array inside the path: decodePath(t, 1, 2, []int{3, 4})
// decodes [1, 2, [3, 4]].
func decodePath(t *testing.T, elements ...any) ristypes.ASPath {
t.Helper()
data, err := json.Marshal(elements)
if err != nil {
t.Fatalf("failed to marshal path: %v", err)
}
var path ristypes.ASPath
err = json.Unmarshal(data, &path)
if err != nil {
t.Fatalf("failed to decode path %s: %v", data, err)
}
return path
}
// TestPeeringHandlerLeavesATrailingASSetOutOfThePeerings handles AS paths that
// end in an AS set. The members of a set are not neighbours of each other or of
// the AS before the set, so a set of several members is in no peering; a set of
// one member counts as that AS.
func TestPeeringHandlerLeavesATrailingASSetOutOfThePeerings(t *testing.T) {
t.Parallel()
tests := []struct {
name string
path []any
// The peerings recorded, keyed as the mock store keys them.
wantPeerings []string
}{
{
name: "a set of several members",
path: []any{testASNA, testASNB, []int{testASNC, testASND}},
wantPeerings: []string{fmt.Sprintf("%d_%d", testASNA, testASNB)},
},
{
name: "a set of one member",
path: []any{testASNA, testASNB, []int{testASNC}},
wantPeerings: []string{
fmt.Sprintf("%d_%d", testASNA, testASNB),
fmt.Sprintf("%d_%d", testASNB, testASNC),
},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
t.Parallel()
store := newMockStore()
h := newTestHandler(store)
h.HandleMessage(&ristypes.RISMessage{
Path: decodePath(t, tt.path...),
ParsedTimestamp: time.Now().UTC(),
})
h.ProcessPeeringsNow()
got := slices.Sorted(maps.Keys(store.Peerings))
if !slices.Equal(got, tt.wantPeerings) {
t.Errorf("peerings recorded = %v, want %v", got, tt.wantPeerings)
}
})
}
}
// TestPeeringHandlerProcessesEachRunAndEmpties verifies that a processing run
// empties the path map (the swap) and that a path seen again after a run is
// recorded in the next run too.
func TestPeeringHandlerProcessesEachRunAndEmpties(t *testing.T) {
t.Parallel()
store := &recordingStore{mockStore: newMockStore()}
h := newTestHandler(store)
now := time.Now().UTC()
// Run 1: one path, one peering recorded, map emptied afterwards.
handle(h, now, testASNA, testASNB)
h.ProcessPeeringsNow()
if tracked, _ := snapshot(h); tracked != 0 {
t.Fatalf("map not empty after first run: %d paths remain", tracked)
}
if got := store.callCount(); got != 1 {
t.Fatalf("want 1 RecordPeering call after first run, got %d", got)
}
// Run 2: the same path again is recorded again (RecordPeering upserts).
handle(h, now.Add(time.Second), testASNA, testASNB)
h.ProcessPeeringsNow()
if tracked, _ := snapshot(h); tracked != 0 {
t.Fatalf("map not empty after second run: %d paths remain", tracked)
}
if got := store.callCount(); got != 2 {
t.Fatalf("want 2 RecordPeering calls after second run, got %d", got)
}
}
// TestPeeringHandlerStopWritesTheRemainingPaths hands a running handler one
// path and stops it. The peering must be in the database when Stop returns.
func TestPeeringHandlerStopWritesTheRemainingPaths(t *testing.T) {
t.Parallel()
store := &recordingStore{mockStore: newMockStore()}
h := NewPeeringHandler(store, logger.New())
handle(h, time.Now().UTC(), testASNA, testASNB)
h.Stop(t.Context())
if got := store.callCount(); got != 1 {
t.Errorf("want 1 RecordPeering call when Stop returns, got %d", got)
}
}
// TestPeeringHandlerCapDropsAndCounts verifies that a full map drops new paths
// and counts them, while a path already tracked is refreshed rather than
// dropped.
func TestPeeringHandlerCapDropsAndCounts(t *testing.T) {
t.Parallel()
store := &recordingStore{mockStore: newMockStore()}
h := newTestHandler(store)
now := time.Now().UTC()
// Fill the map to exactly maxTrackedPaths, including one real path key so
// the "already tracked" branch can be exercised. The filler keys are never
// processed in this test, so their contents do not matter.
existing := pathKey(t, testASNA, testASNB)
h.mu.Lock()
h.asPaths[existing] = database.SeenTimes{First: now, Last: now}
for i := 0; len(h.asPaths) < maxTrackedPaths; i++ {
h.asPaths[strconv.Itoa(i)] = database.SeenTimes{First: now, Last: now}
}
h.mu.Unlock()
// A new path is dropped and counted because the map is full.
handle(h, now.Add(time.Second), testASNA, testASNC)
tracked, dropped := snapshot(h)
if tracked != maxTrackedPaths {
t.Fatalf("want map size %d after drop, got %d", maxTrackedPaths, tracked)
}
if dropped != 1 {
t.Fatalf("want dropped count 1, got %d", dropped)
}
// A path already tracked is refreshed, not dropped.
refreshed := now.Add(2 * time.Second)
handle(h, refreshed, testASNA, testASNB)
tracked, dropped = snapshot(h)
if tracked != maxTrackedPaths {
t.Fatalf("want map size %d after refresh, got %d", maxTrackedPaths, tracked)
}
if dropped != 1 {
t.Fatalf("want dropped count still 1 after refresh, got %d", dropped)
}
h.mu.Lock()
gotSeen := h.asPaths[existing]
h.mu.Unlock()
if !gotSeen.Last.Equal(refreshed) {
t.Fatalf("existing path last seen not refreshed: want %v, got %v",
refreshed, gotSeen.Last)
}
}
// TestPeeringHandlerStoresTheEarliestAndLatestTimesOfANewPeering handles two
// updates whose path carries a peering the database does not hold yet, the
// second 5 seconds after the first, in one run into a real database. The
// peering is stored first seen at the time of the first update and last seen
// at the time of the second.
func TestPeeringHandlerStoresTheEarliestAndLatestTimesOfANewPeering(
t *testing.T,
) {
t.Parallel()
db := newTestDatabase(t)
h := newTestHandler(db)
first := peerAnnouncement(testPeerIP, testV4Prefix)
second := peerAnnouncement(testPeerIP, testV4Prefix)
second.ParsedTimestamp = first.ParsedTimestamp.Add(secondUpdateAfter)
h.HandleMessage(first)
h.HandleMessage(second)
h.ProcessPeeringsNow()
peers, err := db.GetASPeers(testASNA)
if err != nil {
t.Fatalf("GetASPeers(%d): %v", testASNA, err)
}
if len(peers) != 1 || peers[0].ASN != testASNB {
t.Fatalf("peers of AS%d = %+v, want AS%d alone", testASNA, peers,
testASNB)
}
if !peers[0].FirstSeen.Equal(first.ParsedTimestamp) {
t.Errorf("first_seen of peering AS%d-AS%d = %s, want %s", testASNA,
testASNB, peers[0].FirstSeen, first.ParsedTimestamp)
}
if !peers[0].LastSeen.Equal(second.ParsedTimestamp) {
t.Errorf("last_seen of peering AS%d-AS%d = %s, want %s", testASNA,
testASNB, peers[0].LastSeen, second.ParsedTimestamp)
}
}
internal/routewatch/peeringhandler_test.go
-163
View File
@@ -1,163 +0,0 @@
package routewatch
import (
"encoding/json"
"strconv"
"sync"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
)
const (
testASNA = 64500
testASNB = 64501
testASNC = 64502
)
// recordingStore wraps mockStore to count every RecordPeering call, so a
// test can tell how many times a peering was written across separate runs.
type recordingStore struct {
*mockStore
mu sync.Mutex
calls int
}
func (r *recordingStore) RecordPeering(asA, asB int, ts time.Time) error {
r.mu.Lock()
r.calls++
r.mu.Unlock()
return r.mockStore.RecordPeering(asA, asB, ts)
}
func (r *recordingStore) callCount() int {
r.mu.Lock()
defer r.mu.Unlock()
return r.calls
}
// newTestHandler builds a PeeringHandler without starting the periodic
// processing goroutine, so tests drive processing explicitly.
func newTestHandler(db database.Store) *PeeringHandler {
return &PeeringHandler{
db: db,
logger: logger.New(),
asPaths: make(map[string]time.Time),
stopCh: make(chan struct{}),
}
}
func snapshot(h *PeeringHandler) (tracked, dropped int) {
h.mu.Lock()
defer h.mu.Unlock()
return len(h.asPaths), h.droppedPaths
}
func pathKey(t *testing.T, asns ...int) string {
t.Helper()
b, err := json.Marshal(ristypes.ASPath(asns))
if err != nil {
t.Fatalf("failed to marshal path: %v", err)
}
return string(b)
}
func handle(h *PeeringHandler, ts time.Time, asns ...int) {
h.HandleMessage(&ristypes.RISMessage{
Path: ristypes.ASPath(asns),
ParsedTimestamp: ts,
})
}
// TestPeeringHandlerProcessesEachRunAndEmpties verifies that a processing run
// empties the path map (the swap) and that a path seen again after a run is
// recorded in the next run too.
func TestPeeringHandlerProcessesEachRunAndEmpties(t *testing.T) {
store := &recordingStore{mockStore: newMockStore()}
h := newTestHandler(store)
now := time.Now().UTC()
// Run 1: one path, one peering recorded, map emptied afterwards.
handle(h, now, testASNA, testASNB)
h.ProcessPeeringsNow()
if tracked, _ := snapshot(h); tracked != 0 {
t.Fatalf("map not empty after first run: %d paths remain", tracked)
}
if got := store.callCount(); got != 1 {
t.Fatalf("want 1 RecordPeering call after first run, got %d", got)
}
// Run 2: the same path again is recorded again (RecordPeering upserts).
handle(h, now.Add(time.Second), testASNA, testASNB)
h.ProcessPeeringsNow()
if tracked, _ := snapshot(h); tracked != 0 {
t.Fatalf("map not empty after second run: %d paths remain", tracked)
}
if got := store.callCount(); got != 2 {
t.Fatalf("want 2 RecordPeering calls after second run, got %d", got)
}
}
// TestPeeringHandlerCapDropsAndCounts verifies that a full map drops new paths
// and counts them, while a path already tracked is refreshed rather than
// dropped.
func TestPeeringHandlerCapDropsAndCounts(t *testing.T) {
store := &recordingStore{mockStore: newMockStore()}
h := newTestHandler(store)
now := time.Now().UTC()
// Fill the map to exactly maxTrackedPaths, including one real path key so
// the "already tracked" branch can be exercised. The filler keys are never
// processed in this test, so their contents do not matter.
existing := pathKey(t, testASNA, testASNB)
h.mu.Lock()
h.asPaths[existing] = now
for i := 0; len(h.asPaths) < maxTrackedPaths; i++ {
h.asPaths[strconv.Itoa(i)] = now
}
h.mu.Unlock()
// A new path is dropped and counted because the map is full.
handle(h, now.Add(time.Second), testASNA, testASNC)
tracked, dropped := snapshot(h)
if tracked != maxTrackedPaths {
t.Fatalf("want map size %d after drop, got %d", maxTrackedPaths, tracked)
}
if dropped != 1 {
t.Fatalf("want dropped count 1, got %d", dropped)
}
// A path already tracked is refreshed, not dropped.
refreshed := now.Add(2 * time.Second)
handle(h, refreshed, testASNA, testASNB)
tracked, dropped = snapshot(h)
if tracked != maxTrackedPaths {
t.Fatalf("want map size %d after refresh, got %d", maxTrackedPaths, tracked)
}
if dropped != 1 {
t.Fatalf("want dropped count still 1 after refresh, got %d", dropped)
}
h.mu.Lock()
gotTS := h.asPaths[existing]
h.mu.Unlock()
if !gotTS.Equal(refreshed) {
t.Fatalf("existing path timestamp not refreshed: want %v, got %v", refreshed, gotTS)
}
}
internal/routewatch/prefixhandler.go
+410 -291
View File
@@ -1,7 +1,10 @@
package routewatch
import (
"context"
"net"
"net/netip"
"slices"
"strings"
"sync"
"time"
@@ -10,19 +13,27 @@ import (
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/metrics"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
"github.com/google/uuid"
)
const (
// prefixHandlerQueueSize is the queue capacity for prefix tracking
// operations, about 4 seconds of feed at peak. The streamer drops rather
// than blocks when a queue is full, so this bounds memory. Batches still
// flush on a timer (prefixBatchTimeout).
// operations, about 4 seconds of feed at peak. The streamer drops updates
// rather than blocks when a queue is full (a STATE message is held until
// there is room), so this bounds memory. Batches still flush on a timer
// (prefixBatchTimeout).
prefixHandlerQueueSize = 20000
// prefixBatchSize is the number of prefix updates to batch together
prefixBatchSize = 25000
// prefixBatchMaxSize is the most prefix updates the batch holds while the
// flush goroutine writes the previous one: ten batches, about 60 MiB
// (README "Memory"). At this size HandleMessage waits until the flush
// goroutine takes the batch. On top of these, the handler holds at most
// prefixBatchSize updates, about 6 MiB, that a flush kept for the peers
// whose route delete failed (held).
prefixBatchMaxSize = 250000
// prefixBatchTimeout is the maximum time to wait before flushing a batch
// DO NOT reduce this timeout - larger batches are more efficient
prefixBatchTimeout = 1 * time.Second
@@ -32,19 +43,46 @@ const (
ipv6Version = 6
)
// PrefixHandler tracks BGP prefixes and maintains a live routing table in the database.
// Routes are added on announcement and deleted on withdrawal.
// PrefixHandler tracks BGP prefixes and maintains a live routing table in the
// database. Routes are added on announcement and deleted on withdrawal or on an
// announcement whose AS path is empty or ends in AS 0, and all routes of a peer
// are deleted when its BGP session goes down.
type PrefixHandler struct {
db database.Store
logger *logger.Logger
metrics *metrics.Tracker
// Batching
mu sync.Mutex
batch []prefixUpdate
lastFlush time.Time
stopCh chan struct{}
wg sync.WaitGroup
// Batching. mu guards batch, held, peersDown, writing and deleting only:
// the flush goroutine takes them under mu and writes them after releasing
// mu, so HandleMessage waits for a database write only while batch is at
// prefixBatchMaxSize.
mu sync.Mutex
batch []prefixUpdate
// held holds the updates that the peers whose route delete failed in the
// last flush sent after their session went down, for the next flush to
// write after it retries the delete. They are not in batch, so they do
// not count toward a full batch: a delete that keeps failing is retried
// at the pace of the flush timer.
held []prefixUpdate
// peersDown holds, each once, the peers whose BGP session went down
// since the batch was taken, and those whose route delete failed in the
// last flush.
peersDown []string
// writing is how many updates the flush goroutine took and has not
// finished writing.
writing int
// deleting is how many peers from peersDown the flush goroutine took and
// has not yet deleted the routes of.
deleting int
// batchTaken is signalled when the flush goroutine takes batch; its L is
// &mu.
batchTaken sync.Cond
batchFullCh chan struct{} // wakes the flush goroutine when batch is full
lastFlush time.Time // used only by the flush goroutine
// stopCh hands Stop's context to the flush goroutine. It holds one, so
// Stop does not wait for a write in progress to hand it over.
stopCh chan context.Context
done chan struct{} // closed when the flush goroutine returns
}
type prefixUpdate struct {
@@ -54,20 +92,25 @@ type prefixUpdate struct {
messageType string
timestamp time.Time
path []int
// endsInSet says the path ends in an AS set of several members, so it
// gives no origin.
endsInSet bool
}
// NewPrefixHandler creates a new batched prefix tracking handler
func NewPrefixHandler(db database.Store, logger *logger.Logger) *PrefixHandler {
h := &PrefixHandler{
db: db,
logger: logger,
batch: make([]prefixUpdate, 0, prefixBatchSize),
lastFlush: time.Now(),
stopCh: make(chan struct{}),
db: db,
logger: logger,
batch: make([]prefixUpdate, 0, prefixBatchSize),
batchFullCh: make(chan struct{}, 1),
lastFlush: time.Now(),
stopCh: make(chan context.Context, 1),
done: make(chan struct{}),
}
h.batchTaken.L = &h.mu
// Start the flush timer goroutine
h.wg.Add(1)
go h.flushLoop()
return h
@@ -78,10 +121,13 @@ func (h *PrefixHandler) SetMetricsTracker(metrics *metrics.Tracker) {
h.metrics = metrics
}
// WantsMessage returns true if this handler wants to process messages of the given type
// WantsMessage returns true if this handler wants to process messages of the
// given type
func (h *PrefixHandler) WantsMessage(messageType string) bool {
// We only care about UPDATE messages for the routing table
return messageType == "UPDATE"
// UPDATE messages change routes; a STATE message can say a peer's
// session went down, which removes all of its routes.
return messageType == ristypes.MessageTypeUpdate ||
messageType == ristypes.MessageTypeState
}
// QueueCapacity returns the desired queue capacity for this handler
@@ -92,28 +138,61 @@ func (h *PrefixHandler) QueueCapacity() int {
// HandleMessage processes a message to track prefix information
func (h *PrefixHandler) HandleMessage(msg *ristypes.RISMessage) {
if msg.Type == ristypes.MessageTypeState {
if msg.State == ristypes.StateDown {
h.peerSessionDown(msg.Peer)
}
return
}
// Use the pre-parsed timestamp
timestamp := msg.ParsedTimestamp
// Get origin ASN from path (last element)
var originASN int
if len(msg.Path) > 0 {
originASN = msg.Path[len(msg.Path)-1]
}
originASN := msg.Path.Origin()
// The stored and shown AS path lists the members of an AS set that ends
// the path after the rest of it.
path := slices.Concat(msg.Path.Sequence, msg.Path.Set)
h.mu.Lock()
defer h.mu.Unlock()
// Process announcements
// While the batch is at its maximum size, wait until the flush goroutine
// takes it. Meanwhile the handler takes nothing off its queue, which
// fills, and the streamer drops messages for it.
for len(h.batch) >= prefixBatchMaxSize {
h.batchTaken.Wait()
}
// Withdrawals are queued before announcements, because the batch keeps
// each peer's last update for a prefix: a message that lists a prefix as
// both withdrawn and announced keeps the route (RFC 4271 section 4.3).
// A withdrawal names only the prefix and peer: the message's AS path, and
// so its origin AS, belongs to the announced prefixes.
for _, prefix := range msg.Withdrawals {
h.batch = append(h.batch, prefixUpdate{
prefix: canonicalPrefix(prefix),
peer: msg.Peer,
messageType: "withdrawal",
timestamp: timestamp,
})
// Record withdrawal in metrics
if h.metrics != nil {
h.metrics.RecordWithdrawal()
}
}
for _, announcement := range msg.Announcements {
for _, prefix := range announcement.Prefixes {
h.batch = append(h.batch, prefixUpdate{
prefix: prefix,
prefix: canonicalPrefix(prefix),
originASN: originASN,
peer: msg.Peer,
messageType: "announcement",
timestamp: timestamp,
path: msg.Path,
path: path,
endsInSet: len(msg.Path.Set) > 0,
})
// Record announcement in metrics
if h.metrics != nil {
@@ -122,96 +201,195 @@ func (h *PrefixHandler) HandleMessage(msg *ristypes.RISMessage) {
}
}
// Process withdrawals
for _, prefix := range msg.Withdrawals {
h.batch = append(h.batch, prefixUpdate{
prefix: prefix,
originASN: originASN, // Use the originASN from path if available
peer: msg.Peer,
messageType: "withdrawal",
timestamp: timestamp,
path: msg.Path,
})
// Record withdrawal in metrics
if h.metrics != nil {
h.metrics.RecordWithdrawal()
}
}
// Check if we need to flush
// A full batch wakes the flush goroutine. If it is writing, the wake-up
// waits in batchFullCh and the batch keeps growing until the write ends
// or the batch reaches prefixBatchMaxSize.
if len(h.batch) >= prefixBatchSize {
h.flushBatchLocked()
select {
case h.batchFullCh <- struct{}{}:
default: // already woken
}
}
}
// flushLoop runs in a goroutine and periodically flushes batches
// Stop stops the flush goroutine, which writes the updates and peer downs the
// handler holds before it returns, and waits for it until ctx ends. Once ctx
// has ended the flush goroutine skips that last write, and a write still
// running is lost when the process exits. A peer whose route delete fails in
// the last write is not retried either. Stop logs how many updates are not
// written and how many peer downs have not had their routes deleted.
func (h *PrefixHandler) Stop(ctx context.Context) {
h.stopCh <- ctx
select {
case <-h.done:
case <-ctx.Done():
}
h.mu.Lock()
unwritten := h.writing + len(h.batch) + len(h.held)
peersDown := h.deleting + len(h.peersDown)
h.mu.Unlock()
if unwritten == 0 && peersDown == 0 {
return
}
if ctx.Err() != nil {
h.logger.Error("Stop deadline passed before the prefix handler wrote "+
"its updates", "updates", unwritten, "peers_down", peersDown)
return
}
h.logger.Error("The prefix handler stopped without deleting the routes "+
"of peers whose session went down", "updates", unwritten,
"peers_down", peersDown)
}
// peerSessionDown notes that the peer's BGP session went down, so that the next
// flush deletes all of the peer's live routes before it writes the batch. The
// peer's updates still in the batch or held came before the down, which
// replaces them, so they are dropped; updates the peer sends later stay in the
// batch and are written after the delete. A peer already in peersDown is not
// added again: one delete removes all of its routes.
func (h *PrefixHandler) peerSessionDown(peer string) {
h.mu.Lock()
defer h.mu.Unlock()
isPeers := func(update prefixUpdate) bool { return update.peer == peer }
h.batch = slices.DeleteFunc(h.batch, isPeers)
h.held = slices.DeleteFunc(h.held, isPeers)
if !slices.Contains(h.peersDown, peer) {
h.peersDown = append(h.peersDown, peer)
}
}
// flushLoop runs in a goroutine, the only one that writes the batch: it
// flushes when the batch is full, periodically, and once more on Stop.
func (h *PrefixHandler) flushLoop() {
defer h.wg.Done()
defer close(h.done)
ticker := time.NewTicker(prefixBatchTimeout)
defer ticker.Stop()
for {
select {
case <-ticker.C:
h.mu.Lock()
if time.Since(h.lastFlush) >= prefixBatchTimeout {
h.flushBatchLocked()
h.flushBatch()
}
case <-h.batchFullCh:
h.flushBatch()
case ctx := <-h.stopCh:
// Final flush, unless the stop deadline has passed
if ctx.Err() == nil {
h.flushBatch()
}
h.mu.Unlock()
case <-h.stopCh:
// Final flush
h.mu.Lock()
h.flushBatchLocked()
h.mu.Unlock()
return
}
}
}
// flushBatchLocked flushes the prefix batch to the database (must be called with mutex held)
func (h *PrefixHandler) flushBatchLocked() {
if len(h.batch) == 0 {
// takeBatch returns the pending updates, the held ones first, and the peers in
// peersDown, and leaves an empty batch, no held updates and no peers in their
// place for HandleMessage to fill. The updates count in writing until
// flushBatch has written them, and the peers in deleting until their routes
// are deleted.
func (h *PrefixHandler) takeBatch() ([]prefixUpdate, []string) {
h.mu.Lock()
defer h.mu.Unlock()
if len(h.held) == 0 && len(h.batch) == 0 && len(h.peersDown) == 0 {
return nil, nil
}
batch, peersDown := h.batch, h.peersDown
if len(h.held) > 0 {
batch = slices.Concat(h.held, h.batch)
}
h.writing = len(batch)
h.deleting = len(peersDown)
h.held = nil
h.batch = make([]prefixUpdate, 0, prefixBatchSize)
h.peersDown = nil
h.batchTaken.Broadcast()
return batch, peersDown
}
// flushBatch takes the pending batch and writes it to the database, without
// holding the mutex during the write.
func (h *PrefixHandler) flushBatch() {
batch, peersDown := h.takeBatch()
if len(batch) == 0 && len(peersDown) == 0 {
return
}
startTime := time.Now()
batchSize := len(h.batch)
batchSize := len(batch)
// Group updates by prefix to deduplicate
// For each prefix, keep the latest update
prefixMap := make(map[string]prefixUpdate)
for _, update := range h.batch {
key := update.prefix
if existing, ok := prefixMap[key]; !ok || update.timestamp.After(existing.timestamp) {
prefixMap[key] = update
// The routes of a peer whose session went down are deleted first: the
// batch holds only the updates the peer sent after the down. A peer whose
// delete failed is kept, with those updates, for the next flush to retry.
failedPeers := h.deleteLiveRoutesOfPeers(peersDown)
if len(failedPeers) > 0 {
batch = h.keepForNextFlush(batch, failedPeers)
}
// For each prefix and peer, keep only the last update the feed delivered.
// Updates from different peers for the same prefix are all kept. Arrival
// order decides, not the timestamp: timestamps are whole seconds, so
// updates within one second carry the same one.
type prefixAndPeer struct{ prefix, peer string }
lastUpdates := make(map[prefixAndPeer]prefixUpdate)
// Each prefix the batch names, announced or withdrawn, with the earliest
// timestamp among all of its updates, not only each peer's last one: the
// prefix tables store only first_seen, written when the prefix is new and
// never changed after.
prefixesToUpdate := make(map[string]time.Time)
for _, update := range batch {
lastUpdates[prefixAndPeer{update.prefix, update.peer}] = update
seen, exists := prefixesToUpdate[update.prefix]
if !exists || update.timestamp.Before(seen) {
prefixesToUpdate[update.prefix] = update.timestamp
}
}
// Collect routes to upsert and delete
var routesToUpsert []*database.LiveRoute
var routesToDelete []database.LiveRouteDeletion
var (
routesToUpsert []*database.LiveRoute
routesToDelete []database.LiveRouteDeletion
)
// Collect unique prefixes to update
prefixesToUpdate := make(map[string]time.Time)
for _, update := range prefixMap {
// Track prefix for both announcements and withdrawals
if _, exists := prefixesToUpdate[update.prefix]; !exists || update.timestamp.After(prefixesToUpdate[update.prefix]) {
prefixesToUpdate[update.prefix] = update.timestamp
}
if update.messageType == "announcement" && update.originASN > 0 {
for _, update := range lastUpdates {
// An announcement is stored when its path gives an origin, or when it
// ends in an AS set of several members: that route is stored with
// origin 0 (none). Any other update deletes the peer's route for the
// prefix: a withdrawal, or an announcement whose path is empty or ends
// in AS 0. Such an announcement replaces the peer's route, as any
// announcement does, with none (RFC 7607 treats AS 0 in a path as a
// withdrawal).
if update.messageType == "announcement" &&
(update.originASN > 0 || update.endsInSet) {
// Create live route for batch upsert
route := h.createLiveRoute(update)
if route != nil {
routesToUpsert = append(routesToUpsert, route)
}
} else if update.messageType == "withdrawal" {
} else {
// Parse CIDR to get IP version
_, ipVersion, err := parseCIDR(update.prefix)
if err != nil {
h.logger.Error("Failed to parse CIDR for withdrawal", "prefix", update.prefix, "error", err)
h.logger.Error("Failed to parse CIDR for withdrawal",
"prefix", update.prefix, "error", err)
continue
}
@@ -219,26 +397,148 @@ func (h *PrefixHandler) flushBatchLocked() {
// Create deletion record for batch delete
routesToDelete = append(routesToDelete, database.LiveRouteDeletion{
Prefix: update.prefix,
OriginASN: update.originASN,
PeerIP: update.peer,
IPVersion: ipVersion,
})
}
}
// Process batch operations
successCount := h.writeBatch(routesToUpsert, routesToDelete, prefixesToUpdate)
h.mu.Lock()
h.writing = 0
h.mu.Unlock()
elapsed := time.Since(startTime)
h.logger.Debug("Flushed prefix batch",
"batch_size", batchSize,
"peers_down", len(peersDown),
"unique_prefixes", len(prefixesToUpdate),
"success", successCount,
"duration_ms", elapsed.Milliseconds(),
)
h.lastFlush = time.Now()
}
// deleteLiveRoutesOfPeers deletes all live routes of each of peers, the peers
// whose BGP session went down, and returns those whose delete failed. Most
// downs come from peers that have no routes, so only a delete that removed
// routes is logged. A peer whose delete succeeded no longer counts in deleting;
// a failed one counts until keepForNextFlush puts it back in peersDown.
func (h *PrefixHandler) deleteLiveRoutesOfPeers(peers []string) []string {
var failed []string
for _, peer := range peers {
deleted, err := h.db.DeleteLiveRoutesOfPeer(peer)
if err != nil {
h.logger.Error(
"Failed to delete the live routes of a peer whose session "+
"went down; the next flush retries",
"peer", peer, "routes_deleted", deleted, "error", err)
failed = append(failed, peer)
continue
}
h.mu.Lock()
h.deleting--
h.mu.Unlock()
if deleted > 0 {
h.logger.Info(
"Deleted the live routes of a peer whose session went down",
"peer", peer, "routes", deleted)
}
}
return failed
}
// keepForNextFlush puts the peers whose route delete failed back in peersDown,
// and their updates from batch in held, so that the next flush retries each
// delete before it writes the peer's updates; written now, they would be
// deleted by the retry. The updates of a peer whose session went down again
// since the batch was taken came before that down, so they are dropped, as
// peerSessionDown drops them, and the peer is already in peersDown. It returns
// the updates of the other peers, for this flush to write.
//
// While a delete keeps failing, the held updates would grow with every flush,
// so only the last prefixBatchSize of them are kept and the older ones are
// dropped. This never leaves in place a route the peer withdrew: a prefix with
// a kept update keeps its last one, and the retry deletes all of the peer's
// routes, so a prefix whose updates were all dropped has no route from the peer
// until the peer announces it again.
func (h *PrefixHandler) keepForNextFlush(
batch []prefixUpdate, failedPeers []string,
) []prefixUpdate {
var kept, toWrite []prefixUpdate
for _, update := range batch {
if slices.Contains(failedPeers, update.peer) {
kept = append(kept, update)
} else {
toWrite = append(toWrite, update)
}
}
h.mu.Lock()
defer h.mu.Unlock()
kept = slices.DeleteFunc(kept, func(update prefixUpdate) bool {
return slices.Contains(h.peersDown, update.peer)
})
if len(kept) > prefixBatchSize {
h.logger.Error(
"Dropped the oldest updates of peers whose route delete failed",
"peers", failedPeers, "dropped", len(kept)-prefixBatchSize)
// A copy, so that the dropped updates are freed.
kept = slices.Clone(kept[len(kept)-prefixBatchSize:])
}
// takeBatch left held empty, and only this flush fills it. The kept
// updates now count in held, not in writing, and the failed peers in
// peersDown, not in deleting.
h.held = kept
h.writing = len(toWrite)
h.deleting = 0
for _, peer := range failedPeers {
if !slices.Contains(h.peersDown, peer) {
h.peersDown = append(h.peersDown, peer)
}
}
return toWrite
}
// writeBatch writes the collected route upserts, route deletions and prefix
// updates to the database and returns how many routes it wrote.
func (h *PrefixHandler) writeBatch(
routesToUpsert []*database.LiveRoute,
routesToDelete []database.LiveRouteDeletion,
prefixesToUpdate map[string]time.Time,
) int {
successCount := 0
if len(routesToUpsert) > 0 {
if err := h.db.UpsertLiveRouteBatch(routesToUpsert); err != nil {
h.logger.Error("Failed to upsert route batch", "error", err, "count", len(routesToUpsert))
err := h.db.UpsertLiveRouteBatch(routesToUpsert)
if err != nil {
h.logger.Error("Failed to upsert route batch",
"error", err, "count", len(routesToUpsert))
} else {
successCount += len(routesToUpsert)
}
}
if len(routesToDelete) > 0 {
if err := h.db.DeleteLiveRouteBatch(routesToDelete); err != nil {
h.logger.Error("Failed to delete route batch", "error", err, "count", len(routesToDelete))
err := h.db.DeleteLiveRouteBatch(routesToDelete)
if err != nil {
h.logger.Error("Failed to delete route batch",
"error", err, "count", len(routesToDelete))
} else {
successCount += len(routesToDelete)
}
@@ -246,26 +546,33 @@ func (h *PrefixHandler) flushBatchLocked() {
// Update prefix tables
if len(prefixesToUpdate) > 0 {
if err := h.db.UpdatePrefixesBatch(prefixesToUpdate); err != nil {
h.logger.Error("Failed to update prefix batch", "error", err, "count", len(prefixesToUpdate))
err := h.db.UpdatePrefixesBatch(prefixesToUpdate)
if err != nil {
h.logger.Error("Failed to update prefix batch",
"error", err, "count", len(prefixesToUpdate))
}
}
elapsed := time.Since(startTime)
h.logger.Debug("Flushed prefix batch",
"batch_size", batchSize,
"unique_prefixes", len(prefixMap),
"success", successCount,
"duration_ms", elapsed.Milliseconds(),
)
// Clear batch
h.batch = h.batch[:0]
h.lastFlush = time.Now()
return successCount
}
// parseCIDR extracts the mask length and IP version from a prefix string
func parseCIDR(prefix string) (maskLength int, ipVersion int, err error) {
// canonicalPrefix returns prefix in the text form net/netip prints, the form
// the IP lookup builds when it looks a prefix up. The feed sends IPv6
// announcements compressed ("2001:db8::/32") but IPv6 withdrawals uncompressed
// ("2001:db8:0:0:0:0:0:0/32"); stored as received, a withdrawal would not match
// the route its announcement stored. A prefix that does not parse is returned
// unchanged, and the batch flush reports it.
func canonicalPrefix(prefix string) string {
p, err := netip.ParsePrefix(prefix)
if err != nil {
return prefix
}
return p.Masked().String()
}
// parseCIDR returns the mask length and the IP version of a prefix string
func parseCIDR(prefix string) (int, int, error) {
_, ipNet, err := net.ParseCIDR(prefix)
if err != nil {
return 0, 0, err
@@ -279,64 +586,6 @@ func parseCIDR(prefix string) (maskLength int, ipVersion int, err error) {
return ones, ipv4Version, nil
}
// processAnnouncement handles storing an announcement in the database
// nolint:unused // kept for potential future use
func (h *PrefixHandler) processAnnouncement(_ *database.Prefix, update prefixUpdate) {
// Parse CIDR to get mask length
maskLength, ipVersion, err := parseCIDR(update.prefix)
if err != nil {
h.logger.Error("Failed to parse CIDR",
"prefix", update.prefix,
"error", err,
)
return
}
// Track route update metrics
if h.metrics != nil {
if ipVersion == ipv4Version {
h.metrics.RecordIPv4Update()
} else {
h.metrics.RecordIPv6Update()
}
}
// Create live route record
liveRoute := &database.LiveRoute{
ID: uuid.New(),
Prefix: update.prefix,
MaskLength: maskLength,
IPVersion: ipVersion,
OriginASN: update.originASN,
PeerIP: update.peer,
ASPath: update.path,
NextHop: update.peer, // Using peer as next hop
LastUpdated: update.timestamp,
}
// For IPv4, calculate the IP range
if ipVersion == ipv4Version {
start, end, err := database.CalculateIPv4Range(update.prefix)
if err == nil {
liveRoute.V4IPStart = &start
liveRoute.V4IPEnd = &end
} else {
h.logger.Error("Failed to calculate IPv4 range",
"prefix", update.prefix,
"error", err,
)
}
}
if err := h.db.UpsertLiveRoute(liveRoute); err != nil {
h.logger.Error("Failed to upsert live route",
"prefix", update.prefix,
"error", err,
)
}
}
// createLiveRoute creates a LiveRoute from a prefix update
func (h *PrefixHandler) createLiveRoute(update prefixUpdate) *database.LiveRoute {
// Parse CIDR to get mask length
@@ -361,7 +610,6 @@ func (h *PrefixHandler) createLiveRoute(update prefixUpdate) *database.LiveRoute
// Create live route record
liveRoute := &database.LiveRoute{
ID: uuid.New(),
Prefix: update.prefix,
MaskLength: maskLength,
IPVersion: ipVersion,
@@ -372,134 +620,5 @@ func (h *PrefixHandler) createLiveRoute(update prefixUpdate) *database.LiveRoute
LastUpdated: update.timestamp,
}
// For IPv4, calculate the IP range
if ipVersion == ipv4Version {
start, end, err := database.CalculateIPv4Range(update.prefix)
if err == nil {
liveRoute.V4IPStart = &start
liveRoute.V4IPEnd = &end
} else {
h.logger.Error("Failed to calculate IPv4 range",
"prefix", update.prefix,
"error", err,
)
}
}
return liveRoute
}
// processAnnouncementDirect handles storing an announcement directly without prefix table
// nolint:unused // kept for potential future use
func (h *PrefixHandler) processAnnouncementDirect(update prefixUpdate) {
// Parse CIDR to get mask length
maskLength, ipVersion, err := parseCIDR(update.prefix)
if err != nil {
h.logger.Error("Failed to parse CIDR",
"prefix", update.prefix,
"error", err,
)
return
}
// Track route update metrics
if h.metrics != nil {
if ipVersion == ipv4Version {
h.metrics.RecordIPv4Update()
} else {
h.metrics.RecordIPv6Update()
}
}
// Create live route record
liveRoute := &database.LiveRoute{
ID: uuid.New(),
Prefix: update.prefix,
MaskLength: maskLength,
IPVersion: ipVersion,
OriginASN: update.originASN,
PeerIP: update.peer,
ASPath: update.path,
NextHop: update.peer, // Using peer as next hop
LastUpdated: update.timestamp,
}
// For IPv4, calculate the IP range
if ipVersion == ipv4Version {
start, end, err := database.CalculateIPv4Range(update.prefix)
if err == nil {
liveRoute.V4IPStart = &start
liveRoute.V4IPEnd = &end
} else {
h.logger.Error("Failed to calculate IPv4 range",
"prefix", update.prefix,
"error", err,
)
}
}
if err := h.db.UpsertLiveRoute(liveRoute); err != nil {
h.logger.Error("Failed to upsert live route",
"prefix", update.prefix,
"error", err,
)
}
}
// processWithdrawalDirect handles removing a route directly without prefix table
// nolint:unused // kept for potential future use
func (h *PrefixHandler) processWithdrawalDirect(update prefixUpdate) {
// For withdrawals, we need to delete the route from live_routes
if update.originASN > 0 {
if err := h.db.DeleteLiveRoute(update.prefix, update.originASN, update.peer); err != nil {
h.logger.Error("Failed to delete live route",
"prefix", update.prefix,
"origin_asn", update.originASN,
"peer", update.peer,
"error", err,
)
}
} else {
// If no origin ASN, just delete all routes for this prefix from this peer
if err := h.db.DeleteLiveRoute(update.prefix, 0, update.peer); err != nil {
h.logger.Error("Failed to delete live route (no origin ASN)",
"prefix", update.prefix,
"peer", update.peer,
"error", err,
)
}
}
}
// processWithdrawal handles removing a route from the live routing table
// nolint:unused // kept for potential future use
func (h *PrefixHandler) processWithdrawal(_ *database.Prefix, update prefixUpdate) {
// For withdrawals, we need to delete the route from live_routes
// Since we have the origin ASN from the update, we can delete the specific route
if update.originASN > 0 {
if err := h.db.DeleteLiveRoute(update.prefix, update.originASN, update.peer); err != nil {
h.logger.Error("Failed to delete live route",
"prefix", update.prefix,
"origin_asn", update.originASN,
"peer", update.peer,
"error", err,
)
}
} else {
// If no origin ASN, just delete all routes for this prefix from this peer
if err := h.db.DeleteLiveRoute(update.prefix, 0, update.peer); err != nil {
h.logger.Error("Failed to delete live route (no origin ASN)",
"prefix", update.prefix,
"peer", update.peer,
"error", err,
)
}
}
}
// Stop gracefully stops the handler and flushes remaining batches
func (h *PrefixHandler) Stop() {
close(h.stopCh)
h.wg.Wait()
}
internal/routewatch/prefixhandler_internal_test.go
+1114
View File
File diff suppressed because it is too large Load Diff
internal/server/handlers.go
+896 -625
View File
File diff suppressed because it is too large Load Diff
internal/server/handlers_internal_test.go
+1199
View File
File diff suppressed because it is too large Load Diff
internal/server/handlers_test.go
-99
View File
@@ -1,99 +0,0 @@
package server
import (
"context"
"net/http"
"net/http/httptest"
"runtime"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/metrics"
"git.eeqj.de/sneak/routewatch/internal/streamer"
)
// blockingStatsDB embeds database.Store (left nil) and overrides only
// GetStatsContext, which blocks until release is closed. The stats handlers
// call it in a goroutine; every other Store method is unused on the timeout
// path and would panic if called.
type blockingStatsDB struct {
database.Store
release chan struct{}
}
func (d blockingStatsDB) GetStatsContext(_ context.Context) (database.Stats, error) {
<-d.release
return database.Stats{}, nil
}
// TestStatsHandlersDoNotLeakOnTimeout drives each stats handler repeatedly with
// a request whose context times out before the database responds, then releases
// the blocked queries and asserts the goroutine count returns to its starting
// value. Before the fix the per-request goroutine sent on an unbuffered channel
// that nothing received once the timeout won, so it blocked forever and every
// poll leaked one goroutine.
func TestStatsHandlersDoNotLeakOnTimeout(t *testing.T) {
release := make(chan struct{})
db := blockingStatsDB{release: release}
s := New(db, streamer.New(logger.New(), metrics.New()), logger.New())
handlers := map[string]http.HandlerFunc{
"status.json": s.handleStatusJSON(),
"stats": s.handleStats(),
}
baseline := settledGoroutineCount()
const (
iterations = 20
requestTimeout = 50 * time.Millisecond
)
for _, handler := range handlers {
for range iterations {
ctx, cancel := context.WithTimeout(context.Background(), requestTimeout)
req := httptest.NewRequest(http.MethodGet, "/", nil).WithContext(ctx)
handler(httptest.NewRecorder(), req)
cancel()
}
}
// Let the blocked queries finish; with buffered channels each goroutine's
// send now succeeds and the goroutine exits.
close(release)
if !waitForGoroutines(baseline) {
t.Fatalf("goroutines did not return to baseline %d, got %d",
baseline, runtime.NumGoroutine())
}
}
// settledGoroutineCount lets transient goroutines finish, then reports the
// current count.
func settledGoroutineCount() int {
prev := runtime.NumGoroutine()
for range 20 {
time.Sleep(10 * time.Millisecond)
cur := runtime.NumGoroutine()
if cur == prev {
return cur
}
prev = cur
}
return prev
}
// waitForGoroutines waits until the goroutine count drops to target or below.
func waitForGoroutines(target int) bool {
for range 100 {
if runtime.NumGoroutine() <= target {
return true
}
time.Sleep(10 * time.Millisecond)
}
return false
}
internal/server/middleware.go
+169 -144
View File
@@ -3,17 +3,25 @@ package server
import (
"bytes"
"context"
"crypto/rand"
"encoding/json"
"errors"
"fmt"
"log/slog"
"net/http"
"net/netip"
"slices"
"strings"
"sync"
"time"
"github.com/go-chi/chi/v5"
)
// responseWriter wraps http.ResponseWriter to capture the response
type responseWriter struct {
http.ResponseWriter
body *bytes.Buffer
statusCode int
written bool
@@ -86,8 +94,10 @@ func JSONResponseMiddleware(next http.Handler) http.Handler {
}
// Parse the original response
var originalResponse map[string]interface{}
if err := json.Unmarshal(rw.body.Bytes(), &originalResponse); err != nil {
var originalResponse map[string]any
err := json.Unmarshal(rw.body.Bytes(), &originalResponse)
if err != nil {
// If we can't parse it, just write original
w.WriteHeader(rw.statusCode)
_, _ = w.Write(rw.body.Bytes())
@@ -96,7 +106,7 @@ func JSONResponseMiddleware(next http.Handler) http.Handler {
}
// Add @meta field
originalResponse["@meta"] = map[string]interface{}{
originalResponse["@meta"] = map[string]any{
"time_zone": "UTC",
"api_version": 1,
"execution_time": fmt.Sprintf("%d ms", execTime.Milliseconds()),
@@ -109,142 +119,32 @@ func JSONResponseMiddleware(next http.Handler) http.Handler {
})
}
// timeoutWriter wraps ResponseWriter to prevent concurrent writes after timeout
type timeoutWriter struct {
http.ResponseWriter
mu sync.Mutex
written bool
header http.Header // cached header to prevent concurrent access
}
// requestTimeoutMessage is the error message sent with 504 Gateway Timeout.
const requestTimeoutMessage = "Request timeout"
func (tw *timeoutWriter) Write(b []byte) (int, error) {
tw.mu.Lock()
defer tw.mu.Unlock()
// internalErrorMessage is the error message sent with 500 Internal Server
// Error, whatever went wrong; the handler logs the detail.
const internalErrorMessage = "Internal server error"
if tw.written {
return 0, nil // Discard writes after timeout
}
return tw.ResponseWriter.Write(b)
}
func (tw *timeoutWriter) WriteHeader(statusCode int) {
tw.mu.Lock()
defer tw.mu.Unlock()
if tw.written {
return // Discard writes after timeout
}
tw.ResponseWriter.WriteHeader(statusCode)
}
func (tw *timeoutWriter) Header() http.Header {
tw.mu.Lock()
defer tw.mu.Unlock()
if tw.written {
// Return a copy to prevent modifications after timeout
if tw.header == nil {
tw.header = make(http.Header)
}
return tw.header
}
return tw.ResponseWriter.Header()
}
func (tw *timeoutWriter) markWritten() {
tw.mu.Lock()
defer tw.mu.Unlock()
tw.written = true
}
// TimeoutMiddleware creates an HTTP middleware that enforces a request timeout.
// If the handler does not complete within the specified duration, the middleware
// returns a JSON error response with HTTP status 408 (Request Timeout).
// deadlinePassed reports whether ctx ended because its deadline passed.
//
// The timeout parameter specifies the maximum duration allowed for request processing.
// The returned middleware handles panics from the wrapped handler by re-panicking
// after cleanup, and prevents concurrent writes to the response after timeout occurs.
func TimeoutMiddleware(timeout time.Duration, logger *slog.Logger) func(http.Handler) http.Handler {
return func(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
startTime := time.Now()
ctx, cancel := context.WithTimeout(r.Context(), timeout)
defer cancel()
tw := &timeoutWriter{
ResponseWriter: w,
header: make(http.Header),
}
done := make(chan struct{})
panicChan := make(chan interface{}, 1)
go func() {
defer func() {
if p := recover(); p != nil {
panicChan <- p
}
}()
next.ServeHTTP(tw, r.WithContext(ctx))
close(done)
}()
select {
case p := <-panicChan:
panic(p)
case <-done:
return
case <-ctx.Done():
tw.markWritten() // Prevent the handler from writing after timeout
execTime := time.Since(startTime)
// Log the timeout as a warning
logger.Warn("Request timeout",
"method", r.Method,
"path", r.URL.Path,
"duration_ms", execTime.Milliseconds(),
"remote_addr", r.RemoteAddr,
)
// Write directly to the underlying writer since we've marked tw as written
// This is safe because markWritten() prevents the handler from writing
tw.mu.Lock()
w.Header().Set("Content-Type", "application/json")
w.WriteHeader(http.StatusRequestTimeout)
tw.mu.Unlock()
response := map[string]interface{}{
"status": "error",
"error": map[string]interface{}{
"msg": "Request timeout",
"code": http.StatusRequestTimeout,
},
"@meta": map[string]interface{}{
"time_zone": "UTC",
"api_version": 1,
"execution_time": fmt.Sprintf("%d ms", execTime.Milliseconds()),
},
}
_ = json.NewEncoder(w).Encode(response)
}
})
}
// A handler calls it right after each database read or host name lookup,
// before it looks at what came back, error or not, and answers 504 if it
// reports true. A read cut off by the time limit can return an error, or part
// of its rows or none of them without one, so only this check tells a cut-off
// read from a complete one.
func deadlinePassed(ctx context.Context) bool {
return errors.Is(ctx.Err(), context.DeadlineExceeded)
}
// JSONValidationMiddleware is an HTTP middleware that validates JSON API responses.
// It ensures that responses with Content-Type "application/json" contain valid JSON.
//
// If a response is not valid JSON or is empty when JSON is expected, the middleware
// returns a properly formatted JSON error response. For timeout errors (status 408),
// the error message will be "Request timeout". For other errors, it returns
// "Internal server error" with status 500 if the original status was 200.
// returns a properly formatted JSON error response. If the request's time limit
// has passed, it answers 504 with the error message "Request timeout". For other
// errors, it returns "Internal server error" with status 500 if the original
// status was 200.
//
// Non-JSON responses are passed through unchanged.
func JSONValidationMiddleware(next http.Handler) http.Handler {
@@ -273,8 +173,10 @@ func JSONValidationMiddleware(next http.Handler) http.Handler {
// For JSON responses, validate the JSON
if rw.body.Len() > 0 {
var testParse interface{}
if err := json.Unmarshal(rw.body.Bytes(), &testParse); err == nil {
var testParse any
err := json.Unmarshal(rw.body.Bytes(), &testParse)
if err == nil {
// Valid JSON, write it out
w.Header().Set("Content-Type", "application/json")
w.WriteHeader(rw.statusCode)
@@ -293,18 +195,23 @@ func JSONValidationMiddleware(next http.Handler) http.Handler {
if statusCode == http.StatusOK {
statusCode = http.StatusInternalServerError
}
w.WriteHeader(statusCode)
errorMsg := "Internal server error"
if statusCode == http.StatusRequestTimeout {
errorMsg = "Request timeout"
if deadlinePassed(r.Context()) {
statusCode = http.StatusGatewayTimeout
}
response := map[string]interface{}{
"status": "error",
"error": map[string]interface{}{
"msg": errorMsg,
"code": statusCode,
w.WriteHeader(statusCode)
errorMsg := internalErrorMessage
if statusCode == http.StatusGatewayTimeout {
errorMsg = requestTimeoutMessage
}
response := map[string]any{
jsonKeyStatus: jsonStatusError,
jsonKeyError: map[string]any{
jsonKeyMsg: errorMsg,
jsonKeyCode: statusCode,
},
}
_ = json.NewEncoder(w).Encode(response)
@@ -314,6 +221,7 @@ func JSONValidationMiddleware(next http.Handler) http.Handler {
// statusWriter wraps http.ResponseWriter to capture the status code
type statusWriter struct {
http.ResponseWriter
statusCode int
written bool
}
@@ -323,6 +231,7 @@ func (sw *statusWriter) WriteHeader(statusCode int) {
sw.statusCode = statusCode
sw.written = true
}
sw.ResponseWriter.WriteHeader(statusCode)
}
@@ -362,11 +271,12 @@ func RequestLoggerMiddleware(logger *slog.Logger) func(http.Handler) http.Handle
// Slow query threshold (1 second)
const slowQueryThreshold = 1 * time.Second
if sw.statusCode >= http.StatusInternalServerError {
switch {
case sw.statusCode >= http.StatusInternalServerError:
logLevel = slog.LevelError
} else if sw.statusCode >= http.StatusBadRequest {
case sw.statusCode >= http.StatusBadRequest:
logLevel = slog.LevelWarn
} else if duration >= slowQueryThreshold {
case duration >= slowQueryThreshold:
logLevel = slog.LevelWarn
}
@@ -381,3 +291,118 @@ func RequestLoggerMiddleware(logger *slog.Logger) func(http.Handler) http.Handle
})
}
}
// DecodedPathMiddleware makes the router match routes against the decoded
// path, so that the handlers read their URL parameters decoded. Without it,
// chi matches against the escaped path whenever the request has one and hands
// the handlers their parameters still escaped: /ip/2001%3Adb8%3A%3A1 would
// look up "2001%3Adb8%3A%3A1".
func DecodedPathMiddleware(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
chi.RouteContext(r.Context()).RoutePath = r.URL.Path
next.ServeHTTP(w, r)
})
}
// The values of the headers SecurityHeadersMiddleware sets.
const (
// strictTransportSecurity tells browsers to reach this host and its
// subdomains only over HTTPS for a year. The server listens on plain
// HTTP behind a TLS-terminating proxy, which passes the header on.
strictTransportSecurity = "max-age=31536000; includeSubDomains"
// contentSecurityPolicy lets a page load content only from this server,
// and run or apply an inline script or style block only when the block
// carries the nonce made for that answer, which replaces %[1]s. No other
// site may frame the pages.
contentSecurityPolicy = "default-src 'self'; " +
"script-src 'nonce-%[1]s'; style-src 'nonce-%[1]s'; " +
"frame-ancestors 'none'; base-uri 'none'; form-action 'self'"
frameOptions = "DENY"
contentTypeOptions = "nosniff"
referrerPolicy = "strict-origin-when-cross-origin"
// permissionsPolicy turns off browser features the pages do not use.
permissionsPolicy = "accelerometer=(), camera=(), geolocation=(), " +
"gyroscope=(), magnetometer=(), microphone=(), payment=(), usb=()"
)
// cspNonceKey is the request context key of the nonce SecurityHeadersMiddleware
// made for the answer.
type cspNonceKey struct{}
// SecurityHeadersMiddleware sets the security headers REPO_POLICIES.md requires
// on every answer. Each answer's Content-Security-Policy names a nonce of its
// own; a page reads it with cspNonce and writes it on its script and style
// blocks.
func SecurityHeadersMiddleware(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
nonce := rand.Text()
header := w.Header()
header.Set("Strict-Transport-Security", strictTransportSecurity)
header.Set("Content-Security-Policy",
fmt.Sprintf(contentSecurityPolicy, nonce))
header.Set("X-Frame-Options", frameOptions)
header.Set("X-Content-Type-Options", contentTypeOptions)
header.Set("Referrer-Policy", referrerPolicy)
header.Set("Permissions-Policy", permissionsPolicy)
ctx := context.WithValue(r.Context(), cspNonceKey{}, nonce)
next.ServeHTTP(w, r.WithContext(ctx))
})
}
// cspNonce returns the nonce SecurityHeadersMiddleware made for the answer to
// the request with ctx.
func cspNonce(ctx context.Context) string {
nonce, _ := ctx.Value(cspNonceKey{}).(string)
return nonce
}
// RealIPMiddleware gives a request from one of trustedProxies the address of
// the client the proxies forwarded it for as its RemoteAddr: the last
// X-Forwarded-For entry that is not a trusted proxy. Each proxy appends the
// address it took the request from, so the entries left of that one are
// whatever the client sent. A request from any other peer keeps the peer's
// address, whatever headers it carries.
func RealIPMiddleware(trustedProxies []netip.Prefix) func(http.Handler) http.Handler {
isTrusted := func(addr netip.Addr) bool {
return slices.ContainsFunc(trustedProxies, func(proxy netip.Prefix) bool {
return proxy.Contains(addr.Unmap())
})
}
return func(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
peer, err := netip.ParseAddrPort(r.RemoteAddr)
if err != nil || !isTrusted(peer.Addr()) {
next.ServeHTTP(w, r)
return
}
// A header sent on several lines is one list.
entries := strings.Split(
strings.Join(r.Header.Values("X-Forwarded-For"), ","), ",")
for _, entry := range slices.Backward(entries) {
addr, err := netip.ParseAddr(strings.TrimSpace(entry))
if err != nil {
break
}
r.RemoteAddr = addr.String()
if !isTrusted(addr) {
break
}
}
next.ServeHTTP(w, r)
})
}
}
internal/server/middleware_internal_test.go
+205
View File
@@ -0,0 +1,205 @@
package server
import (
"encoding/json"
"net/http"
"net/http/httptest"
"net/netip"
"strings"
"testing"
"time"
)
// The addresses the forwarding tests use: the client a proxy names in
// X-Forwarded-For, a peer inside the trusted range of
// TestRealIPMiddlewareTrustsOnlyConfiguredProxies, and a peer outside it,
// which lies in testPrefixV4 so that an IP lookup of it finds a route.
const (
forwardedClient = "203.0.113.9"
trustedPeer = "10.0.0.2:1234"
untrustedPeerAddr = "192.0.2.1"
untrustedPeer = untrustedPeerAddr + ":1234"
)
// TestResponsesCarrySecurityHeaders checks that a page, a JSON route and a
// path no route matches each answer with the security headers REPO_POLICIES.md
// requires, and that the Content-Security-Policy allows no inline content
// without a nonce (https://git.eeqj.de/sneak/routewatch/issues/108).
func TestResponsesCarrySecurityHeaders(t *testing.T) {
t.Parallel()
s := newServerWithTestRoutes(t)
headers := []string{
"Strict-Transport-Security",
"Content-Security-Policy",
"X-Frame-Options",
"X-Content-Type-Options",
"Referrer-Policy",
"Permissions-Policy",
}
for _, path := range []string{"/", statsPath, "/no-such-page"} {
rec := httptest.NewRecorder()
s.router.ServeHTTP(rec, httptest.NewRequestWithContext(
t.Context(), http.MethodGet, path, nil))
for _, name := range headers {
if rec.Header().Get(name) == "" {
t.Errorf("%s: no %s header", path, name)
}
}
policy := rec.Header().Get("Content-Security-Policy")
if strings.Contains(policy, "'unsafe-") {
t.Errorf("%s: Content-Security-Policy %q allows unsafe content",
path, policy)
}
}
}
// TestPagesCarryTheCSPNonce checks that every script and style block of each
// page carries the nonce its Content-Security-Policy header names, so that the
// browser runs and applies them, that each answer has a nonce of its own, and
// that no page uses a style attribute, which the policy does not allow.
func TestPagesCarryTheCSPNonce(t *testing.T) {
t.Parallel()
s := newServerWithTestRoutes(t)
_, err := s.db.GetOrCreateASN(64500, time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC))
if err != nil {
t.Fatalf("GetOrCreateASN: %v", err)
}
paths := []string{
"/", "/status", asPagePath, "/prefix/" + testPrefixV4,
prefixLengthPagePath, "/prefixlength6/32",
}
seen := make(map[string]bool)
for _, path := range paths {
rec := httptest.NewRecorder()
s.router.ServeHTTP(rec, httptest.NewRequestWithContext(
t.Context(), http.MethodGet, path, nil))
if rec.Code != http.StatusOK {
t.Errorf("%s: status %d, want %d; body %s",
path, rec.Code, http.StatusOK, rec.Body)
continue
}
policy := rec.Header().Get("Content-Security-Policy")
_, nonce, found := strings.Cut(policy, "'nonce-")
if !found {
t.Errorf("%s: Content-Security-Policy %q names no nonce", path, policy)
continue
}
nonce, _, _ = strings.Cut(nonce, "'")
if seen[nonce] {
t.Errorf("%s: nonce %q was already used for another answer", path, nonce)
}
seen[nonce] = true
page := rec.Body.String()
for _, tag := range []string{"<script", "<style"} {
withNonce := tag + ` nonce="` + nonce + `"`
if strings.Count(page, tag) != strings.Count(page, withNonce) {
t.Errorf("%s: a %s tag does not carry the nonce %q", path, tag, nonce)
}
}
if strings.Contains(page, "style=") {
t.Errorf("%s: the page uses a style attribute", path)
}
}
}
// TestForwardedForFromUntrustedPeerIsIgnored asks /ip/ for the caller's own
// address from a peer that is not a trusted proxy, with forwarding headers
// naming another address, and checks that the lookup is for the TCP peer.
// Before the fix any client could choose the address it was looked up and
// logged as (https://git.eeqj.de/sneak/routewatch/issues/108).
func TestForwardedForFromUntrustedPeerIsIgnored(t *testing.T) {
t.Parallel()
s := newServerWithTestRoutes(t)
req := httptest.NewRequestWithContext(t.Context(), http.MethodGet, "/ip/", nil)
req.RemoteAddr = untrustedPeer
req.Header.Set("X-Forwarded-For", forwardedClient)
req.Header.Set("X-Real-IP", forwardedClient)
req.Header.Set("True-Client-IP", forwardedClient)
rec := httptest.NewRecorder()
s.router.ServeHTTP(rec, req)
var body struct {
Data IPLookupResponse `json:"data"`
}
err := json.Unmarshal(rec.Body.Bytes(), &body)
if err != nil {
t.Fatalf("decoding the answer %q: %v", rec.Body, err)
}
if body.Data.Query != untrustedPeerAddr {
t.Errorf("looked up %q, want the TCP peer %s; answer %s",
body.Data.Query, untrustedPeerAddr, rec.Body)
}
}
// TestRealIPMiddlewareTrustsOnlyConfiguredProxies checks which address
// RealIPMiddleware gives a request: the TCP peer, unless the peer is a trusted
// proxy; then the last X-Forwarded-For entry that is not a trusted proxy,
// since the entries left of it are whatever the client sent.
func TestRealIPMiddlewareTrustsOnlyConfiguredProxies(t *testing.T) {
t.Parallel()
trusted := []netip.Prefix{netip.MustParsePrefix("10.0.0.0/8")}
tests := []struct {
name string
peer string
forwardedFor string
want string
}{
{"untrusted peer", untrustedPeer, forwardedClient, untrustedPeer},
{"trusted peer", trustedPeer, forwardedClient, forwardedClient},
{
"entries the client wrote", trustedPeer,
"198.51.100.7, " + forwardedClient, forwardedClient,
},
{
"two trusted proxies", trustedPeer,
forwardedClient + ", 10.0.0.3", forwardedClient,
},
{"no header", trustedPeer, "", trustedPeer},
{"entry not an address", trustedPeer, "client", trustedPeer},
}
for _, tt := range tests {
var got string
handler := RealIPMiddleware(trusted)(http.HandlerFunc(
func(_ http.ResponseWriter, r *http.Request) { got = r.RemoteAddr }))
req := httptest.NewRequestWithContext(t.Context(), http.MethodGet, "/", nil)
req.RemoteAddr = tt.peer
if tt.forwardedFor != "" {
req.Header.Set("X-Forwarded-For", tt.forwardedFor)
}
handler.ServeHTTP(httptest.NewRecorder(), req)
if got != tt.want {
t.Errorf("%s: RemoteAddr %q, want %q", tt.name, got, tt.want)
}
}
}
internal/server/routes.go
+12 -4
View File
@@ -12,19 +12,27 @@ func (s *Server) setupRoutes() {
r := chi.NewRouter()
// Middleware
r.Use(DecodedPathMiddleware)
r.Use(SecurityHeadersMiddleware)
r.Use(middleware.RequestID)
r.Use(middleware.RealIP)
r.Use(RealIPMiddleware(s.trustedProxies))
r.Use(RequestLoggerMiddleware(s.logger.Logger)) // Structured request logging
r.Use(middleware.Recoverer)
const requestTimeout = 30 * time.Second // Increased from 8s for slow queries
r.Use(TimeoutMiddleware(requestTimeout, s.logger.Logger))
r.Use(JSONResponseMiddleware)
// Timeout must come after JSONResponseMiddleware, which sends its answer
// when the handler returns and would send a 200 ahead of Timeout's 504 for
// a handler that wrote nothing.
const requestTimeout = 30 * time.Second // Increased from 8s for slow queries
r.Use(middleware.Timeout(requestTimeout))
// Routes
r.Get("/", s.handleIndex())
r.Get("/status", s.handleStatusHTML())
r.Get("/status.json", JSONValidationMiddleware(s.handleStatusJSON()).ServeHTTP)
r.Get("/.well-known/healthcheck.json", JSONValidationMiddleware(s.handleHealthCheck()).ServeHTTP)
r.Get(
"/.well-known/healthcheck.json",
JSONValidationMiddleware(s.handleHealthCheck()).ServeHTTP,
)
// AS and prefix detail pages
r.Get("/as/{asn}", s.handleASDetail())
internal/server/server.go
+61 -19
View File
@@ -3,10 +3,15 @@ package server
import (
"context"
"errors"
"fmt"
"net"
"net/http"
"os"
"net/netip"
"strconv"
"time"
"git.eeqj.de/sneak/routewatch/internal/config"
"git.eeqj.de/sneak/routewatch/internal/database"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/streamer"
@@ -27,22 +32,46 @@ type ASNFetcher interface {
GetStats() ASNFetcherStats
}
// dnsResolver looks up the addresses of host names and the names of addresses
// in DNS, as *net.Resolver does.
type dnsResolver interface {
LookupHost(ctx context.Context, host string) ([]string, error)
LookupAddr(ctx context.Context, addr string) ([]string, error)
}
// Server provides HTTP endpoints for status monitoring
type Server struct {
router *chi.Mux
db database.Store
streamer *streamer.Streamer
logger *logger.Logger
port int
srv *http.Server
asnFetcher ASNFetcher
// trustedProxies are the reverse proxies whose X-Forwarded-For header
// RealIPMiddleware believes
trustedProxies []netip.Prefix
// resolver makes the IP lookup's DNS lookups: net.DefaultResolver, or a
// fake in tests
resolver dnsResolver
}
// New creates a new HTTP server
func New(db database.Store, streamer *streamer.Streamer, logger *logger.Logger) *Server {
func New(
db database.Store,
streamer *streamer.Streamer,
logger *logger.Logger,
cfg *config.Config,
) *Server {
s := &Server{
db: db,
streamer: streamer,
logger: logger,
db: db,
streamer: streamer,
logger: logger,
port: cfg.Port,
trustedProxies: cfg.TrustedProxies,
resolver: net.DefaultResolver,
}
s.setupRoutes()
@@ -50,14 +79,12 @@ func New(db database.Store, streamer *streamer.Streamer, logger *logger.Logger)
return s
}
// Start starts the HTTP server
func (s *Server) Start() error {
port := os.Getenv("PORT")
if port == "" {
port = "8080"
}
// Start listens on the configured port and serves HTTP in the background.
// It returns an error if it cannot listen, for example because the port is
// already in use. ctx bounds only opening the port, not serving.
func (s *Server) Start(ctx context.Context) error {
const (
network = "tcp"
readHeaderTimeout = 40 * time.Second
readTimeout = 60 * time.Second
writeTimeout = 60 * time.Second
@@ -65,7 +92,7 @@ func (s *Server) Start() error {
)
s.srv = &http.Server{
Addr: ":" + port,
Addr: ":" + strconv.Itoa(s.port),
Handler: s.router,
ReadHeaderTimeout: readHeaderTimeout,
ReadTimeout: readTimeout,
@@ -73,12 +100,20 @@ func (s *Server) Start() error {
IdleTimeout: idleTimeout,
}
s.logger.Info("Starting HTTP server", "port", port, "addr", s.srv.Addr)
s.logger.Info("Starting HTTP server", "port", s.port, "addr", s.srv.Addr)
var listenConfig net.ListenConfig
listener, err := listenConfig.Listen(ctx, network, s.srv.Addr)
if err != nil {
return fmt.Errorf("starting HTTP server: %w", err)
}
s.logger.Info("HTTP server listening", "addr", listener.Addr().String())
// Start in goroutine but log when actually listening
go func() {
s.logger.Info("HTTP server listening", "addr", s.srv.Addr)
if err := s.srv.ListenAndServe(); err != nil && err != http.ErrServerClosed {
err := s.srv.Serve(listener)
if err != nil && !errors.Is(err, http.ErrServerClosed) {
s.logger.Error("HTTP server error", "error", err)
}
}()
@@ -86,7 +121,9 @@ func (s *Server) Start() error {
return nil
}
// Stop gracefully stops the HTTP server
// Stop stops the HTTP server. It waits, until ctx ends, for the requests in
// progress to finish, and then closes the connections of those still running
// and returns ctx's error.
func (s *Server) Stop(ctx context.Context) error {
if s.srv == nil {
return nil
@@ -94,7 +131,12 @@ func (s *Server) Stop(ctx context.Context) error {
s.logger.Info("Stopping HTTP server")
return s.srv.Shutdown(ctx)
err := s.srv.Shutdown(ctx)
if err != nil {
return errors.Join(err, s.srv.Close())
}
return nil
}
// SetASNFetcher sets the ASN WHOIS fetcher for on-demand lookups.
internal/server/server_internal_test.go
+92
View File
@@ -0,0 +1,92 @@
package server
import (
"context"
"net"
"net/http"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/logger"
)
// TestStopClosesTheConnectionOfARequestStillRunning serves a request whose
// handler does not return until the test ends, and stops the server with a
// deadline that passes while the request runs. The client must then see its
// connection closed. Before the fix Stop returned at the deadline and left the
// connection open until the handler returned.
func TestStopClosesTheConnectionOfARequestStillRunning(t *testing.T) {
t.Parallel()
const (
stopDeadline = 100 * time.Millisecond
// waitTimeout is how long the test waits for the request to reach
// the handler, and then for the client to see its connection closed.
waitTimeout = 2 * time.Second
// readHeaderTimeout bounds the test server's read of the request
// header, as every http.Server must.
readHeaderTimeout = time.Second
)
release := make(chan struct{})
defer close(release)
started := make(chan struct{})
var listenConfig net.ListenConfig
listener, err := listenConfig.Listen(t.Context(), "tcp", "127.0.0.1:0")
if err != nil {
t.Fatalf("failed to open the test listener: %v", err)
}
s := &Server{logger: logger.New(), srv: &http.Server{
Handler: http.HandlerFunc(func(http.ResponseWriter, *http.Request) {
close(started)
<-release
}),
ReadHeaderTimeout: readHeaderTimeout,
}}
go func() { _ = s.srv.Serve(listener) }()
answered := make(chan error, 1)
go func() {
req, err := http.NewRequestWithContext(t.Context(), http.MethodGet,
"http://"+listener.Addr().String()+"/", nil)
if err != nil {
answered <- err
return
}
resp, err := http.DefaultClient.Do(req)
if err == nil {
_ = resp.Body.Close()
}
answered <- err
}()
select {
case <-started:
case <-time.After(waitTimeout):
t.Fatal("the request did not reach the handler")
}
ctx, cancel := context.WithTimeout(t.Context(), stopDeadline)
defer cancel()
_ = s.Stop(ctx)
select {
case err := <-answered:
if err == nil {
t.Error("the request still running at the deadline was answered")
}
case <-time.After(waitTimeout):
t.Fatal("the connection of the request still running at the " +
"deadline was open after Stop returned")
}
}
internal/server/server_test.go
+47
View File
@@ -0,0 +1,47 @@
package server_test
import (
"context"
"net"
"testing"
"git.eeqj.de/sneak/routewatch/internal/config"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/metrics"
"git.eeqj.de/sneak/routewatch/internal/server"
"git.eeqj.de/sneak/routewatch/internal/streamer"
)
// TestStartFailsWhenPortIsHeld holds a port with a listener of its own and
// checks that Start on that port returns an error. Before the fix Start
// listened in a goroutine, only logged the failure and returned nil, so the
// daemon kept running without its web server.
func TestStartFailsWhenPortIsHeld(t *testing.T) {
t.Parallel()
var listenConfig net.ListenConfig
held, err := listenConfig.Listen(t.Context(), "tcp", ":0")
if err != nil {
t.Fatalf("failed to open the test listener: %v", err)
}
defer func() { _ = held.Close() }()
addr, ok := held.Addr().(*net.TCPAddr)
if !ok {
t.Fatalf("test listener address %v is not a TCP address", held.Addr())
}
port := addr.Port
log := logger.New()
s := server.New(nil, streamer.New(log, metrics.New()), log,
&config.Config{Port: port})
defer func() { _ = s.Stop(context.Background()) }()
err = s.Start(t.Context())
if err == nil {
t.Fatalf("Start returned nil while port %d was held by another listener",
port)
}
}
internal/streamer/streamer.go
+445 -222
View File
@@ -1,5 +1,6 @@
// Package streamer implements an HTTP client that connects to the RIPE RIS Live streaming API,
// parses BGP UPDATE messages from the JSON stream, and dispatches them to registered handlers.
// Package streamer implements an HTTP client that connects to the RIPE RIS Live
// streaming API, parses BGP UPDATE messages from the JSON stream, and
// dispatches them to registered handlers.
package streamer
import (
@@ -7,6 +8,7 @@ import (
"compress/gzip"
"context"
"encoding/json"
"errors"
"fmt"
"io"
"math"
@@ -24,33 +26,37 @@ import (
// countingReader wraps an io.Reader and counts bytes read
type countingReader struct {
reader io.Reader
count int64
count atomic.Int64
}
// Read implements io.Reader and counts bytes
func (c *countingReader) Read(p []byte) (int, error) {
n, err := c.reader.Read(p)
atomic.AddInt64(&c.count, int64(n))
c.count.Add(int64(n))
return n, err
}
// Count returns the total bytes read
func (c *countingReader) Count() int64 {
return atomic.LoadInt64(&c.count)
return c.count.Load()
}
// Configuration constants for the RIS Live streamer.
const (
risLiveURL = "https://ris-live.ripe.net/v1/stream/?format=json&" +
"client=https%3A%2F%2Fgit.eeqj.de%2Fsneak%2Froutewatch"
metricsWindowSize = 60 // seconds for rolling average
metricsUpdateRate = time.Second
minBackoffDelay = 5 * time.Second
maxBackoffDelay = 320 * time.Second
metricsLogInterval = 10 * time.Second
bytesPerKB = 1024
bytesPerMB = 1024 * 1024
metricsWindowSize = 60 // seconds for rolling average
metricsUpdateRate = time.Second
minBackoffDelay = 5 * time.Second
maxBackoffDelay = 320 * time.Second
// Reconnect when no line arrives for this long
streamIdleTimeout = 60 * time.Second
metricsLogInterval = 10 * time.Second
bytesPerKB = 1024
bytesPerMB = 1024 * 1024
// Well above an extended BGP message (RFC 8654) in hex plus its prefixes
maxLineSize = bytesPerMB
maxConcurrentHandlers = 800 // Maximum number of concurrent message handlers
// Backpressure constants
@@ -58,11 +64,26 @@ const (
backpressureSlope = 2.0 // Slope for linear drop probability increase
)
// messageTypeNotification is the type of a BGP NOTIFICATION message, which ends
// the session with the peer.
const messageTypeNotification = "NOTIFICATION"
var (
errAlreadyRunning = errors.New("streamer already running")
errUnexpectedStatus = errors.New("unexpected status code")
// errStreamEnded is returned by stream when RIS Live ends the response.
errStreamEnded = errors.New("RIS Live ended the stream")
// errStreamIdle is returned by stream when no line arrived for the idle
// timeout.
errStreamIdle = errors.New("no line received")
)
// MessageHandler defines the interface for processing RIS messages.
// Implementations must specify which message types they want to receive,
// how to process messages, and their desired queue capacity.
type MessageHandler interface {
// WantsMessage returns true if this handler wants to process messages of the given type.
// WantsMessage returns true if this handler wants to process messages of
// the given type.
WantsMessage(messageType string) bool
// HandleMessage processes a RIS message. This method is called from a dedicated
@@ -73,7 +94,8 @@ type MessageHandler interface {
// QueueCapacity returns the desired queue capacity for this handler.
// Handlers that process quickly can have larger queues to buffer bursts.
// When the queue fills up, messages will be dropped according to the
// backpressure algorithm.
// backpressure algorithm, except STATE messages, which are held until
// there is room.
QueueCapacity() int
}
@@ -97,6 +119,9 @@ type handlerInfo struct {
handler MessageHandler
queue chan *ristypes.RISMessage
metrics handlerMetrics
// heldState holds, in feed order, the STATE messages that found queue
// full (see queueMessage). Only the read loop uses it, and then Stop.
heldState []*ristypes.RISMessage
}
// Streamer manages a connection to the RIPE RIS Live streaming API for receiving
@@ -104,16 +129,21 @@ type handlerInfo struct {
// backoff, dispatches messages to registered handlers via per-handler queues, and
// implements backpressure to prevent queue overflow during high traffic periods.
type Streamer struct {
logger *logger.Logger
client *http.Client
url string
logger *logger.Logger
client *http.Client
url string
// Drop the connection when no line arrives for this long
idleTimeout time.Duration
// First wait before reconnecting; doubles up to maxBackoffDelay
minBackoff time.Duration
handlers []*handlerInfo
rawHandler RawMessageHandler
mu sync.RWMutex
cancel context.CancelFunc
running bool
workers sync.WaitGroup // the handler workers; Stop waits for them
metrics *metrics.Tracker
totalDropped uint64 // Total dropped messages across all handlers
totalDropped atomic.Uint64 // Total dropped messages across all handlers
random *rand.Rand // Random number generator for backpressure drops
bgpPeers map[string]bool // Track active BGP peers by peer IP
bgpPeersMu sync.RWMutex // Protects bgpPeers map
@@ -121,11 +151,14 @@ type Streamer struct {
// New creates a new Streamer instance configured to connect to the RIS Live API.
// The logger is used for structured logging of connection events and errors.
// The metrics tracker is used to record message counts, bytes received, and connection status.
// The metrics tracker is used to record message counts, bytes received, and
// connection status.
func New(logger *logger.Logger, metrics *metrics.Tracker) *Streamer {
return &Streamer{
logger: logger,
url: risLiveURL,
logger: logger,
url: risLiveURL,
idleTimeout: streamIdleTimeout,
minBackoff: minBackoffDelay,
client: &http.Client{
Timeout: 0, // No timeout for streaming
Transport: &http.Transport{
@@ -153,7 +186,8 @@ func (s *Streamer) RegisterHandler(handler MessageHandler) {
handler: handler,
queue: make(chan *ristypes.RISMessage, handler.QueueCapacity()),
metrics: handlerMetrics{
minTime: time.Duration(math.MaxInt64), // Initialize to max so first value sets the floor
// Initialize to max so first value sets the floor
minTime: time.Duration(math.MaxInt64),
},
}
@@ -161,37 +195,49 @@ func (s *Streamer) RegisterHandler(handler MessageHandler) {
// If we're already running, start a worker for this handler
if s.running {
s.workers.Add(1)
go s.runHandlerWorker(info)
}
}
// SetURL makes the streamer read from url instead of RIS Live. Call it before
// Start. Tests use it to read from a local server.
func (s *Streamer) SetURL(url string) {
s.url = url
}
// RegisterRawHandler sets a callback to receive raw JSON lines from the stream
// before they are parsed. Only one raw handler can be registered at a time;
// subsequent calls will replace the previous handler.
func (s *Streamer) RegisterRawHandler(handler RawMessageHandler) {
s.mu.Lock()
defer s.mu.Unlock()
s.rawHandler = handler
}
// Start begins streaming BGP updates from the RIS Live API in a background goroutine.
// It starts worker goroutines for each registered handler and manages automatic
// reconnection with exponential backoff on connection failures.
// Returns an error if the streamer is already running.
func (s *Streamer) Start() error {
// Returns an error if the streamer is already running. It streams until Stop:
// the end of ctx does not stop it.
func (s *Streamer) Start(ctx context.Context) error {
s.mu.Lock()
defer s.mu.Unlock()
if s.running {
return fmt.Errorf("streamer already running")
return errAlreadyRunning
}
ctx, cancel := context.WithCancel(context.Background())
ctx, cancel := context.WithCancel(context.WithoutCancel(ctx))
s.cancel = cancel
s.running = true
// Start workers for each handler
for _, info := range s.handlers {
s.workers.Add(1)
go s.runHandlerWorker(info)
}
@@ -205,50 +251,74 @@ func (s *Streamer) Start() error {
return nil
}
// Stop halts the streaming connection and shuts down all handler workers.
// It cancels the streaming context, closes all handler queues, and updates
// the connection status in metrics. This method is safe to call multiple times.
func (s *Streamer) Stop() {
// Stop ends the stream and stops the handler workers. It returns once each
// worker has handed its handler every message left on its queue and then the
// STATE messages held for it (see queueMessage), or once ctx ends, and then
// logs how many messages were not handed over. This method is safe to call
// multiple times.
func (s *Streamer) Stop(ctx context.Context) {
s.mu.Lock()
if s.cancel != nil {
s.cancel()
if s.cancel == nil {
// Not started, or already stopped: closing the queues again would panic.
s.mu.Unlock()
return
}
// Close all handler queues to signal workers to stop
// Once the stream's context is cancelled the read loop queues and holds
// nothing more (see dispatchMessage).
s.cancel()
s.cancel = nil
notHandedOver := 0
for _, info := range s.handlers {
// The held messages came after everything on the queue. Each waits
// for the worker, which does not take s.mu, to make room, until ctx
// ends.
for len(info.heldState) > 0 && ctx.Err() == nil {
select {
case info.queue <- info.heldState[0]:
info.heldState = info.heldState[1:]
case <-ctx.Done():
}
}
notHandedOver += len(info.heldState)
info.heldState = nil
// The worker returns once it has emptied the closed queue.
close(info.queue)
}
handlers := s.handlers
s.running = false
s.mu.Unlock()
s.metrics.SetConnected(false)
}
// runHandlerWorker processes messages for a specific handler
func (s *Streamer) runHandlerWorker(info *handlerInfo) {
for msg := range info.queue {
start := time.Now()
info.handler.HandleMessage(msg)
elapsed := time.Since(start)
workersDone := make(chan struct{})
// Update metrics
info.metrics.mu.Lock()
info.metrics.processedCount++
info.metrics.totalTime += elapsed
go func() {
s.workers.Wait()
close(workersDone)
}()
// Update min time
if elapsed < info.metrics.minTime {
info.metrics.minTime = elapsed
select {
case <-workersDone:
case <-ctx.Done():
for _, info := range handlers {
notHandedOver += len(info.queue)
}
}
// Update max time
if elapsed > info.metrics.maxTime {
info.metrics.maxTime = elapsed
}
info.metrics.mu.Unlock()
if notHandedOver > 0 {
s.logger.Error("Stop deadline passed before the handlers took every "+
"message", "messages", notHandedOver)
}
}
// IsRunning reports whether the streamer is currently connected and processing messages.
// This is safe to call concurrently from multiple goroutines.
// IsRunning reports whether the streamer is currently connected and processing
// messages. This is safe to call concurrently from multiple goroutines.
func (s *Streamer) IsRunning() bool {
s.mu.RLock()
defer s.mu.RUnlock()
@@ -309,12 +379,15 @@ func (s *Streamer) GetHandlerStats() []HandlerStats {
// Calculate average time
if info.metrics.processedCount > 0 {
processedCount := info.metrics.processedCount
const maxInt64 = 1<<63 - 1
if processedCount > maxInt64 {
processedCount = maxInt64
}
//nolint:gosec // processedCount is explicitly bounded above
hs.AvgProcessTime = info.metrics.totalTime / time.Duration(processedCount)
hs.AvgProcessTime = info.metrics.totalTime /
time.Duration(processedCount)
}
info.metrics.mu.Unlock()
@@ -325,11 +398,40 @@ func (s *Streamer) GetHandlerStats() []HandlerStats {
return stats
}
// GetDroppedMessages returns the total number of messages dropped across all handlers
// due to queue overflow or backpressure. This counter is monotonically increasing
// and is safe to call concurrently.
// GetDroppedMessages returns the total number of messages dropped across all
// handlers due to queue overflow or backpressure. This counter is monotonically
// increasing and is safe to call concurrently.
func (s *Streamer) GetDroppedMessages() uint64 {
return atomic.LoadUint64(&s.totalDropped)
return s.totalDropped.Load()
}
// runHandlerWorker processes messages for a specific handler
func (s *Streamer) runHandlerWorker(info *handlerInfo) {
defer s.workers.Done()
for msg := range info.queue {
start := time.Now()
info.handler.HandleMessage(msg)
elapsed := time.Since(start)
// Update metrics
info.metrics.mu.Lock()
info.metrics.processedCount++
info.metrics.totalTime += elapsed
// Update min time
if elapsed < info.metrics.minTime {
info.metrics.minTime = elapsed
}
// Update max time
if elapsed > info.metrics.maxTime {
info.metrics.maxTime = elapsed
}
info.metrics.mu.Unlock()
}
}
// logMetrics logs the current streaming statistics
@@ -338,7 +440,8 @@ func (s *Streamer) logMetrics() {
uptime := time.Since(metrics.ConnectedSince)
const bitsPerMegabit = 1000000
totalDropped := atomic.LoadUint64(&s.totalDropped)
totalDropped := s.totalDropped.Load()
s.logger.Info(
"Stream statistics",
@@ -364,16 +467,19 @@ func (s *Streamer) logMetrics() {
// Log per-handler statistics
s.mu.RLock()
for i, info := range s.handlers {
info.metrics.mu.Lock()
if info.metrics.processedCount > 0 {
// Safe conversion: processedCount is bounded by maxInt64
processedCount := info.metrics.processedCount
const maxInt64 = 1<<63 - 1
if processedCount > maxInt64 {
processedCount = maxInt64
}
//nolint:gosec // processedCount is explicitly bounded above
avgTime := info.metrics.totalTime / time.Duration(processedCount)
s.logger.Info(
"Handler statistics",
@@ -390,6 +496,7 @@ func (s *Streamer) logMetrics() {
}
info.metrics.mu.Unlock()
}
s.mu.RUnlock()
}
@@ -398,9 +505,10 @@ func (s *Streamer) updateMetrics(messageBytes int) {
s.metrics.RecordMessage(int64(messageBytes))
}
// streamWithReconnect handles streaming with automatic reconnection and exponential backoff
// streamWithReconnect handles streaming with automatic reconnection and
// exponential backoff
func (s *Streamer) streamWithReconnect(ctx context.Context) {
backoffDelay := minBackoffDelay
backoffDelay := s.minBackoff
consecutiveFailures := 0
for {
@@ -417,8 +525,9 @@ func (s *Streamer) streamWithReconnect(ctx context.Context) {
err := s.stream(ctx)
streamDuration := time.Since(startTime)
if err == nil {
// Clean exit (context cancelled)
// Only Stop ends the streamer. A stream that ends any other way,
// with or without an error, is a disconnect and is reconnected.
if ctx.Err() != nil {
return
}
@@ -429,17 +538,14 @@ func (s *Streamer) streamWithReconnect(ctx context.Context) {
"stream_duration", streamDuration)
s.metrics.SetConnected(false)
// Check if context is cancelled
if ctx.Err() != nil {
return
}
// If we streamed for more than 30 seconds, reset the backoff
// This indicates we had a successful connection that received data
// A stream that lasted more than 30 seconds resets the backoff. That
// includes a stream dropped after streamIdleTimeout with no line, so
// a silent feed is retried after the first backoff delay every time.
if streamDuration > 30*time.Second {
s.logger.Info("Resetting backoff delay due to successful connection",
"stream_duration", streamDuration)
backoffDelay = minBackoffDelay
backoffDelay = s.minBackoff
consecutiveFailures = 0
} else {
// Increment consecutive failures
@@ -472,7 +578,13 @@ func (s *Streamer) stream(ctx context.Context) error {
connCtx, connCancel := context.WithCancel(ctx)
defer connCancel()
req, err := http.NewRequestWithContext(ctx, "GET", s.url, nil)
// The idle timer drops the connection, by cancelling connCtx, when no
// line arrives for s.idleTimeout. Each line read starts it over. It
// also covers the wait for the server to answer.
idleTimer := time.AfterFunc(s.idleTimeout, connCancel)
defer idleTimer.Stop()
req, err := http.NewRequestWithContext(connCtx, http.MethodGet, s.url, nil)
if err != nil {
return fmt.Errorf("failed to create request: %w", err)
}
@@ -485,13 +597,14 @@ func (s *Streamer) stream(ctx context.Context) error {
return fmt.Errorf("failed to connect to RIS Live: %w", err)
}
defer func() {
if err := resp.Body.Close(); err != nil {
err := resp.Body.Close()
if err != nil {
s.logger.Error("Failed to close response body", "error", err)
}
}()
if resp.StatusCode != http.StatusOK {
return fmt.Errorf("unexpected status code: %d", resp.StatusCode)
return fmt.Errorf("%w: %d", errUnexpectedStatus, resp.StatusCode)
}
// Wrap body with counting reader to track actual wire bytes
@@ -505,7 +618,9 @@ func (s *Streamer) stream(ctx context.Context) error {
return fmt.Errorf("failed to create gzip reader: %w", err)
}
defer func() { _ = gzReader.Close() }()
reader = gzReader
s.logger.Info("Connected to RIS Live stream", "compression", "gzip")
} else {
s.logger.Info("Connected to RIS Live stream", "compression", "none")
@@ -513,9 +628,6 @@ func (s *Streamer) stream(ctx context.Context) error {
s.metrics.SetConnected(true)
// Track wire bytes for metrics updates
var lastWireBytes int64
// Start metrics logging goroutine
metricsTicker := time.NewTicker(metricsLogInterval)
defer metricsTicker.Stop()
@@ -532,28 +644,50 @@ func (s *Streamer) stream(ctx context.Context) error {
}()
// Wire byte update ticker - update metrics with actual wire bytes periodically
go s.recordWireBytes(connCtx, wireCounter)
return s.readStream(ctx, connCtx, reader, idleTimer)
}
// recordWireBytes adds the bytes wireCounter has counted to the metrics once a
// second until ctx is cancelled.
func (s *Streamer) recordWireBytes(
ctx context.Context, wireCounter *countingReader,
) {
wireUpdateTicker := time.NewTicker(time.Second)
defer wireUpdateTicker.Stop()
go func() {
for {
select {
case <-wireUpdateTicker.C:
currentBytes := wireCounter.Count()
delta := currentBytes - lastWireBytes
if delta > 0 {
s.metrics.RecordWireBytes(delta)
lastWireBytes = currentBytes
}
case <-connCtx.Done():
return
}
}
}()
var lastWireBytes int64
for {
select {
case <-wireUpdateTicker.C:
currentBytes := wireCounter.Count()
delta := currentBytes - lastWireBytes
if delta > 0 {
s.metrics.RecordWireBytes(delta)
lastWireBytes = currentBytes
}
case <-ctx.Done():
return
}
}
}
// readStream hands each line of the stream to processLine until the stream
// ends, a line fails, or ctx is cancelled. Each line restarts idleTimer, which
// cancels connCtx when it fires.
func (s *Streamer) readStream(
ctx, connCtx context.Context, reader io.Reader, idleTimer *time.Timer,
) error {
scanner := bufio.NewScanner(reader)
scanner.Buffer(nil, maxLineSize)
for scanner.Scan() {
idleTimer.Reset(s.idleTimeout)
select {
case <-ctx.Done():
s.logger.Info("Stream stopped by context")
@@ -567,142 +701,231 @@ func (s *Streamer) stream(ctx context.Context) error {
continue
}
// Update metrics with decompressed message size
s.updateMetrics(len(line))
// Call raw handler if registered
s.mu.RLock()
rawHandler := s.rawHandler
s.mu.RUnlock()
if rawHandler != nil {
// Call raw handler synchronously to preserve order
rawHandler(string(line))
err := s.processLine(ctx, line)
if err != nil {
return err
}
// Parse the message first
var wrapper ristypes.RISLiveMessage
if err := json.Unmarshal(line, &wrapper); err != nil {
// Log the error and return to trigger reconnection
s.logger.Error("Failed to parse JSON",
"error", err,
"line", string(line),
"line_length", len(line))
return fmt.Errorf("JSON parse error: %w", err)
}
// Check if it's a ris_message wrapper
if wrapper.Type != "ris_message" {
s.logger.Error("Unexpected wrapper type",
"type", wrapper.Type,
"line", string(line),
)
continue
}
// Get the actual message
msg := wrapper.Data
// Parse the timestamp
msg.ParsedTimestamp = time.Unix(int64(msg.Timestamp), 0).UTC()
// Process based on message type
switch msg.Type {
case "UPDATE":
// Process BGP UPDATE messages
// Will be dispatched to handlers
case "RIS_PEER_STATE":
// RIS peer state messages - silently ignore
continue
case "KEEPALIVE":
// BGP keepalive messages - silently process
continue
case "OPEN":
// BGP open messages - track peer as active
s.bgpPeersMu.Lock()
s.bgpPeers[msg.Peer] = true
peerCount := len(s.bgpPeers)
s.bgpPeersMu.Unlock()
s.metrics.SetBGPPeerCount(peerCount)
s.logger.Info("BGP session opened",
"peer", msg.Peer,
"peer_asn", msg.PeerASN,
"total_peers", peerCount,
)
continue
case "NOTIFICATION":
// BGP notification messages (session closed)
s.bgpPeersMu.Lock()
delete(s.bgpPeers, msg.Peer)
peerCount := len(s.bgpPeers)
s.bgpPeersMu.Unlock()
s.metrics.SetBGPPeerCount(peerCount)
s.logger.Warn("BGP notification",
"peer", msg.Peer,
"peer_asn", msg.PeerASN,
"total_peers", peerCount,
)
continue
case "STATE":
// Peer state changes - silently ignore
continue
default:
s.logger.Warn("Unknown message type, skipping",
"type", msg.Type,
)
continue
}
// Dispatch to interested handlers
s.mu.RLock()
for _, info := range s.handlers {
if !info.handler.WantsMessage(msg.Type) {
continue
}
// Check if we should drop due to backpressure
if s.shouldDropForBackpressure(info) {
atomic.AddUint64(&info.metrics.droppedCount, 1)
atomic.AddUint64(&s.totalDropped, 1)
continue
}
// Try to queue the message
select {
case info.queue <- &msg:
// Message queued successfully
// Update high water mark if needed
queueLen := len(info.queue)
info.metrics.mu.Lock()
if queueLen > info.metrics.queueHighWaterMark {
info.metrics.queueHighWaterMark = queueLen
}
info.metrics.mu.Unlock()
default:
// Queue is full, drop the message
atomic.AddUint64(&info.metrics.droppedCount, 1)
atomic.AddUint64(&s.totalDropped, 1)
}
}
s.mu.RUnlock()
}
if err := scanner.Err(); err != nil {
if ctx.Err() != nil {
return ctx.Err()
}
if connCtx.Err() != nil {
return fmt.Errorf("%w for %s", errStreamIdle, s.idleTimeout)
}
err := scanner.Err()
if err != nil {
return fmt.Errorf("scanner error: %w", err)
}
return errStreamEnded
}
// processLine passes one line of the stream to the raw handler, parses it,
// updates the peer set (updatePeers), and dispatches an UPDATE or STATE
// message to the handlers. An error ends the connection.
func (s *Streamer) processLine(ctx context.Context, line []byte) error {
// Update metrics with decompressed message size
s.updateMetrics(len(line))
// Call raw handler if registered
s.mu.RLock()
rawHandler := s.rawHandler
s.mu.RUnlock()
if rawHandler != nil {
// Call raw handler synchronously to preserve order
rawHandler(string(line))
}
// Parse the message first
var wrapper ristypes.RISLiveMessage
err := json.Unmarshal(line, &wrapper)
if err != nil {
// Log the error and return to trigger reconnection
s.logger.Error("Failed to parse JSON",
"error", err,
"line", string(line),
"line_length", len(line))
return fmt.Errorf("JSON parse error: %w", err)
}
// Check if it's a ris_message wrapper
if wrapper.Type != "ris_message" {
s.logger.Error("Unexpected wrapper type",
"type", wrapper.Type,
"line", string(line),
)
return nil
}
// Get the actual message
msg := wrapper.Data
// Parse the timestamp
msg.ParsedTimestamp = time.Unix(int64(msg.Timestamp), 0).UTC()
peerCount := s.updatePeers(&msg)
// Process based on message type
switch msg.Type {
case ristypes.MessageTypeUpdate:
// Process BGP UPDATE messages
// Will be dispatched to handlers
case "RIS_PEER_STATE":
// RIS peer state messages - silently ignore
return nil
case "KEEPALIVE":
// BGP keepalive messages - silently process
return nil
case "OPEN":
s.logger.Info("BGP session opened",
"peer", msg.Peer,
"peer_asn", msg.PeerASN,
"total_peers", peerCount,
)
return nil
case messageTypeNotification:
s.logger.Warn("BGP notification",
"peer", msg.Peer,
"peer_asn", msg.PeerASN,
"total_peers", peerCount,
)
return nil
case ristypes.MessageTypeState:
// The BGP session with the peer changed state, for example went
// down. Dispatched to the handlers that want it.
default:
s.logger.Warn("Unknown message type, skipping",
"type", msg.Type,
)
return nil
}
return s.dispatchMessage(ctx, &msg)
}
// updatePeers adds the peer msg came from to the peer set, or removes it if
// msg ends the peer's session: a NOTIFICATION, or a STATE message whose state
// is down. It records the size of the set as the BGP peer count and returns
// it.
func (s *Streamer) updatePeers(msg *ristypes.RISMessage) int {
sessionEnded := msg.Type == messageTypeNotification ||
(msg.Type == ristypes.MessageTypeState && msg.State == ristypes.StateDown)
s.bgpPeersMu.Lock()
if sessionEnded {
delete(s.bgpPeers, msg.Peer)
} else {
s.bgpPeers[msg.Peer] = true
}
peerCount := len(s.bgpPeers)
s.bgpPeersMu.Unlock()
s.metrics.SetBGPPeerCount(peerCount)
return peerCount
}
// dispatchMessage queues msg for each handler that wants it, unless
// queueMessage drops it for that handler.
func (s *Streamer) dispatchMessage(
ctx context.Context, msg *ristypes.RISMessage,
) error {
// Stop cancels ctx and closes the queues under the write lock, so if ctx
// is cancelled here, under the read lock, the queues are closed and must
// not be sent to.
s.mu.RLock()
if ctx.Err() != nil {
s.mu.RUnlock()
return ctx.Err()
}
for _, info := range s.handlers {
if !info.handler.WantsMessage(msg.Type) {
continue
}
if !s.queueMessage(info, msg) {
atomic.AddUint64(&info.metrics.droppedCount, 1)
s.totalDropped.Add(1)
continue
}
// Update high water mark if needed
queueLen := len(info.queue)
info.metrics.mu.Lock()
if queueLen > info.metrics.queueHighWaterMark {
info.metrics.queueHighWaterMark = queueLen
}
info.metrics.mu.Unlock()
}
s.mu.RUnlock()
return nil
}
// shouldDropForBackpressure determines if a message should be dropped based on queue utilization
// queueMessage puts msg on the handler's queue and reports false if it dropped
// msg instead. It never waits for room, so a full queue does not hold up the
// read loop.
//
// A message is dropped while the queue is filling up (shouldDropForBackpressure)
// or full, except a STATE message: one whose state is down is the only notice
// that all of a peer's routes are gone. A STATE message skips the random drop,
// and one that finds the queue full is held in info.heldState. Held messages
// go into the queue, oldest first, before anything else for the handler, as
// soon as there is room, or when Stop is called; while any is still held, the
// handler's other messages are dropped, so none of them overtakes a held down.
//
// The held list stays small: a STATE message is a few hundred bytes and the
// feed sends about ten a second, so a queue full for a minute holds about 600.
func (s *Streamer) queueMessage(
info *handlerInfo, msg *ristypes.RISMessage,
) bool {
for len(info.heldState) > 0 && trySend(info.queue, info.heldState[0]) {
info.heldState = info.heldState[1:]
}
if msg.Type == ristypes.MessageTypeState {
if len(info.heldState) > 0 || !trySend(info.queue, msg) {
info.heldState = append(info.heldState, msg)
}
return true
}
if len(info.heldState) > 0 || s.shouldDropForBackpressure(info) {
return false
}
return trySend(info.queue, msg)
}
// trySend puts msg on queue if there is room and reports whether it did.
func trySend(queue chan *ristypes.RISMessage, msg *ristypes.RISMessage) bool {
select {
case queue <- msg:
return true
default:
return false
}
}
// shouldDropForBackpressure determines if a message should be dropped based on
// queue utilization
func (s *Streamer) shouldDropForBackpressure(info *handlerInfo) bool {
// Calculate queue utilization
queueLen := len(info.queue)
internal/streamer/streamer_internal_test.go
+788
View File
@@ -0,0 +1,788 @@
package streamer
import (
"bytes"
"context"
"encoding/json"
"errors"
"io"
"log/slog"
"net/http"
"net/http/httptest"
"runtime"
"strings"
"sync/atomic"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/metrics"
"git.eeqj.de/sneak/routewatch/internal/ristypes"
)
func TestNewStreamer(t *testing.T) {
t.Parallel()
logger := logger.New()
metricsTracker := metrics.New()
s := New(logger, metricsTracker)
if s == nil {
t.Fatal("New() returned nil")
}
if s.logger != logger {
t.Error("logger not set correctly")
}
if s.client == nil {
t.Error("HTTP client not initialized")
}
if s.handlers == nil {
t.Error("handlers slice not initialized")
}
if s.metrics != metricsTracker {
t.Error("metrics tracker not set correctly")
}
if s.idleTimeout != streamIdleTimeout {
t.Errorf("idleTimeout = %s, want %s", s.idleTimeout, streamIdleTimeout)
}
if s.minBackoff != minBackoffDelay {
t.Errorf("minBackoff = %s, want %s", s.minBackoff, minBackoffDelay)
}
}
// TestStreamDoesNotLeakTickersAcrossReconnects drives many short-lived
// connections (each stream call is one reconnect cycle) and asserts the
// goroutine count returns to its starting value. Each connection starts two
// ticker goroutines; before the fix they lived until the streamer's lifetime
// context was cancelled, so every reconnect leaked two.
//
//nolint:paralleltest // counts goroutines, so no other test may run alongside
func TestStreamDoesNotLeakTickersAcrossReconnects(t *testing.T) {
// The handler returns immediately, so the response body is empty and each
// stream call ends at once, standing in for a dropped connection.
srv := httptest.NewServer(http.HandlerFunc(
func(_ http.ResponseWriter, _ *http.Request) {}))
defer srv.Close()
s := New(logger.New(), metrics.New())
s.url = srv.URL
// One warm-up connection so any persistent HTTP transport goroutine exists
// before we take the baseline.
err := s.stream(context.Background())
if !errors.Is(err, errStreamEnded) {
t.Fatalf("warm-up stream returned %v, want %v", err, errStreamEnded)
}
s.client.CloseIdleConnections()
baseline := settledGoroutineCount()
const reconnects = 20
for range reconnects {
err := s.stream(context.Background())
if !errors.Is(err, errStreamEnded) {
t.Fatalf("stream returned %v, want %v", err, errStreamEnded)
}
}
s.client.CloseIdleConnections()
if !waitForGoroutines(baseline) {
t.Fatalf("goroutines did not return to baseline %d after %d reconnects, got %d",
baseline, reconnects, runtime.NumGoroutine())
}
}
// updateLine is one UPDATE message as RIS Live sends it.
const updateLine = `{"type":"ris_message","data":{"type":"UPDATE",` +
`"peer":"192.0.2.1","peer_asn":"64496","timestamp":1700000000}}` + "\n"
// updateHandler wants UPDATE messages. For each one it handles it sends on
// handled if there is room, and otherwise does nothing; a nil handled is
// never sent on.
type updateHandler struct {
handled chan struct{}
}
func (updateHandler) WantsMessage(messageType string) bool {
return messageType == "UPDATE"
}
func (updateHandler) QueueCapacity() int { return 10 }
func (h updateHandler) HandleMessage(*ristypes.RISMessage) {
select {
case h.handled <- struct{}{}:
default:
}
}
// flush sends what w has buffered to the client. The writers httptest servers
// hand to handlers can always do that.
func flush(w http.ResponseWriter) {
flusher, ok := w.(http.Flusher)
if ok {
flusher.Flush()
}
}
// stateDownLine is the STATE message RIS Live sends when the collector's BGP
// session with the peer stateDownPeer goes down.
const (
stateDownPeer = "192.0.2.1"
stateDownLine = `{"type":"ris_message","data":{"type":"STATE",` +
`"peer":"` + stateDownPeer + `",` +
`"peer_asn":"64496","timestamp":1700000000,"state":"down"}}` + "\n"
)
// stateHandler wants STATE messages only. Tests read what it is handed from
// its queue, so HandleMessage does nothing.
type stateHandler struct{}
func (stateHandler) WantsMessage(messageType string) bool {
return messageType == ristypes.MessageTypeState
}
func (stateHandler) QueueCapacity() int { return 10 }
func (stateHandler) HandleMessage(*ristypes.RISMessage) {}
// TestStreamHandsStateMessagesToHandlersThatWantThem has the server send one
// STATE message whose state is down. The streamer must queue it, with its
// peer and state, for a handler that wants STATE messages.
func TestStreamHandsStateMessagesToHandlersThatWantThem(t *testing.T) {
t.Parallel()
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, _ *http.Request) {
_, _ = io.WriteString(w, stateDownLine)
}))
defer srv.Close()
s := New(logger.New(), metrics.New())
s.url = srv.URL
s.RegisterHandler(stateHandler{})
err := s.stream(context.Background())
if !errors.Is(err, errStreamEnded) {
t.Fatalf("stream returned %v, want %v", err, errStreamEnded)
}
select {
case msg := <-s.handlers[0].queue:
if msg.Type != ristypes.MessageTypeState || msg.Peer != stateDownPeer ||
msg.State != ristypes.StateDown {
t.Errorf("queued message has type %q, peer %q, state %q; "+
"want %q, %q, %q", msg.Type, msg.Peer, msg.State,
ristypes.MessageTypeState, stateDownPeer, ristypes.StateDown)
}
default:
t.Fatal("the STATE message was not queued for the handler that wants it")
}
}
// TestHeldStateMessagesKeepTheirPlaceInTheQueue hands queueMessage two STATE
// messages and two updates while the queue is full, making room in between.
// Neither STATE message may be dropped: each must reach the queue after
// everything queued before it and before anything handed over after it. The
// update handed over while a STATE message is held is dropped.
func TestHeldStateMessagesKeepTheirPlaceInTheQueue(t *testing.T) {
t.Parallel()
s := New(logger.New(), metrics.New())
s.RegisterHandler(stateHandler{})
info := s.handlers[0]
before := &ristypes.RISMessage{Type: ristypes.MessageTypeUpdate}
for len(info.queue) < cap(info.queue) {
info.queue <- before
}
firstDown := &ristypes.RISMessage{
Type: ristypes.MessageTypeState, State: ristypes.StateDown,
}
secondDown := &ristypes.RISMessage{
Type: ristypes.MessageTypeState, State: ristypes.StateDown,
}
updateWhileHeld := &ristypes.RISMessage{Type: ristypes.MessageTypeUpdate}
after := &ristypes.RISMessage{Type: ristypes.MessageTypeUpdate}
if !s.queueMessage(info, firstDown) {
t.Fatal("the first STATE message was dropped on a full queue")
}
<-info.queue // the handler makes room for one message
if !s.queueMessage(info, secondDown) {
t.Fatal("the second STATE message was dropped")
}
if s.queueMessage(info, updateWhileHeld) {
t.Fatal("an update was queued while a STATE message was held")
}
// The handler takes every message queued before the first STATE message.
for range cap(info.queue) - 1 {
if <-info.queue != before {
t.Fatal("a message overtook those queued before the first STATE message")
}
}
if !s.queueMessage(info, after) {
t.Fatal("an update was dropped on a queue with room")
}
if <-info.queue != firstDown {
t.Fatal("the first STATE message did not come next")
}
if <-info.queue != secondDown {
t.Fatal("the second STATE message did not come after the first")
}
if <-info.queue != after {
t.Fatal("the update handed over after the STATE messages did not come " +
"after them")
}
if len(info.queue) != 0 {
t.Errorf("%d messages left on the queue, want none", len(info.queue))
}
}
// takeAll takes the messages off queue, as a handler's worker would, and sends
// them on the returned channel once queue is closed.
func takeAll(queue chan *ristypes.RISMessage) chan []*ristypes.RISMessage {
taken := make(chan []*ristypes.RISMessage, 1)
go func() {
var msgs []*ristypes.RISMessage
for msg := range queue {
msgs = append(msgs, msg)
}
taken <- msgs
}()
return taken
}
// newStreamerWithFullStateQueue returns a streamer reading from url with one
// handler, a stateHandler whose queue is full.
func newStreamerWithFullStateQueue(url string) *Streamer {
s := New(logger.New(), metrics.New())
s.url = url
s.RegisterHandler(stateHandler{})
queue := s.handlers[0].queue
for len(queue) < cap(queue) {
queue <- &ristypes.RISMessage{Type: ristypes.MessageTypeState}
}
return s
}
// TestOtherHandlersKeepReceivingWhileAStateMessageIsHeld has the server send a
// STATE message whose state is down, then an UPDATE, while the queue of the
// handler that wants STATE messages is full. The STATE message must be held
// for that handler, and the read loop must go on to queue the UPDATE for the
// handler that wants it.
func TestOtherHandlersKeepReceivingWhileAStateMessageIsHeld(t *testing.T) {
t.Parallel()
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, _ *http.Request) {
_, _ = io.WriteString(w, stateDownLine+updateLine)
}))
defer srv.Close()
s := newStreamerWithFullStateQueue(srv.URL)
s.RegisterHandler(updateHandler{})
err := s.stream(context.Background())
if !errors.Is(err, errStreamEnded) {
t.Fatalf("stream returned %v, want %v", err, errStreamEnded)
}
held := s.handlers[0].heldState
if len(held) != 1 || held[0].Peer != stateDownPeer ||
held[0].State != ristypes.StateDown {
t.Errorf("held %d STATE messages, want the down from %s alone",
len(held), stateDownPeer)
}
got := len(s.handlers[1].queue)
if got != 1 {
t.Errorf("%d messages queued for the handler that wants UPDATE "+
"messages, want 1", got)
}
}
// TestStopQueuesTheHeldStateMessages has the server send two STATE messages
// and an UPDATE and hold the response open while the queue of the handler that
// wants STATE messages is full, so both are held. Stop must put both on the
// queue, after the messages already on it, before it closes the queue, and
// must end the stream.
func TestStopQueuesTheHeldStateMessages(t *testing.T) {
t.Parallel()
const (
held = 2
lines = held + 1
maxWait = 5 * time.Second
)
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, r *http.Request) {
_, _ = io.WriteString(w, stateDownLine+stateDownLine+updateLine)
flush(w)
<-r.Context().Done()
}))
defer srv.Close()
s := newStreamerWithFullStateQueue(srv.URL)
// The raw handler sees each line before the read loop hands it to the
// queues, so once it has seen the UPDATE, both STATE messages are held.
linesSeen := make(chan struct{}, lines)
s.RegisterRawHandler(func(string) { linesSeen <- struct{}{} })
// As in TestStopBeforeMessageReachesQueues, setting cancel as Start does
// lets Stop cancel the stream run here.
ctx, cancel := context.WithCancel(context.Background())
s.cancel = cancel
streamErr := make(chan error, 1)
go func() { streamErr <- s.stream(ctx) }()
for range lines {
select {
case <-linesSeen:
case <-time.After(maxWait):
t.Fatalf("the read loop did not read %d lines within %s",
lines, maxWait)
}
}
// No worker runs here: the test takes from the full queue in its place.
queue := s.handlers[0].queue
taken := takeAll(queue)
stopped := make(chan struct{})
go func() {
s.Stop(t.Context())
close(stopped)
}()
select {
case <-stopped:
case <-time.After(maxWait):
t.Fatalf("Stop did not return within %s while a STATE message was held",
maxWait)
}
// Stop has closed the queue, so takeAll is done or about to be.
msgs := <-taken
if len(msgs) != cap(queue)+held {
t.Fatalf("%d messages came off the queue, want the %d on it and the "+
"%d held", len(msgs), cap(queue), held)
}
for _, msg := range msgs[cap(queue):] {
if msg.Peer != stateDownPeer || msg.State != ristypes.StateDown {
t.Errorf("message after those on the queue has peer %q, state %q; "+
"want the down from %s", msg.Peer, msg.State, stateDownPeer)
}
}
select {
case err := <-streamErr:
if !errors.Is(err, context.Canceled) {
t.Errorf("stream returned %v, want %v", err, context.Canceled)
}
case <-time.After(maxWait):
t.Fatalf("stream did not end within %s of Stop", maxWait)
}
}
// TestStopReturnsByItsDeadline stops a streamer whose handler has a full queue
// and a STATE message held for it, and a worker that stays busy until the test
// ends, as one does whose handler waits for a database write. Stop must return
// within its deadline plus a small margin instead of waiting for room on the
// queue and for the worker, and log as not handed over the held message and
// every message on the queue.
func TestStopReturnsByItsDeadline(t *testing.T) {
t.Parallel()
const (
deadline = 100 * time.Millisecond
margin = 2 * time.Second
)
var logs bytes.Buffer
s := newStreamerWithFullStateQueue("")
s.logger = &logger.Logger{Logger: slog.New(slog.NewJSONHandler(&logs,
&slog.HandlerOptions{Level: slog.LevelError}))}
s.handlers[0].heldState = []*ristypes.RISMessage{
{Type: ristypes.MessageTypeState},
}
wantNotHandedOver := 1 + len(s.handlers[0].queue)
// Stop cancels the stream Start would have started; there is none here.
s.cancel = func() {}
release := make(chan struct{})
defer close(release)
s.workers.Go(func() { <-release })
ctx, cancel := context.WithTimeout(t.Context(), deadline)
defer cancel()
stopped := make(chan struct{})
go func() {
s.Stop(ctx)
close(stopped)
}()
select {
case <-stopped:
case <-time.After(deadline + margin):
t.Fatalf("Stop did not return within %s of its deadline", margin)
}
var logged struct {
Messages int `json:"messages"`
}
err := json.Unmarshal(logs.Bytes(), &logged)
if err != nil {
t.Fatalf("want the error Stop logs, got %q: %v", logs.String(), err)
}
if logged.Messages != wantNotHandedOver {
t.Errorf("Stop logged %d messages not handed over, want %d",
logged.Messages, wantNotHandedOver)
}
}
// TestStopBeforeMessageReachesQueues stops the streamer after the read loop
// has checked for cancellation but before it hands the message to the handler
// queues. That is the gap a stop from another goroutine can land in, and it
// used to end in "send on closed channel". The raw handler runs in that gap on
// the read loop itself, so calling Stop from it hits the gap every time.
func TestStopBeforeMessageReachesQueues(t *testing.T) {
t.Parallel()
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, _ *http.Request) {
_, _ = io.WriteString(w, updateLine)
}))
defer srv.Close()
s := New(logger.New(), metrics.New())
s.url = srv.URL
s.RegisterHandler(updateHandler{})
s.RegisterRawHandler(func(string) { s.Stop(t.Context()) })
// Start would run the stream in the background, where the test cannot
// wait for it. Setting cancel as Start does lets Stop cancel the stream
// run here instead.
ctx, cancel := context.WithCancel(context.Background())
s.cancel = cancel
err := s.stream(ctx)
if !errors.Is(err, context.Canceled) {
t.Fatalf("stream returned %v, want %v", err, context.Canceled)
}
// A second Stop must not close the queues again.
s.Stop(t.Context())
}
// TestReconnectWhenStreamEnds has the server end the first response
// normally, as when RIS Live restarts. The streamer must treat that as a
// disconnect and connect again.
func TestReconnectWhenStreamEnds(t *testing.T) {
t.Parallel()
endResponse := func(http.ResponseWriter, *http.Request) {}
checkReconnect(t, endResponse, streamIdleTimeout)
}
// TestReconnectWhenStreamGoesIdle has the server hold the first response
// open without sending anything. The streamer must drop it once the idle
// timeout passes and connect again.
func TestReconnectWhenStreamGoesIdle(t *testing.T) {
t.Parallel()
const idleTimeout = 50 * time.Millisecond
holdResponse := func(w http.ResponseWriter, r *http.Request) {
flush(w)
<-r.Context().Done()
}
checkReconnect(t, holdResponse, idleTimeout)
}
// TestReconnectWhenServerNeverAnswers has the server accept the first
// request and never answer it, not even with headers. The streamer must give
// up on it once the idle timeout passes and connect again.
func TestReconnectWhenServerNeverAnswers(t *testing.T) {
t.Parallel()
const idleTimeout = 50 * time.Millisecond
neverAnswer := func(_ http.ResponseWriter, r *http.Request) {
<-r.Context().Done()
}
checkReconnect(t, neverAnswer, idleTimeout)
}
// TestStreamNotDroppedWhileLinesArrive has the server send lines well within
// the idle timeout for several times the timeout, then end the response. Each
// line must start the idle timeout over, so the stream lasts until the server
// ends it instead of being dropped as idle.
func TestStreamNotDroppedWhileLinesArrive(t *testing.T) {
t.Parallel()
const (
idleTimeout = 100 * time.Millisecond
lineInterval = idleTimeout / 10
sendFor = 5 * idleTimeout
)
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, _ *http.Request) {
deadline := time.Now().Add(sendFor)
for time.Now().Before(deadline) {
_, _ = io.WriteString(w, updateLine)
flush(w)
time.Sleep(lineInterval)
}
}))
defer srv.Close()
s := New(logger.New(), metrics.New())
s.url = srv.URL
s.idleTimeout = idleTimeout
err := s.stream(context.Background())
if !errors.Is(err, errStreamEnded) {
t.Fatalf("stream returned %v, want %v", err, errStreamEnded)
}
}
// TestStreamReadsLongLine has the server send one UPDATE line of about
// 200 KiB, padded with raw hex the way RIS Live writes an extended BGP
// message, and then end the response. The line is longer than a
// bufio.Scanner reads by default. The streamer must hand the message to the
// handler's queue and read on until the server ends the stream, instead of
// dropping the connection at the long line.
func TestStreamReadsLongLine(t *testing.T) {
t.Parallel()
const rawHexLength = 200 * bytesPerKB
longLine := `{"type":"ris_message","data":{"type":"UPDATE","peer":"192.0.2.1",` +
`"peer_asn":"64496","timestamp":1700000000,"raw":"` +
strings.Repeat("0", rawHexLength) + `"}}` + "\n"
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, _ *http.Request) {
_, _ = io.WriteString(w, longLine)
}))
defer srv.Close()
s := New(logger.New(), metrics.New())
s.url = srv.URL
s.RegisterHandler(updateHandler{})
err := s.stream(context.Background())
if !errors.Is(err, errStreamEnded) {
t.Fatalf("stream returned %v, want %v", err, errStreamEnded)
}
queued := len(s.handlers[0].queue)
if queued != 1 {
t.Errorf("handler queue holds %d messages, want 1", queued)
}
}
// TestPeerCountCountsEveryPeerHeardFrom has the server send an UPDATE from
// stateDownPeer and a KEEPALIVE from another peer, with no OPEN from either,
// and then a message that ends stateDownPeer's session: a NOTIFICATION, or a
// STATE message whose state is down. The peer count must be 2 after the first
// two messages and 1 after the third.
func TestPeerCountCountsEveryPeerHeardFrom(t *testing.T) {
t.Parallel()
const (
otherPeer = "192.0.2.2"
peerUpdateLine = `{"type":"ris_message","data":{"type":"UPDATE",` +
`"peer":"` + stateDownPeer + `",` +
`"peer_asn":"64496","timestamp":1700000000}}` + "\n"
otherPeerKeepaliveLine = `{"type":"ris_message","data":{"type":"KEEPALIVE",` +
`"peer":"` + otherPeer + `",` +
`"peer_asn":"64497","timestamp":1700000000}}` + "\n"
peerNotificationLine = `{"type":"ris_message","data":{"type":"NOTIFICATION",` +
`"peer":"` + stateDownPeer + `",` +
`"peer_asn":"64496","timestamp":1700000000}}` + "\n"
bothPeers = 2
)
sessionEndLines := map[string]string{
"NOTIFICATION": peerNotificationLine,
"STATE down": stateDownLine,
}
for name, sessionEndLine := range sessionEndLines {
t.Run(name, func(t *testing.T) {
t.Parallel()
tracker := metrics.New()
s := New(logger.New(), tracker)
streamLines(t, s, peerUpdateLine+otherPeerKeepaliveLine)
got := tracker.GetBGPPeerCount()
if got != bothPeers {
t.Fatalf("peer count is %d after messages from two peers, want %d",
got, bothPeers)
}
streamLines(t, s, sessionEndLine)
got = tracker.GetBGPPeerCount()
if got != 1 {
t.Errorf("peer count is %d after one of two peers' sessions "+
"ended, want 1", got)
}
})
}
}
// streamLines has a server send lines as its whole response and runs one
// stream of s against it.
func streamLines(t *testing.T, s *Streamer, lines string) {
t.Helper()
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, _ *http.Request) {
_, _ = io.WriteString(w, lines)
}))
defer srv.Close()
s.url = srv.URL
err := s.stream(context.Background())
if !errors.Is(err, errStreamEnded) {
t.Fatalf("stream returned %v, want %v", err, errStreamEnded)
}
}
// checkReconnect runs a streamer against a fake RIS Live server. The first
// response is served by firstResponse; every later one carries one UPDATE
// line and then stays open. The streamer must come back for a second
// response, must not be marked connected when it does, and must handle the
// UPDATE in it.
func checkReconnect(
t *testing.T, firstResponse http.HandlerFunc, idleTimeout time.Duration,
) {
t.Helper()
const maxWait = 5 * time.Second
tracker := metrics.New()
connectedAtReconnect := make(chan bool, 1)
var requests atomic.Int32
srv := httptest.NewServer(http.HandlerFunc(
func(w http.ResponseWriter, r *http.Request) {
if requests.Add(1) == 1 {
firstResponse(w, r)
return
}
select {
case connectedAtReconnect <- tracker.IsConnected():
default:
}
_, _ = io.WriteString(w, updateLine)
flush(w)
<-r.Context().Done()
}))
defer srv.Close()
s := New(logger.New(), tracker)
s.url = srv.URL
s.idleTimeout = idleTimeout
s.minBackoff = time.Millisecond
handler := updateHandler{handled: make(chan struct{}, 1)}
s.RegisterHandler(handler)
err := s.Start(t.Context())
if err != nil {
t.Fatalf("Start returned error: %v", err)
}
defer s.Stop(t.Context())
select {
case <-handler.handled:
case <-time.After(maxWait):
t.Fatalf("no message from a second response handled within %s", maxWait)
}
if <-connectedAtReconnect {
t.Error("streamer was still marked connected when it came back " +
"for a second response")
}
}
// settledGoroutineCount lets transient goroutines finish, then reports the
// current count.
func settledGoroutineCount() int {
prev := runtime.NumGoroutine()
for range 20 {
time.Sleep(10 * time.Millisecond)
cur := runtime.NumGoroutine()
if cur == prev {
return cur
}
prev = cur
}
return prev
}
// waitForGoroutines waits until the goroutine count drops to target or below.
func waitForGoroutines(target int) bool {
for range 100 {
if runtime.NumGoroutine() <= target {
return true
}
time.Sleep(10 * time.Millisecond)
}
return false
}
internal/streamer/streamer_test.go
-104
View File
@@ -1,104 +0,0 @@
package streamer
import (
"context"
"net/http"
"net/http/httptest"
"runtime"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/logger"
"git.eeqj.de/sneak/routewatch/internal/metrics"
)
func TestNewStreamer(t *testing.T) {
logger := logger.New()
metricsTracker := metrics.New()
s := New(logger, metricsTracker)
if s == nil {
t.Fatal("New() returned nil")
}
if s.logger != logger {
t.Error("logger not set correctly")
}
if s.client == nil {
t.Error("HTTP client not initialized")
}
if s.handlers == nil {
t.Error("handlers slice not initialized")
}
if s.metrics != metricsTracker {
t.Error("metrics tracker not set correctly")
}
}
// TestStreamDoesNotLeakTickersAcrossReconnects drives many short-lived
// connections (each stream call is one reconnect cycle) and asserts the
// goroutine count returns to its starting value. Each connection starts two
// ticker goroutines; before the fix they lived until the streamer's lifetime
// context was cancelled, so every reconnect leaked two.
func TestStreamDoesNotLeakTickersAcrossReconnects(t *testing.T) {
// The handler returns immediately, so the response body is empty and each
// stream call ends at once, standing in for a dropped connection.
srv := httptest.NewServer(http.HandlerFunc(func(_ http.ResponseWriter, _ *http.Request) {}))
defer srv.Close()
s := New(logger.New(), metrics.New())
s.url = srv.URL
// One warm-up connection so any persistent HTTP transport goroutine exists
// before we take the baseline.
if err := s.stream(context.Background()); err != nil {
t.Fatalf("warm-up stream returned error: %v", err)
}
s.client.CloseIdleConnections()
baseline := settledGoroutineCount()
const reconnects = 20
for range reconnects {
if err := s.stream(context.Background()); err != nil {
t.Fatalf("stream returned error: %v", err)
}
}
s.client.CloseIdleConnections()
if !waitForGoroutines(baseline) {
t.Fatalf("goroutines did not return to baseline %d after %d reconnects, got %d",
baseline, reconnects, runtime.NumGoroutine())
}
}
// settledGoroutineCount lets transient goroutines finish, then reports the
// current count.
func settledGoroutineCount() int {
prev := runtime.NumGoroutine()
for range 20 {
time.Sleep(10 * time.Millisecond)
cur := runtime.NumGoroutine()
if cur == prev {
return cur
}
prev = cur
}
return prev
}
// waitForGoroutines waits until the goroutine count drops to target or below.
func waitForGoroutines(target int) bool {
for range 100 {
if runtime.NumGoroutine() <= target {
return true
}
time.Sleep(10 * time.Millisecond)
}
return false
}
internal/templates/as_detail.html
+1 -1
View File
@@ -4,7 +4,7 @@
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>AS{{.ASN.ASN}} - {{.ASN.Handle}} - RouteWatch</title>
<style>
<style nonce="{{.CSPNonce}}">
* {
box-sizing: border-box;
}
internal/templates/index.html
+11 -4
View File
@@ -4,7 +4,7 @@
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>RouteWatch - BGP Route Monitor</title>
<style>
<style nonce="{{.CSPNonce}}">
* {
box-sizing: border-box;
}
@@ -167,6 +167,13 @@
color: #95a5a6;
margin-top: 8px;
}
#asname-results p {
color: #7f8c8d;
margin-top: 12px;
}
#asname-results .search-message {
font-size: 13px;
}
/* IP Lookup result */
.ip-result {
@@ -329,7 +336,7 @@
<span><a href="{{appGitCommitURL}}">{{appGitRevision}}</a></span>
</footer>
<script>
<script nonce="{{.CSPNonce}}">
function formatNumber(num) {
if (num >= 1000000) {
return (num / 1000000).toFixed(1) + 'M';
@@ -387,7 +394,7 @@
if (!query) return;
const resultsDiv = document.getElementById('asname-results');
resultsDiv.innerHTML = '<p class="loading" style="color: #7f8c8d; margin-top: 12px;">Searching...</p>';
resultsDiv.innerHTML = '<p class="loading">Searching...</p>';
// Use a simple client-side search against the asinfo data
// For now, redirect to AS page if it looks like an ASN
@@ -397,7 +404,7 @@
}
// Show a message that server-side search is coming
resultsDiv.innerHTML = '<p style="color: #7f8c8d; margin-top: 12px; font-size: 13px;">AS name search coming soon. For now, try an AS number.</p>';
resultsDiv.innerHTML = '<p class="search-message">AS name search coming soon. For now, try an AS number.</p>';
});
// IP lookup
internal/templates/prefix_detail.html
+15 -2
View File
@@ -4,7 +4,7 @@
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>{{.Prefix}} - RouteWatch</title>
<style>
<style nonce="{{.CSPNonce}}">
* {
box-sizing: border-box;
}
@@ -221,6 +221,10 @@
border-radius: 4px;
border: 1px solid #e0e0e0;
}
.origin-item .peer-count {
color: #7f8c8d;
margin-left: 10px;
}
.empty-state {
text-align: center;
padding: 40px;
@@ -281,9 +285,14 @@
<div class="origin-list">
{{range .Origins}}
<div class="origin-item">
{{if .ASN}}
<a href="/as/{{.ASN}}" class="as-link">AS{{.ASN}}</a>
{{if .Handle}} ({{.Handle}}){{end}}
<span style="color: #7f8c8d; margin-left: 10px;">{{.PeerCount}} peer{{if ne .PeerCount 1}}s{{end}}</span>
{{if .Description}} {{.Description}}{{end}}
{{else}}
No known origin
{{end}}
<span class="peer-count">{{.PeerCount}} peer{{if ne .PeerCount 1}}s{{end}}</span>
</div>
{{end}}
</div>
@@ -311,7 +320,11 @@
{{range .Routes}}
<tr>
<td>
{{if .OriginASN}}
<a href="/as/{{.OriginASN}}" class="as-link">AS{{.OriginASN}}</a>
{{else}}
No known origin
{{end}}
</td>
<td class="peer-ip">{{.PeerIP}}</td>
<td class="as-path">{{range $i, $as := .ASPathWithHandle}}{{if $i}} → {{end}}<a href="/as/{{$as.ASN}}" class="as-link">{{if $as.Handle}}{{$as.Handle}}{{else}}AS{{$as.ASN}}{{end}}</a>{{end}}</td>
internal/templates/prefix_length.html
+5 -1
View File
@@ -4,7 +4,7 @@
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Prefixes with /{{ .MaskLength }} - RouteWatch</title>
<style>
<style nonce="{{.CSPNonce}}">
* {
box-sizing: border-box;
}
@@ -172,9 +172,13 @@
<td><a href="{{ .Prefix | prefixURL }}" class="prefix-link">{{ .Prefix }}</a></td>
<td class="age">{{ .Age }}</td>
<td>
{{ if .OriginASN }}
<a href="/as/{{ .OriginASN }}" class="as-link">
AS{{ .OriginASN }}{{ if .OriginASDescription }} ({{ .OriginASDescription }}){{ end }}
</a>
{{ else }}
No known origin
{{ end }}
</td>
</tr>
{{ end }}
internal/templates/status.html
+5 -4
View File
@@ -4,7 +4,7 @@
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>RouteWatch Status</title>
<style>
<style nonce="{{.CSPNonce}}">
* {
box-sizing: border-box;
}
@@ -130,6 +130,7 @@
color: #ef4444;
}
.error {
display: none;
background: #fee;
color: #c00;
padding: 10px;
@@ -173,7 +174,7 @@
</nav>
<main class="main-content">
<div id="error" class="error" style="display: none;"></div>
<div id="error" class="error"></div>
<div class="status-grid">
<div class="status-card">
<h2>RouteWatch</h2>
@@ -383,8 +384,8 @@
<div id="handler-stats-container" class="status-grid">
<!-- Handler stats will be dynamically added here -->
</div>
<script>
<script nonce="{{.CSPNonce}}">
function formatBytes(bytes) {
if (bytes === 0) return '0 B';
const k = 1024;
internal/templates/templates.go
+11 -4
View File
@@ -60,6 +60,7 @@ func timeSince(t time.Time) string {
if duration < time.Minute {
return "just now"
}
if duration < time.Hour {
minutes := int(duration.Minutes())
if minutes == 1 {
@@ -68,6 +69,7 @@ func timeSince(t time.Time) string {
return duration.Truncate(time.Minute).String() + " ago"
}
if duration < hoursPerDay*time.Hour {
hours := int(duration.Hours())
if hours == 1 {
@@ -76,10 +78,12 @@ func timeSince(t time.Time) string {
return duration.Truncate(time.Hour).String() + " ago"
}
days := int(duration.Hours() / hoursPerDay)
if days == 1 {
return "1 day ago"
}
if days < daysPerMonth {
return duration.Truncate(hoursPerDay*time.Hour).String() + " ago"
}
@@ -118,7 +122,7 @@ func initTemplates() {
"appLicense": func() string { return version.License },
"appRepoURL": func() string { return version.RepoURL },
"appGitRevision": func() string { return version.GitRevisionShort },
"appGitCommitURL": func() string { return version.CommitURL() },
"appGitCommitURL": version.CommitURL,
}
// Parse index template
@@ -134,19 +138,22 @@ func initTemplates() {
}
// Parse AS detail template
defaultTemplates.ASDetail, err = template.New("asDetail").Funcs(funcs).Parse(asDetailHTML)
defaultTemplates.ASDetail, err = template.New("asDetail").
Funcs(funcs).Parse(asDetailHTML)
if err != nil {
panic("failed to parse AS detail template: " + err.Error())
}
// Parse prefix detail template
defaultTemplates.PrefixDetail, err = template.New("prefixDetail").Funcs(funcs).Parse(prefixDetailHTML)
defaultTemplates.PrefixDetail, err = template.New("prefixDetail").
Funcs(funcs).Parse(prefixDetailHTML)
if err != nil {
panic("failed to parse prefix detail template: " + err.Error())
}
// Parse prefix length template
defaultTemplates.PrefixLength, err = template.New("prefixLength").Funcs(funcs).Parse(prefixLengthHTML)
defaultTemplates.PrefixLength, err = template.New("prefixLength").
Funcs(funcs).Parse(prefixLengthHTML)
if err != nil {
panic("failed to parse prefix length template: " + err.Error())
}
internal/version/version.go
+3 -2
View File
@@ -3,11 +3,12 @@ package version
// Build-time variables set via ldflags
//
//nolint:gochecknoglobals // These must be variables to allow ldflags injection at build time
//nolint:gochecknoglobals // variables so that ldflags can set them at build time
var (
// GitRevision is the git commit hash
GitRevision = "unknown"
// GitRevisionShort is the short git commit hash (7 chars)
// GitRevisionShort is the version the page footer shows: the tag or
// short commit hash from `git describe --tags --always`
GitRevisionShort = "unknown"
)
internal/whois/whois.go
+166 -118
View File
@@ -4,7 +4,9 @@ package whois
import (
"bufio"
"context"
"errors"
"fmt"
"io"
"net"
"regexp"
"strings"
@@ -18,6 +20,10 @@ const (
writeTimeout = 5 * time.Second
)
// maxAnswerSize caps the bytes read of one WHOIS answer: a real one is a
// few KiB, so one this large is from a server that does not stop sending.
const maxAnswerSize = 1 << 20
// Parsing constants.
const (
keyValueParts = 2 // Expected parts when splitting "key: value"
@@ -43,6 +49,19 @@ const (
RIRAFRNIC = "AFRINIC"
)
// Errors LookupASN returns when a WHOIS answer has nothing to store.
var (
// ErrNoReferral is returned when IANA refers the AS to no RIR, as it
// does for a reserved, private-use or unallocated AS number.
ErrNoReferral = errors.New("IANA refers the AS to no RIR")
errNoASBlock = errors.New("IANA answered with neither a referral " +
"nor an AS block")
errNoFields = errors.New("RIR answered with no WHOIS fields")
errUnknownServer = errors.New("IANA refers the AS to a server no RIR runs")
errAnswerTooLarge = errors.New("WHOIS answer reached the size cap")
)
// ASNInfo contains parsed WHOIS information for an ASN.
type ASNInfo struct {
ASN int
@@ -63,20 +82,25 @@ type ASNInfo struct {
// Client performs WHOIS lookups for ASNs.
type Client struct {
// Dialer for creating connections (can be overridden for testing)
dialer *net.Dialer
// DialContext opens the connection to a WHOIS server, given as
// host:port. NewClient sets a net.Dialer's; tests replace it to reach
// local servers instead.
DialContext func(ctx context.Context, network, address string) (net.Conn, error)
}
// NewClient creates a new WHOIS client.
func NewClient() *Client {
return &Client{
dialer: &net.Dialer{
Timeout: dialTimeout,
},
}
dialer := &net.Dialer{Timeout: dialTimeout}
return &Client{DialContext: dialer.DialContext}
}
// LookupASN queries WHOIS for the given ASN and returns parsed information.
// LookupASN asks IANA which RIR holds the given ASN, asks that RIR, and
// returns what the RIR's answer says. It returns ErrNoReferral when IANA
// refers the AS to no RIR, and another error when a query fails, when IANA's
// answer has neither a referral nor the AS's block, when IANA refers the AS
// to a server none of the five RIRs run, or when the RIR's answer carries
// none of ASNInfo's fields, as a refusal of service does.
func (c *Client) LookupASN(ctx context.Context, asn int) (*ASNInfo, error) {
// Query IANA first to find the authoritative RIR
query := fmt.Sprintf("AS%d", asn)
@@ -87,72 +111,106 @@ func (c *Client) LookupASN(ctx context.Context, asn int) (*ASNInfo, error) {
}
// Determine RIR from IANA response
rir, whoisServer := c.parseIANAReferral(ianaResp)
rir, whoisServer, err := c.parseIANAReferral(ianaResp)
if err != nil {
return nil, fmt.Errorf("%s: %w", query, err)
}
if whoisServer == "" {
// No referral, try to parse what we have
return c.parseResponse(asn, rir, ianaResp), nil
// IANA names the block of an AS it refers to no RIR. An answer
// without that, such as an empty one, says nothing about the AS.
if !strings.Contains(ianaResp, "as-block:") {
return nil, fmt.Errorf("%s: %w", query, errNoASBlock)
}
return nil, fmt.Errorf("%s: %w", query, ErrNoReferral)
}
// Query the authoritative RIR
rirResp, err := c.query(ctx, whoisServer, query)
if err != nil {
// Return partial data from IANA if RIR query fails
info := c.parseResponse(asn, rir, ianaResp)
info.RawResponse = ianaResp + "\n--- RIR query failed: " + err.Error() + " ---\n"
return info, nil
return nil, fmt.Errorf("RIR query failed: %w", err)
}
// Combine responses and parse
fullResponse := ianaResp + "\n" + rirResp
info := c.parseResponse(asn, rir, fullResponse)
info.RawResponse = fullResponse
// The fields come from the RIR's answer alone: IANA's describes the
// block the AS is in.
info := c.parseResponse(asn, rir, rirResp)
if info.isEmpty() {
return nil, fmt.Errorf("%s from %s: %w", query, whoisServer, errNoFields)
}
info.RawResponse = ianaResp + "\n" + rirResp
return info, nil
}
// query performs a raw WHOIS query to the specified server.
// isEmpty reports whether info holds none of the fields a WHOIS answer
// fills, as when the answer is only comment and error lines.
func (info *ASNInfo) isEmpty() bool {
return info.ASName == "" && info.OrgName == "" && info.OrgID == "" &&
info.Address == "" && info.CountryCode == "" &&
info.AbuseEmail == "" && info.AbusePhone == "" &&
info.TechEmail == "" && info.TechPhone == "" &&
info.RegDate == nil && info.LastMod == nil
}
// query performs a raw WHOIS query to the specified server. An answer that
// reaches maxAnswerSize is an error.
func (c *Client) query(ctx context.Context, server, query string) (string, error) {
conn, err := c.dialer.DialContext(ctx, "tcp", server)
conn, err := c.DialContext(ctx, "tcp", server)
if err != nil {
return "", fmt.Errorf("dial %s: %w", server, err)
}
defer func() { _ = conn.Close() }()
// Set deadlines
if err := conn.SetWriteDeadline(time.Now().Add(writeTimeout)); err != nil {
err = conn.SetWriteDeadline(time.Now().Add(writeTimeout))
if err != nil {
return "", fmt.Errorf("set write deadline: %w", err)
}
// Send query
if _, err := fmt.Fprintf(conn, "%s\r\n", query); err != nil {
_, err = fmt.Fprintf(conn, "%s\r\n", query)
if err != nil {
return "", fmt.Errorf("write query: %w", err)
}
// Read response
if err := conn.SetReadDeadline(time.Now().Add(readTimeout)); err != nil {
err = conn.SetReadDeadline(time.Now().Add(readTimeout))
if err != nil {
return "", fmt.Errorf("set read deadline: %w", err)
}
var sb strings.Builder
scanner := bufio.NewScanner(conn)
// The reader ends at the cap; N, the bytes it has left, is 0 once the
// answer has reached it.
answer := &io.LimitedReader{R: conn, N: maxAnswerSize}
scanner := bufio.NewScanner(answer)
for scanner.Scan() {
sb.WriteString(scanner.Text())
sb.WriteString("\n")
}
if err := scanner.Err(); err != nil {
err = scanner.Err()
if err != nil {
return sb.String(), fmt.Errorf("read response: %w", err)
}
if answer.N == 0 {
return "", fmt.Errorf("read response: %w", errAnswerTooLarge)
}
return sb.String(), nil
}
// parseIANAReferral extracts the RIR and WHOIS server from an IANA response.
func (c *Client) parseIANAReferral(response string) (rir, whoisServer string) {
lines := strings.Split(response, "\n")
for _, line := range lines {
// The server is "" when the response has no referral. A referral to a server
// that is none of the five RIRs' is an error, so a forged or broken answer
// cannot send the client to any other host or port.
func (c *Client) parseIANAReferral(response string) (string, string, error) {
for line := range strings.SplitSeq(response, "\n") {
line = strings.TrimSpace(line)
// Look for "refer:" line
@@ -162,44 +220,22 @@ func (c *Client) parseIANAReferral(response string) (rir, whoisServer string) {
switch {
case strings.Contains(server, "arin"):
return RIRARIN, whoisServerARIN
return RIRARIN, whoisServerARIN, nil
case strings.Contains(server, "ripe"):
return RIRRIPE, whoisServerRIPE
return RIRRIPE, whoisServerRIPE, nil
case strings.Contains(server, "apnic"):
return RIRAPNIC, whoisServerAPNIC
return RIRAPNIC, whoisServerAPNIC, nil
case strings.Contains(server, "lacnic"):
return RIRLACNIC, whoisServerLACNIC
return RIRLACNIC, whoisServerLACNIC, nil
case strings.Contains(server, "afrinic"):
return RIRAFRNIC, whoisServerAFRINIC
return RIRAFRNIC, whoisServerAFRINIC, nil
default:
// Unknown server, add port if missing
if !strings.Contains(server, ":") {
server += ":43"
}
return "", server
}
}
// Also check organisation line for RIR hints
if strings.HasPrefix(strings.ToLower(line), "organisation:") {
org := strings.ToLower(line)
switch {
case strings.Contains(org, "arin"):
rir = RIRARIN
case strings.Contains(org, "ripe"):
rir = RIRRIPE
case strings.Contains(org, "apnic"):
rir = RIRAPNIC
case strings.Contains(org, "lacnic"):
rir = RIRLACNIC
case strings.Contains(org, "afrinic"):
rir = RIRAFRNIC
return "", "", fmt.Errorf("%w: %q", errUnknownServer, server)
}
}
}
return rir, ""
return "", "", nil
}
// parseResponse extracts ASN information from a WHOIS response.
@@ -211,11 +247,13 @@ func (c *Client) parseResponse(asn int, rir, response string) *ASNInfo {
}
lines := strings.Split(response, "\n")
var addressLines []string
for _, line := range lines {
line = strings.TrimSpace(line)
if line == "" || strings.HasPrefix(line, "%") || strings.HasPrefix(line, "#") {
if line == "" || strings.HasPrefix(line, "%") ||
strings.HasPrefix(line, "#") {
continue
}
@@ -232,63 +270,14 @@ func (c *Client) parseResponse(asn int, rir, response string) *ASNInfo {
continue
}
switch key {
// AS Name (varies by RIR)
case "asname", "as-name":
if info.ASName == "" {
info.ASName = value
}
// Organization
case "orgname", "org-name", "owner":
if info.OrgName == "" {
info.OrgName = value
}
case "orgid", "org-id", "org":
if info.OrgID == "" {
info.OrgID = value
}
// Address (collect multiple lines)
case "address":
if key == "address" {
addressLines = append(addressLines, value)
// Country
case "country":
if info.CountryCode == "" && len(value) == 2 {
info.CountryCode = strings.ToUpper(value)
}
// Abuse contact
case "orgabuseemail", "abuse-mailbox":
if info.AbuseEmail == "" {
info.AbuseEmail = value
}
case "orgabusephone":
if info.AbusePhone == "" {
info.AbusePhone = value
}
// Tech contact
case "orgtechemail":
if info.TechEmail == "" {
info.TechEmail = value
}
case "orgtechphone":
if info.TechPhone == "" {
info.TechPhone = value
}
// Registration dates
case "regdate", "created":
if info.RegDate == nil {
info.RegDate = c.parseDate(value)
}
case "updated", "last-modified", "changed":
if info.LastMod == nil {
info.LastMod = c.parseDate(value)
}
continue
}
c.setResponseField(info, key, value)
}
// Combine address lines
@@ -304,6 +293,65 @@ func (c *Client) parseResponse(asn int, rir, response string) *ASNInfo {
return info
}
// setResponseField stores the value of one "key: value" line of a WHOIS
// response in info, keeping the first value seen for each field.
func (c *Client) setResponseField(info *ASNInfo, key, value string) {
switch key {
// Country
case "country":
if info.CountryCode == "" && len(value) == 2 {
info.CountryCode = strings.ToUpper(value)
}
// Registration dates
case "regdate", "created":
if info.RegDate == nil {
info.RegDate = c.parseDate(value)
}
case "updated", "last-modified", "changed":
if info.LastMod == nil {
info.LastMod = c.parseDate(value)
}
default:
field := textFieldForKey(info, key)
if field != nil && *field == "" {
*field = value
}
}
}
// textFieldForKey returns the text field of info that a WHOIS response key
// fills, or nil when the key fills none.
func textFieldForKey(info *ASNInfo, key string) *string {
switch key {
// AS Name (varies by RIR)
case "asname", "as-name":
return &info.ASName
// Organization
case "orgname", "org-name", "owner":
return &info.OrgName
case "orgid", "org-id", "org":
return &info.OrgID
// Abuse contact
case "orgabuseemail", "abuse-mailbox":
return &info.AbuseEmail
case "orgabusephone":
return &info.AbusePhone
// Tech contact
case "orgtechemail":
return &info.TechEmail
case "orgtechphone":
return &info.TechPhone
default:
return nil
}
}
// parseDate attempts to parse various date formats used in WHOIS responses.
func (c *Client) parseDate(value string) *time.Time {
// Common formats
@@ -319,15 +367,15 @@ func (c *Client) parseDate(value string) *time.Time {
value = strings.TrimSpace(value)
// Handle "YYYYMMDD" format from LACNIC
if len(value) == lacnicDateFormatLen {
if _, err := time.Parse("20060102", value); err == nil {
t, _ := time.Parse("20060102", value)
t, err := time.Parse("20060102", value)
if err == nil {
return &t
}
}
for _, format := range formats {
if t, err := time.Parse(format, value); err == nil {
t, err := time.Parse(format, value)
if err == nil {
return &t
}
}
internal/whois/whois_test.go
+246
View File
@@ -0,0 +1,246 @@
package whois_test
import (
"bufio"
"context"
"errors"
"io"
"net"
"strings"
"sync"
"syscall"
"testing"
"time"
"git.eeqj.de/sneak/routewatch/internal/whois"
)
// The servers LookupASN asks: IANA first, then the RIR IANA refers it to.
const (
ianaServer = "whois.iana.org:43"
ripeServer = "whois.ripe.net:43"
)
// testASN is the AS the tests look up.
const testASN = 3333
// ianaReferral is IANA's answer for an AS in a block it gave to RIPE.
const ianaReferral = `% IANA WHOIS server
refer: whois.ripe.net
as-block: 3154-3353
organisation: Assigned by RIPE NCC
whois: whois.ripe.net
changed: 2002-09-01
source: IANA
`
// ripeRecord is RIPE's answer for the AS.
const ripeRecord = `% This is the RIPE Database query service.
aut-num: AS3333
as-name: RIPE-NCC-AS
org: ORG-RIEN1-RIPE
created: 2002-09-17T13:08:32Z
last-modified: 2024-01-09T08:55:41Z
source: RIPE
`
// ripeRefusal is RIPE's answer to a host it no longer serves.
const ripeRefusal = `% This is the RIPE Database query service.
% The objects are in RPSL format.
%
%ERROR:201: access denied for 192.0.2.1
%
% Sorry, access from your host has been permanently
% denied because of a repeated excessive querying.
`
// ianaPrivateUse is IANA's answer for a private-use AS: its block and no
// referral.
const ianaPrivateUse = `% IANA WHOIS server
as-block: 64512-65534
organisation: Reserved for Private Use
descr: Reserved for Private Use
source: IANA
`
// ianaNoObject is IANA's answer when it has nothing for the query.
const ianaNoObject = `% IANA WHOIS server
% This query returned 0 objects.
`
// fillerLine is one of the lines a server that does not stop keeps sending.
const fillerLine = "remarks: and more\n"
// endlessAnswerSize is how many bytes of filler lines that server sends:
// four times the client's cap on an answer.
const endlessAnswerSize = 4 << 20
// fakeServers returns a DialContext for a whois.Client that connects to a
// fake server for each address in answers, which reads the query and sends
// answers[address]. Connecting to any other address is refused.
func fakeServers(
t *testing.T, answers map[string]string,
) func(context.Context, string, string) (net.Conn, error) {
t.Helper()
var servers sync.WaitGroup
t.Cleanup(servers.Wait)
return func(_ context.Context, _, address string) (net.Conn, error) {
answer, ok := answers[address]
if !ok {
return nil, syscall.ECONNREFUSED
}
client, server := net.Pipe()
servers.Go(func() {
defer func() { _ = server.Close() }()
_, _ = bufio.NewReader(server).ReadString('\n')
_, _ = io.WriteString(server, answer)
})
return client, nil
}
}
// lookUp looks testASN up with fake servers sending answers.
func lookUp(
t *testing.T, answers map[string]string,
) (*whois.ASNInfo, error) {
t.Helper()
client := whois.NewClient()
client.DialContext = fakeServers(t, answers)
return client.LookupASN(t.Context(), testASN)
}
// TestLookupASNReadsTheRIRAnswer looks up an AS IANA refers to RIPE. The
// fields come from RIPE's answer, the last-modified date included: IANA's
// "changed:" line dates its block, not the AS.
func TestLookupASNReadsTheRIRAnswer(t *testing.T) {
t.Parallel()
info, err := lookUp(t, map[string]string{
ianaServer: ianaReferral,
ripeServer: ripeRecord,
})
if err != nil {
t.Fatalf("LookupASN: %v", err)
}
if info.ASName != "RIPE-NCC-AS" || info.OrgID != "ORG-RIEN1-RIPE" ||
info.RIR != whois.RIRRIPE {
t.Errorf("LookupASN = AS name %q, org %q, RIR %q; "+
"want RIPE-NCC-AS, ORG-RIEN1-RIPE, RIPE",
info.ASName, info.OrgID, info.RIR)
}
wantLastMod := time.Date(2024, 1, 9, 8, 55, 41, 0, time.UTC)
if info.LastMod == nil || !info.LastMod.Equal(wantLastMod) {
t.Errorf("LookupASN last modified = %v, want %v",
info.LastMod, wantLastMod)
}
}
// TestLookupASNFailsWhenTheRIRGivesNoAnswer looks up an AS IANA refers to
// RIPE when RIPE's server refuses the connection, and when it answers only
// with an error. Neither is a lookup that found the AS.
func TestLookupASNFailsWhenTheRIRGivesNoAnswer(t *testing.T) {
t.Parallel()
for name, answers := range map[string]map[string]string{
"connection refused": {ianaServer: ianaReferral},
"only an error": {ianaServer: ianaReferral, ripeServer: ripeRefusal},
} {
t.Run(name, func(t *testing.T) {
t.Parallel()
_, err := lookUp(t, answers)
if err == nil || errors.Is(err, whois.ErrNoReferral) {
t.Errorf("LookupASN error = %v, want one that is not "+
"ErrNoReferral", err)
}
})
}
}
// TestLookupASNReportsNoReferral looks up a private-use AS, which IANA
// refers to no RIR.
func TestLookupASNReportsNoReferral(t *testing.T) {
t.Parallel()
_, err := lookUp(t, map[string]string{ianaServer: ianaPrivateUse})
if !errors.Is(err, whois.ErrNoReferral) {
t.Errorf("LookupASN error = %v, want ErrNoReferral", err)
}
}
// TestLookupASNFailsWhenIANAAnswersNoBlock looks up an AS when IANA answers
// with neither a referral nor the AS's block. That is not IANA saying no
// RIR holds the AS.
func TestLookupASNFailsWhenIANAAnswersNoBlock(t *testing.T) {
t.Parallel()
_, err := lookUp(t, map[string]string{ianaServer: ianaNoObject})
if err == nil || errors.Is(err, whois.ErrNoReferral) {
t.Errorf("LookupASN error = %v, want one that is not ErrNoReferral",
err)
}
}
// TestLookupASNFailsWhenTheRIRAnswerDoesNotEnd looks up an AS whose RIR
// server follows the record with filler lines far past any real answer.
// The client stops reading at its cap and fails instead of keeping them.
func TestLookupASNFailsWhenTheRIRAnswerDoesNotEnd(t *testing.T) {
t.Parallel()
endless := ripeRecord +
strings.Repeat(fillerLine, endlessAnswerSize/len(fillerLine))
_, err := lookUp(t, map[string]string{
ianaServer: ianaReferral,
ripeServer: endless,
})
if err == nil || errors.Is(err, whois.ErrNoReferral) {
t.Errorf("LookupASN error = %v, want one that is not ErrNoReferral",
err)
}
}
// TestLookupASNFailsOnAReferralToAnotherServer looks up an AS when IANA's
// answer refers it to a server none of the five RIRs run, with a port and
// without. The client follows a referral only to an RIR's server, so it
// fails instead of asking that server, which would answer with a record.
func TestLookupASNFailsOnAReferralToAnotherServer(t *testing.T) {
t.Parallel()
// Each case maps the server on the "refer:" line to the address the
// client would dial to follow it. IANA's answer also names the AS's
// block, so a client that dropped the referral would report no RIR.
for refer, address := range map[string]string{
"whois.example.net:8080": "whois.example.net:8080",
"whois.example.net": "whois.example.net:43",
} {
t.Run(refer, func(t *testing.T) {
t.Parallel()
_, err := lookUp(t, map[string]string{
ianaServer: "refer: " + refer + "\n\n" +
"as-block: 3154-3353\n",
address: ripeRecord,
})
if err == nil || errors.Is(err, whois.ErrNoReferral) {
t.Errorf("LookupASN error = %v, want one that is not "+
"ErrNoReferral", err)
}
})
}
}
package.json
+6
View File
@@ -0,0 +1,6 @@
{
"private": true,
"devDependencies": {
"prettier": "3.8.1"
}
}
pkg/asinfo/asinfo.go
+4 -2
View File
@@ -1,4 +1,3 @@
// Package asinfo provides information about Autonomous Systems (AS)
package asinfo
import (
@@ -54,7 +53,9 @@ func initRegistry() {
}
var asInfos []Info
if err := json.Unmarshal(jsonData, &asInfos); err != nil {
err = json.Unmarshal(jsonData, &asInfos)
if err != nil {
panic("failed to unmarshal embedded AS data: " + err.Error())
}
@@ -137,6 +138,7 @@ func Search(query string) []Info {
defer defaultRegistry.mu.RUnlock()
var results []Info
for _, info := range defaultRegistry.byASN {
if contains(info.Handle, query) || contains(info.Description, query) {
results = append(results, *info)
pkg/asinfo/asinfo_test.go
+52 -16
View File
@@ -1,10 +1,15 @@
package asinfo
package asinfo_test
import (
"strings"
"testing"
"git.eeqj.de/sneak/routewatch/pkg/asinfo"
)
func TestGet(t *testing.T) {
t.Parallel()
tests := []struct {
name string
asn int
@@ -38,20 +43,26 @@ func TestGet(t *testing.T) {
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
info, ok := Get(tt.asn)
t.Parallel()
info, ok := asinfo.Get(tt.asn)
if ok != tt.wantOK {
t.Errorf("Get(%d) ok = %v, want %v", tt.asn, ok, tt.wantOK)
return
}
if ok && info.Description != tt.wantDesc {
t.Errorf("Get(%d) description = %q, want %q", tt.asn, info.Description, tt.wantDesc)
t.Errorf("Get(%d) description = %q, want %q",
tt.asn, info.Description, tt.wantDesc)
}
})
}
}
func TestGetDescription(t *testing.T) {
t.Parallel()
tests := []struct {
asn int
want string
@@ -62,7 +73,10 @@ func TestGetDescription(t *testing.T) {
for _, tt := range tests {
t.Run("", func(t *testing.T) {
if got := GetDescription(tt.asn); got != tt.want {
t.Parallel()
got := asinfo.GetDescription(tt.asn)
if got != tt.want {
t.Errorf("GetDescription(%d) = %q, want %q", tt.asn, got, tt.want)
}
})
@@ -70,6 +84,8 @@ func TestGetDescription(t *testing.T) {
}
func TestGetHandle(t *testing.T) {
t.Parallel()
tests := []struct {
asn int
want string
@@ -81,7 +97,10 @@ func TestGetHandle(t *testing.T) {
for _, tt := range tests {
t.Run("", func(t *testing.T) {
if got := GetHandle(tt.asn); got != tt.want {
t.Parallel()
got := asinfo.GetHandle(tt.asn)
if got != tt.want {
t.Errorf("GetHandle(%d) = %q, want %q", tt.asn, got, tt.want)
}
})
@@ -89,18 +108,23 @@ func TestGetHandle(t *testing.T) {
}
func TestTotal(t *testing.T) {
total := Total()
t.Parallel()
total := asinfo.Total()
if total < 100000 {
t.Errorf("Total() = %d, expected > 100000", total)
}
// Verify it's consistent
if total2 := Total(); total2 != total {
total2 := asinfo.Total()
if total2 != total {
t.Errorf("Total() returned different values: %d vs %d", total, total2)
}
}
func TestSearch(t *testing.T) {
t.Parallel()
tests := []struct {
name string
query string
@@ -130,16 +154,21 @@ func TestSearch(t *testing.T) {
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
results := Search(tt.query)
t.Parallel()
results := asinfo.Search(tt.query)
if len(results) < tt.wantMin {
t.Errorf("Search(%q) returned %d results, want at least %d", tt.query, len(results), tt.wantMin)
t.Errorf("Search(%q) returned %d results, want at least %d",
tt.query, len(results), tt.wantMin)
}
// Verify all results contain the query
if tt.query != "" {
for _, r := range results {
if !contains(r.Handle, tt.query) && !contains(r.Description, tt.query) {
t.Errorf("Search result ASN %d doesn't contain query %q", r.ASN, tt.query)
if !strings.Contains(r.Handle, tt.query) &&
!strings.Contains(r.Description, tt.query) {
t.Errorf("Search result ASN %d doesn't contain query %q",
r.ASN, tt.query)
}
}
}
@@ -148,13 +177,17 @@ func TestSearch(t *testing.T) {
}
func TestDataIntegrity(t *testing.T) {
t.Parallel()
// Verify no duplicate ASNs
all := All()
all := asinfo.All()
seen := make(map[int]bool)
for _, info := range all {
if seen[info.ASN] {
t.Errorf("Duplicate ASN found: %d", info.ASN)
}
seen[info.ASN] = true
}
@@ -163,6 +196,7 @@ func TestDataIntegrity(t *testing.T) {
if info.Handle == "" && info.ASN != 0 {
t.Errorf("ASN %d has empty handle", info.ASN)
}
if info.Description == "" {
t.Errorf("ASN %d has empty description", info.ASN)
}
@@ -174,8 +208,9 @@ func BenchmarkGet(b *testing.B) {
asns := []int{1, 15169, 13335, 32934, 8075, 16509}
b.ResetTimer()
for i := 0; i < b.N; i++ {
Get(asns[i%len(asns)])
for i := range b.N {
asinfo.Get(asns[i%len(asns)])
}
}
@@ -183,7 +218,8 @@ func BenchmarkSearch(b *testing.B) {
queries := []string{"Google", "Amazon", "Microsoft", "University"}
b.ResetTimer()
for i := 0; i < b.N; i++ {
Search(queries[i%len(queries)])
for i := range b.N {
asinfo.Search(queries[i%len(queries)])
}
}
script/bootstrap
+109 -11
View File
@@ -2,14 +2,27 @@
# script/bootstrap: install all dependencies needed to build and develop
# this repo. Idempotent: every install is guarded by a check so already
# installed tools are skipped. Base tooling comes from nix, apt, brew,
# or apk (detected in that order); assumes NOTHING is present (not git,
# make, or go).
# or apk (detected in that order); assumes nothing is present. The
# installed node is used only when its major version is the pinned one;
# otherwise (node missing, or another major) the pinned version is
# installed via nvm (installing nvm itself first, from a hash-verified
# release archive, never curl | sh), and yarn and the packages are
# installed under it. Go formats the code and builds it on the host; the
# linter and the tests run only in Docker, as phases of the Dockerfile.
set -eu
ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
# Pinned versions, 2026-07-06
NODE_VERSION="22.17.0"
NVM_VERSION="0.40.3"
# sha256 of https://github.com/nvm-sh/nvm/archive/refs/tags/v0.40.3.tar.gz
NVM_SHA256="5f4d6aaa04a177dc93c985e31dbc411ab6b8c6e1e21d8015dbc1372625fcd1d0"
YARN_VERSION="1.22.22"
PKGMGR=""
SUDO=""
APT_UPDATED=""
detect_pkgmgr() {
[ -n "$PKGMGR" ] && return 0
@@ -38,7 +51,14 @@ pkg_install() {
detect_pkgmgr
case "$PKGMGR" in
nix) nix-env -iA "nixpkgs.$1" ;;
apt) $SUDO env DEBIAN_FRONTEND=noninteractive apt-get install -y "$2" ;;
apt)
# Package lists may be empty (fresh images); refresh once per run.
if [ -z "$APT_UPDATED" ]; then
$SUDO env DEBIAN_FRONTEND=noninteractive apt-get update
APT_UPDATED=1
fi
$SUDO env DEBIAN_FRONTEND=noninteractive apt-get install -y "$2"
;;
brew) brew install "$3" ;;
apk) apk add --no-cache "$4" ;;
esac
@@ -48,22 +68,100 @@ missing() {
! command -v "$1" >/dev/null 2>&1
}
# verify_sha256 <file> <expected-hash>
verify_sha256() {
if command -v sha256sum >/dev/null 2>&1; then
actual="$(sha256sum "$1" | cut -d' ' -f1)"
else
actual="$(shasum -a 256 "$1" | cut -d' ' -f1)"
fi
if [ "$actual" != "$2" ]; then
echo "bootstrap: sha256 mismatch for $1" >&2
echo " expected: $2" >&2
echo " actual: $actual" >&2
exit 1
fi
}
# nvm is a bash script; run a command in a bash with nvm loaded
nvm_sh() {
bash -c ". \"\$HOME/.nvm/nvm.sh\" && $*"
}
ensure_nvm() {
[ -s "$HOME/.nvm/nvm.sh" ] && return 0
# nvm prerequisites; nvm itself requires bash
if missing bash; then pkg_install bash bash bash bash; fi
if missing curl; then pkg_install curl curl curl curl; fi
if missing git; then pkg_install git git git git; fi
tmp="$(mktemp -d)"
curl -fsSL -o "$tmp/nvm.tar.gz" \
"https://github.com/nvm-sh/nvm/archive/refs/tags/v${NVM_VERSION}.tar.gz"
verify_sha256 "$tmp/nvm.tar.gz" "$NVM_SHA256"
mkdir -p "$HOME/.nvm"
tar -xzf "$tmp/nvm.tar.gz" -C "$HOME/.nvm" --strip-components=1
rm -rf "$tmp"
}
# True when the node on PATH has the pinned major version. Only the
# major is compared, as in the shared script/bootstrap of sneak/prompts.
# Another major is not used: the pinned yarn 1 prints a deprecation
# warning under node 24. script/fmt and script/fmt-check carry this
# function unchanged, to run the yarn installed here.
node_is_pinned_major() {
if ! command -v node >/dev/null 2>&1; then return 1; fi
installed="$(node --version)"
installed="${installed#v}"
[ "${installed%%.*}" = "${NODE_VERSION%%.*}" ]
}
ensure_node() {
if node_is_pinned_major; then
echo "bootstrap: using node $(node --version)"
return 0
fi
ensure_nvm
nvm_sh "nvm install $NODE_VERSION"
nvm_sh "nvm use $NODE_VERSION >/dev/null && \
echo \"bootstrap: using node \$(node --version)\""
}
ensure_yarn() {
if ! node_is_pinned_major; then
nvm_sh "nvm use $NODE_VERSION >/dev/null && corepack enable && \
corepack prepare yarn@$YARN_VERSION --activate"
return 0
fi
if ! missing yarn; then return 0; fi
if ! missing corepack; then
corepack enable
corepack prepare "yarn@$YARN_VERSION" --activate
else
npm install -g "yarn@$YARN_VERSION"
fi
}
install_js_deps() {
if node_is_pinned_major; then
yarn install --frozen-lockfile
else
nvm_sh "nvm use $NODE_VERSION >/dev/null && cd \"$ROOT\" && \
yarn install --frozen-lockfile"
fi
}
main() {
cd "$ROOT"
# Base tooling (every repo)
if missing git; then pkg_install git git git git; fi
if missing make; then pkg_install gnumake make make make; fi
if missing git; then pkg_install git git git git; fi
# Go toolchain
if missing go; then pkg_install go golang go go; fi
# golangci-lint: packaged in nix, brew, and apk. There is no apt
# package; on apt systems install it manually from a hash-verified
# GitHub release archive (never curl | sh).
if missing golangci-lint; then
pkg_install golangci-lint golangci-lint golangci-lint golangci-lint
fi
ensure_node
ensure_yarn
install_js_deps
go mod download
script/check
+3 -2
View File
@@ -1,7 +1,8 @@
#!/bin/sh
# script/check: run all checks (test, lint, fmt-check). Our own
# extension to scripts-to-rule-them-all. Must not modify any files.
# Generic: usually needs no adaptation.
# extension to scripts-to-rule-them-all. test and lint are Docker
# phases; fmt-check is native, because a formatter writes the working
# tree. Must not modify any files.
set -eu
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
script/cibuild
+24 -5
View File
@@ -1,14 +1,33 @@
#!/bin/sh
# script/cibuild: run the CI build. The Dockerfile runs script/check
# (via make check), so a successful build implies all checks pass.
# Generic: needs no adaptation. The Gitea workflow runs this on push.
# script/cibuild: run the CI build. It bootstraps first: a CI runner
# checks out and runs this and nothing else, and script/fmt-check runs
# the formatter on the host, which a pristine checkout cannot do.
# A new CACHEBUST value reruns every RUN step after the ARG CACHEBUST
# line, since a cached gate phase is a check that did not run; unlike
# --no-cache, it keeps the layers above that line and the cache mounts.
# This relies on every stage of the Dockerfile declaring ARG CACHEBUST
# before every step that lints, tests or compiles; a stage without that
# line is served from cache on an unchanged tree, with no warning.
set -eu
ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
main() {
cd "$ROOT"
docker build .
"$SCRIPT_DIR/bootstrap"
"$SCRIPT_DIR/check"
# The version and the tag each get their own line: a failing
# command substitution inside an argument does not trip `set -e`,
# so the inline form degrades silently to an empty constant. The
# VERSION build argument takes precedence over the version a build
# stage derives from the .git in the context.
version="$(git describe --tags --always --dirty 2>/dev/null || true)"
[ -n "$version" ] || version="unknown"
tag="$("$SCRIPT_DIR/projectname")"
docker build --build-arg CACHEBUST="$(date +%s)-$$" \
--build-arg VERSION="$version" \
-t "$tag" .
}
main "$@"
script/docker
+17 -2
View File
@@ -1,7 +1,12 @@
#!/bin/sh
# script/docker: build the Docker image tagged with the project name.
# Identical in all repos; the tag comes from script/projectname.
# Generic: needs no adaptation.
# A new CACHEBUST value reruns every RUN step after the ARG CACHEBUST
# line, since a cached gate phase is a check that did not run; unlike
# --no-cache, it keeps the layers above that line and the cache mounts.
# This relies on every stage of the Dockerfile declaring ARG CACHEBUST
# before every step that lints, tests or compiles; a stage without that
# line is served from cache on an unchanged tree, with no warning.
set -eu
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -9,7 +14,17 @@ ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
main() {
cd "$ROOT"
docker build -t "$("$SCRIPT_DIR/projectname")" .
# The version and the tag each get their own line: a failing
# command substitution inside an argument does not trip `set -e`,
# so the inline form degrades silently to an empty constant. The
# VERSION build argument takes precedence over the version a build
# stage derives from the .git in the context.
version="$(git describe --tags --always --dirty 2>/dev/null || true)"
[ -n "$version" ] || version="unknown"
tag="$("$SCRIPT_DIR/projectname")"
docker build --build-arg CACHEBUST="$(date +%s)-$$" \
--build-arg VERSION="$version" \
-t "$tag" .
}
main "$@"
script/fmt
+32 -1
View File
@@ -1,12 +1,43 @@
#!/bin/sh
# script/fmt: format all files (writes).
# script/fmt: format all files (writes): Go with go fmt, markdown with
# prettier.
set -eu
ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
# Must match the pin in script/bootstrap.
NODE_VERSION="22.17.0"
# True when the node on PATH has the pinned major version: the same
# function as in script/bootstrap, which uses it to decide where it
# installs yarn.
node_is_pinned_major() {
if ! command -v node >/dev/null 2>&1; then return 1; fi
installed="$(node --version)"
installed="${installed#v}"
[ "${installed%%.*}" = "${NODE_VERSION%%.*}" ]
}
# Run the yarn script/bootstrap installed: the yarn on PATH when the
# node on PATH has the pinned major version, and otherwise yarn under
# the pinned node in nvm, which bootstrap leaves off the PATH of the
# shell that called it. nvm is a bash script, hence the subshell.
run_yarn() {
if node_is_pinned_major && command -v yarn >/dev/null 2>&1; then
exec yarn "$@"
fi
if node_is_pinned_major || [ ! -s "$HOME/.nvm/nvm.sh" ]; then
echo "fmt: no yarn; run script/bootstrap first" >&2
exit 1
fi
exec bash -c '. "$HOME/.nvm/nvm.sh" && nvm use "$1" >/dev/null &&
shift && exec yarn "$@"' bash "$NODE_VERSION" "$@"
}
main() {
cd "$ROOT"
go fmt ./...
run_yarn run prettier --write '**/*.md' --tab-width 4 --prose-wrap always
}
main "$@"
script/fmt-check
+30
View File
@@ -5,6 +5,35 @@ set -eu
ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
# Must match the pin in script/bootstrap.
NODE_VERSION="22.17.0"
# True when the node on PATH has the pinned major version: the same
# function as in script/bootstrap, which uses it to decide where it
# installs yarn.
node_is_pinned_major() {
if ! command -v node >/dev/null 2>&1; then return 1; fi
installed="$(node --version)"
installed="${installed#v}"
[ "${installed%%.*}" = "${NODE_VERSION%%.*}" ]
}
# Run the yarn script/bootstrap installed: the yarn on PATH when the
# node on PATH has the pinned major version, and otherwise yarn under
# the pinned node in nvm, which bootstrap leaves off the PATH of the
# shell that called it. nvm is a bash script, hence the subshell.
run_yarn() {
if node_is_pinned_major && command -v yarn >/dev/null 2>&1; then
exec yarn "$@"
fi
if node_is_pinned_major || [ ! -s "$HOME/.nvm/nvm.sh" ]; then
echo "fmt-check: no yarn; run script/bootstrap first" >&2
exit 1
fi
exec bash -c '. "$HOME/.nvm/nvm.sh" && nvm use "$1" >/dev/null &&
shift && exec yarn "$@"' bash "$NODE_VERSION" "$@"
}
main() {
cd "$ROOT"
unformatted="$(gofmt -l .)"
@@ -13,6 +42,7 @@ main() {
echo "$unformatted" >&2
exit 1
fi
run_yarn run prettier --check '**/*.md' --tab-width 4 --prose-wrap always
}
main "$@"
script/lint
+19 -4
View File
@@ -1,13 +1,28 @@
#!/bin/sh
# script/lint: run the linters.
# script/lint: run the linter. Linting is a phase of the Dockerfile and
# this builds that phase alone; the linter is never installed or run on
# a developer host, where a shared result cache and a host-global lock
# make its answer untrustworthy.
#
# The phase is not the last stage in the file, so it is built only when
# --target names it. A new CACHEBUST value reruns every RUN step after
# the ARG CACHEBUST line, since a cached lint layer is a lint that did
# not run; unlike --no-cache, it keeps the layers above that line and
# the cache mounts. This relies on every stage of the Dockerfile
# declaring ARG CACHEBUST before every step that lints, tests or
# compiles; a stage without that line is served from cache on an
# unchanged tree, with no warning.
# --output type=cacheonly writes no image, since nothing uses one.
set -eu
ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
main() {
cd "$ROOT"
go vet ./...
golangci-lint run
docker build --build-arg CACHEBUST="$(date +%s)-$$" \
--target lint \
--output type=cacheonly .
}
main "$@"
script/test
+15 -8
View File
@@ -1,17 +1,24 @@
#!/bin/sh
# script/test: run the test suite. On failure, rerun verbosely so the
# failing tests are visible in the output.
# script/test: run the test suite. Testing is a phase of the Dockerfile
# and this builds that phase alone, on the same terms as script/lint:
# --target because a phase that is not the last stage is built only when
# named, and a new CACHEBUST value because a cached test layer is a test
# that did not run; unlike --no-cache, it keeps the layers above the ARG
# CACHEBUST line and the cache mounts. This relies on every stage of the
# Dockerfile declaring ARG CACHEBUST before every step that lints, tests
# or compiles; a stage without that line is served from cache on an
# unchanged tree, with no warning. --output type=cacheonly writes no
# image, since nothing uses one.
set -eu
ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
main() {
cd "$ROOT"
go test -short -timeout 30s -race -cover ./... || {
echo "--- Rerunning with -v for details ---"
go test -short -timeout 30s -race -v ./...
exit 1
}
docker build --build-arg CACHEBUST="$(date +%s)-$$" \
--target test \
--output type=cacheonly .
}
main "$@"
yarn.lock
+8
View File
@@ -0,0 +1,8 @@
# THIS IS AN AUTOGENERATED FILE. DO NOT EDIT THIS FILE DIRECTLY.
# yarn lockfile v1
prettier@3.8.1:
version "3.8.1"
resolved "https://registry.yarnpkg.com/prettier/-/prettier-3.8.1.tgz#edf48977cf991558f4fcbd8a3ba6015ba2a3a173"
integrity sha512-UOnG6LftzbdaHZcKoPFtOcCKztrQ57WkHDeRD9t/PTQtmT0NHSeWWepj6pS0z/N7+08BHFDQVUrfmfMRcZwbMg==