WIP: re-vendor canonical files and gate phases

Model: opus-5-5

.dockerignore

@@ -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

@@ -10,3 +10,6 @@ insert_final_newline = true
 
 [Makefile]
 indent_style = tab
+
+[*.go]
+indent_style = tab

.gitea/workflows/check.yml

@@ -4,6 +4,6 @@ jobs:
     check:
         runs-on: ubuntu-latest
         steps:
-            # actions/checkout v4.2.2, 2026-02-28
+            # actions/checkout v4.2.2, 2026-02-22
             - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
             - run: script/cibuild

.gitignore

@@ -1,41 +1,77 @@
-# 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 after these lines, 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, kept after the canonical content above.
+
+# 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

@@ -1,92 +1,99 @@
 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:
+    # 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
+  settings:
     lll:
-    line-length: 120
-
-  nestif:
-    min-complexity: 4
-
-  nlreturn:
-    block-size: 2
-
-  revive:
+      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:
-      - name: var-naming
-        arguments:
-          - []
-          - []
-          - "upperCaseConst=true"
-
-  tagliatelle:
-    case:
-      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

Dockerfile

@@ -1,22 +1,37 @@
-# 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
-
+# Lint phase, built alone by script/lint. The linter is invoked directly
+# rather than through `make lint`, which is itself a docker build and
+# would recurse into a daemon that does not exist in a build step. The
+# image is Debian-based and has Go and gcc, so it can check the cgo
+# SQLite package without extra installs.
+# golangci/golangci-lint:v2.14.0 (Go 1.27.0), 2026-10-14
+FROM golangci/golangci-lint@sha256:ad862ba6b3798cbe0fd9fd7408d498fd74fbd2623a92406b2fd3898faf0bf98f AS lint
 WORKDIR /src
-
 COPY go.mod go.sum ./
 RUN go mod download
-
 COPY . .
+RUN golangci-lint run --config .golangci.yml ./...
 
-RUN make fmt-check
-RUN make lint
+# Test phase, built alone by script/test. -race needs cgo and so a C
+# compiler, which the Debian Go image ships. -short skips the live RIPE RIS
+# feed test, so the suite needs no network. The rerun with -v only shows
+# details: the build fails however it ends, because the first run failed.
+# golang:1.24-bookworm, 2026-09-21
+FROM golang@sha256:1a6d4452c65dea36aac2e2d606b01b4a029ec90cc1ae53890540ce6173ea77ac AS test
+WORKDIR /src
+COPY go.mod go.sum ./
+RUN go mod download
+COPY . .
+RUN go test -short -timeout 90s -race -cover ./... || \
+    { echo "--- Rerunning with -v for details ---"; \
+      go test -short -timeout 90s -race -v ./...; exit 1; }
 
-# Build stage
+# Build stage. Nothing is wanted from the lint and test phases; the copies
+# are what make BuildKit build them first, so this stage cannot run unless
+# lint and test passed.
 # golang:1.24-bookworm, 2026-09-21
 FROM golang@sha256:1a6d4452c65dea36aac2e2d606b01b4a029ec90cc1ae53890540ce6173ea77ac AS builder
+COPY --from=lint /src/go.sum /dev/null
+COPY --from=test /src/go.sum /dev/null
 
 # Install build dependencies (zstd for archive, gcc for CGO/sqlite3)
 RUN apt-get update && apt-get install -y --no-install-recommends \
@@ -25,10 +40,10 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     libc6-dev \
     && rm -rf /var/lib/apt/lists/*
 
-WORKDIR /src
+# A tar-stream context keeps the sender's file owners, which git refuses.
+RUN git config --system --add safe.directory /src
 
-# Force BuildKit to run the lint stage before compiling or testing.
-COPY --from=lint /src/go.sum /dev/null
+WORKDIR /src
 
 # Copy everything
 COPY . .
@@ -36,12 +51,23 @@ COPY . .
 # Vendor dependencies (must be after copying source)
 RUN go mod download && 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 argument when one is given, otherwise
+# `git describe --tags --always` of the .git in the build context (git comes
+# with this image): the tag on a tagged commit, tag-N-gHASH after one, the
+# short commit when no tag is reachable. A context that carries .git and still
+# yields no version fails the build. The footer links to the full commit.
+ARG VERSION
+RUN version="${VERSION:-$(git describe --tags --always || echo unknown)}"; \
+    if [ -e .git ] && { [ -z "$version" ] || [ "$version" = dev ] || \
+        [ "$version" = unknown ]; }; then \
+        echo "no version could be derived although the build context carries .git" >&2; \
+        exit 1; \
+    fi; \
+    CGO_ENABLED=1 GOOS=linux go build -o /routewatch -ldflags "\
+        -X git.eeqj.de/sneak/routewatch/internal/version.GitRevision=$(git rev-parse --verify HEAD || echo unknown) \
+        -X git.eeqj.de/sneak/routewatch/internal/version.GitRevisionShort=$version" \
+        ./cmd/routewatch
 
 # Create source archive with vendored dependencies
 RUN tar --zstd -cf /routewatch-source.tar.zst \
@@ -79,8 +105,8 @@ 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 +122,8 @@ 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

@@ -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

@@ -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

@@ -1,6 +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
@@ -139,7 +142,10 @@ routewatch/
 - **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
+- **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 33 lookups for IPv4, 129 for IPv6); prefixes are stored in the
+  text form Go's `net/netip` prints
 
 ### Database Schema
 
@@ -151,7 +157,7 @@ prefixes_v6(id, prefix, mask_length, first_seen, last_seen)
 
 -- 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)
+               next_hop, last_updated)
 live_routes_v6(id, prefix, mask_length, origin_asn, peer_ip, as_path,
                next_hop, last_updated)
 
@@ -166,14 +172,21 @@ Configuration is handled via environment variables and OS-specific paths:
 
 | Variable | Default | Description |
 |----------|----------|-------------|
-| `PORT` | `8080` | HTTP server port |
+| `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; see Memory |
+| `MALLOC_ARENA_MAX` | `2` (in the Docker image) | glibc malloc arena cap, a positive whole number; see Memory |
+
+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.
 
 State directory (database location):
 - macOS: `~/Library/Application Support/routewatch/`
-- Linux: `/var/lib/routewatch/` or `~/.local/share/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
 
@@ -220,6 +233,22 @@ What happens at each limit:
 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: nothing is required. 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`.
+
 ## Development
 
 ```bash
@@ -274,4 +303,8 @@ invoke `go test` directly without `-short`:
 
 ## License
 
-See LICENSE file.
+MIT. See [`LICENSE`](LICENSE).
+
+## Author
+
+[@sneak](https://sneak.berlin)

REPO_POLICIES.md

@@ -1,6 +1,6 @@
 ---
 title: Repository Policies
-last_modified: 2026-07-06
+last_modified: 2026-10-04
 ---
 
 This document covers repository structure, tooling, and workflow standards. Code
@@ -60,17 +60,28 @@ style conventions are in separate documents:
   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
-  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
+  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**:
+  `script/bootstrap` installs node and yarn under nvm and leaves neither on the
+  `PATH` of the shell that called it, so a bare `yarn` still exits 127. The host
+  entrypoints that need yarn — `script/fmt` and `script/fmt-check` — therefore
+  source nvm for the pinned node version before invoking it, exactly as
+  `script/bootstrap`'s own install step does. 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
   must document the provided scripts in an **Entrypoints** section (see the
   README requirements below).
@@ -89,87 +100,198 @@ 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 --no-cache --target lint -t "$(script/projectname)-lint" .
+    docker build --no-cache --target test -t "$(script/projectname)-test" .
+    ```
+
+    **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.
+
+    **Every `docker build` in `script/` is tagged**, here and in
+    `script/cibuild` and `script/docker`. An untagged build leaves a dangling
+    image behind on every invocation, on every developer host and every CI
+    runner; a tagged one replaces the previous image.
+
+    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 cache, its own `TMPDIR` and
+    a digest-pinned binary, so neither is reachable.
+
+- **Any build that runs checks is built with `--no-cache`.** 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 `--no-cache`:
+  `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. 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
+    # 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
     COPY . .
-    RUN make fmt-check
-    RUN make lint
+    RUN 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.
+    # 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
     COPY . .
-    RUN make test
+    RUN go test -timeout 90s -race -cover ./... || \
+        { echo "--- Rerunning with -v for details ---"; \
+          go test -timeout 90s -race -v ./...; exit 1; }
 
-    ARG VERSION=dev
-    RUN CGO_ENABLED=0 go build -trimpath \
+    # Build stage. Nothing is wanted from either phase above; the copies
+    # are what make BuildKit build them first, so this stage cannot run
+    # unless lint and test passed.
+    # golang:1.x-alpine, YYYY-MM-DD
+    FROM golang@sha256:... AS builder
+    COPY --from=lint /src/go.sum /dev/null
+    COPY --from=test /src/go.sum /dev/null
+    RUN apk add --no-cache git
+    # A tar-stream context keeps the sender's file owners, which git refuses.
+    RUN git config --system --add safe.directory /src
+    WORKDIR /src
+    COPY go.mod go.sum ./
+    RUN go mod download
+    COPY . .
+
+    # 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 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
+    # Runtime stage, and the last one
     FROM alpine@sha256:...
     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.
     - 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:
+
+        ```dockerfile
+        RUN apt-get update \
+            && apt-get install -y --no-install-recommends libvips-dev \
+            && rm -rf /var/lib/apt/lists/*
+        ```
+
+    - `.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.
 
 - 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.
+  That script 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 is uncached 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. 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 +311,21 @@ 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 +338,26 @@ 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. That matters on a developer's machine, where this
+    target runs and the directory lasts from one run to the next. The `test`
+    phase of the `Dockerfile` needs no `-count=1`: its base image holds no
+    result for this repo's tests and nothing before its `go test` step runs a
+    test, so there is nothing to replay. `--no-cache` (above) is what makes that
+    step run on an unchanged tree.
+
     Python example:
 
     ```makefile
@@ -239,10 +383,84 @@ 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
-  `https://git.eeqj.de/sneak/prompts/raw/branch/main/.gitignore` when setting up
-  a new repo.
+  editor files (`.swp`, `*~`), in-repo agent scratch directories (`.claude/`),
+  language build artifacts, and `node_modules/`. Fetch the standard `.gitignore`
+  from `https://git.eeqj.de/sneak/prompts/raw/branch/main/.gitignore` when
+  setting up a new repo. 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
+    # 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"
+    docker build --no-cache \
+        --build-arg VERSION="$version" \
+        -t "$(script/projectname)" .
+    ```
+
+    `--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 a repo that embeds a
+    tag-derived version must set `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 +476,56 @@ 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.
+
+    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).
@@ -374,12 +639,14 @@ style conventions are in separate documents:
   settings.
 
 - 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 +673,7 @@ 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

@@ -10,19 +10,95 @@
 
 # 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. The Docker build runs the format check, the linter
+and the tests, and the Gitea workflow runs that build 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
+35-hour run of `3898daa` on the live feed peaked at about 1 GiB, without a
+container memory limit.
 
 # 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.
+`next` waits for sneak to merge it to `main` through
+https://git.eeqj.de/sneak/routewatch/pulls/6. After that, setting
+routewatch up under upaas on fsn1app1 and deploying it are his
+(https://git.eeqj.de/sneak/routewatch/issues/31), and so is the run under a
+real 5 GiB limit (https://git.eeqj.de/sneak/routewatch/issues/3).
 
 # Completed Steps
 
+- 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
@@ -41,8 +117,6 @@ runs make check on main.
 
 # 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
+- Production memory under 5 GiB: whether to test under a real 5 GiB
+  container limit on fsn1app1 is open for sneak
+  (https://git.eeqj.de/sneak/routewatch/issues/3)

entrypoint.sh

@@ -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

internal/config/config.go

@@ -6,6 +6,7 @@ import (
 	"os"
 	"path/filepath"
 	"runtime"
+	"strconv"
 	"time"
 )
 
@@ -18,6 +19,12 @@ 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
 )
 
 // Config holds configuration for the entire application
@@ -25,6 +32,9 @@ 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
+
 	// MaxRuntime is the maximum runtime (0 = run forever)
 	MaxRuntime time.Duration
 
@@ -43,8 +53,14 @@ func New() (*Config, error) {
 		return nil, fmt.Errorf("failed to determine state directory: %w", err)
 	}
 
+	port, err := getPort()
+	if err != nil {
+		return nil, err
+	}
+
 	return &Config{
 		StateDir:                    stateDir,
+		Port:                        port,
 		MaxRuntime:                  0,                                           // Run forever by default
 		EnableBatchedDatabaseWrites: true,                                        // Enable batching by default
 		RouteExpirationTimeout:      defaultRouteExpirationMinutes * time.Minute, // For active route monitoring
@@ -69,13 +85,20 @@ 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("XDG_DATA_HOME must be an absolute path, got %q", 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
 		}
 
@@ -92,6 +115,22 @@ func getStateDirectory() (string, error) {
 	}
 }
 
+// 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("PORT must be a whole number from 1 to %d, got %q", maxPort, value)
+	}
+
+	return int(port), nil
+}
+
 // EnsureDirectories creates all necessary directories if they don't exist
 func (c *Config) EnsureDirectories() error {
 	// Ensure state directory exists

internal/config/config_test.go

@@ -0,0 +1,62 @@
+package config
+
+import (
+	"runtime"
+	"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", "")
+
+			if _, err := New(); err == nil {
+				t.Errorf("New() with PORT=%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")
+	if _, err := New(); err == nil {
+		t.Error("New() with a relative XDG_DATA_HOME returned no error")
+	}
+
+	t.Setenv("XDG_DATA_HOME", "/var/lib")
+	if _, err := New(); err != nil {
+		t.Errorf("New() with XDG_DATA_HOME=/var/lib: %v", err)
+	}
+}

internal/database/counts.go

@@ -0,0 +1,203 @@
+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.
+	if err := d.db.QueryRowContext(ctx, "SELECT COUNT(*) FROM "+table).Scan(&n); 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, "live_routes_v4")
+	if err != nil {
+		return err
+	}
+	routesV6, err := d.countRows(ctx, "live_routes_v6")
+	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_test.go

@@ -0,0 +1,511 @@
+package database
+
+import (
+	"context"
+	"slices"
+	"sync"
+	"testing"
+	"time"
+
+	"git.eeqj.de/sneak/routewatch/internal/config"
+	"git.eeqj.de/sneak/routewatch/internal/logger"
+	"github.com/google/uuid"
+)
+
+// 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{
+		ID:          uuid.New(),
+		Prefix:      prefix,
+		MaskLength:  maskLength,
+		IPVersion:   ipVersionV4,
+		OriginASN:   asn,
+		PeerIP:      "192.0.2.1",
+		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{
+		ID:          uuid.New(),
+		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) {
+	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)
+
+	if err := db.GetOrCreateASNBatch(map[int]time.Time{64500: ts, 64501: ts}); err != nil {
+		t.Fatalf("GetOrCreateASNBatch: %v", err)
+	}
+	if err := db.UpdatePrefixesBatch(map[string]time.Time{
+		"198.51.100.0/24": ts,
+		"2001:db8::/32":   ts,
+	}); err != nil {
+		t.Fatalf("UpdatePrefixesBatch: %v", err)
+	}
+	if err := db.UpdatePeerBatch(map[string]PeerUpdate{
+		"192.0.2.1": {PeerIP: "192.0.2.1", PeerASN: 64500, MessageType: "UPDATE", Timestamp: ts},
+	}); err != nil {
+		t.Fatalf("UpdatePeerBatch: %v", err)
+	}
+	if err := db.RecordPeering(64500, 64501, ts); err != nil {
+		t.Fatalf("RecordPeering: %v", err)
+	}
+
+	routes := []*LiveRoute{
+		mkV4Route(t, "198.51.100.0/24", 64500, ts),
+		mkV4Route(t, "203.0.113.0/24", 64501, ts.Add(time.Minute)),
+		mkV6Route("2001:db8::/32", 64502, ts.Add(2*time.Minute)),
+	}
+	if err := db.UpsertLiveRouteBatch(routes); 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.
+	if err := db.UpsertLiveRouteBatch(routes); 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.
+	if err := db.DeleteLiveRouteBatch([]LiveRouteDeletion{
+		{Prefix: "203.0.113.0/24", OriginASN: 64501, PeerIP: "192.0.2.1", IPVersion: ipVersionV4},
+	}); 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,
+	})
+}
+
+// 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) {
+	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)
+	if err := db.GetOrCreateASNBatch(map[int]time.Time{64500: ts, 64501: ts, 64502: ts}); err != nil {
+		t.Fatalf("GetOrCreateASNBatch: %v", err)
+	}
+	if err := db.UpsertLiveRouteBatch([]*LiveRoute{
+		mkV4Route(t, "198.51.100.0/24", 64500, ts),
+		mkV6Route("2001:db8::/32", 64502, ts),
+	}); err != nil {
+		t.Fatalf("UpsertLiveRouteBatch: %v", err)
+	}
+	if err := db.Close(); 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) {
+	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("2001:db8::/32", 64503, ts)
+	v6SecondPeer := mkV6Route("2001:db8::/32", 64503, ts)
+	v6SecondPeer.PeerIP = "2001:db8::2"
+
+	all := []PrefixDistribution{{MaskLength: 16, Count: 1}, {MaskLength: 24, Count: 2}}
+	v6Only := []PrefixDistribution{{MaskLength: 32, Count: 1}}
+
+	steps := []struct {
+		name     string
+		announce []*LiveRoute
+		withdraw []*LiveRoute
+		wantV4   []PrefixDistribution
+		wantV6   []PrefixDistribution
+	}{
+		{
+			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,
+		},
+	}
+
+	for _, batch := range []bool{true, false} {
+		name := "single-route writes"
+		if batch {
+			name = "batch writes"
+		}
+
+		t.Run(name, func(t *testing.T) {
+			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 {
+				if err := announceRoutes(db, batch, step.announce); err != nil {
+					t.Fatalf("%s: announce: %v", step.name, err)
+				}
+				if err := withdrawRoutes(db, batch, step.withdraw); 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)
+			}
+		})
+	}
+}
+
+// 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 {
+		if err := db.UpsertLiveRoute(route); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// withdrawRoutes removes routes in one DeleteLiveRouteBatch, or with one
+// DeleteLiveRoute each. It names each route by prefix and peer only, with no
+// origin ASN, as a withdrawal from the feed does when its message carries no AS
+// path.
+func withdrawRoutes(db *Database, batch bool, routes []*LiveRoute) error {
+	if !batch {
+		for _, route := range routes {
+			if err := db.DeleteLiveRoute(route.Prefix, 0, route.PeerIP); 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)
+}
+
+// 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) {
+	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)
+	if 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("2001:db8::/32", 64502, newest),
+	}); 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) {
+	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("2001:db8::/32", asn, ts)
+				route.PeerIP = "2001:db8::" + uuid.NewString()
+				if err := db.UpsertLiveRoute(route); 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:
+				if _, err := db.GetStatsContext(context.Background()); 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_test.go

@@ -3,23 +3,36 @@ package database
 import (
 	"context"
 	"database/sql"
-	"net"
+	"errors"
+	"fmt"
+	"net/netip"
 	"sync"
 	"testing"
 	"time"
 
 	"git.eeqj.de/sneak/routewatch/internal/config"
 	"git.eeqj.de/sneak/routewatch/internal/logger"
+	"github.com/google/uuid"
 )
 
 // 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
 
+// autoVacuumIncremental is the PRAGMA auto_vacuum value meaning "incremental".
+const autoVacuumIncremental = 2
+
+// vacuumTestRoutes is how many routes the vacuum test writes and then deletes,
+// enough to leave many free pages in the file.
+const vacuumTestRoutes = 2000
+
 // 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
 
+// testPeerIP is the peer every route in the IP lookup test is learned from.
+const testPeerIP = "192.0.2.254"
+
 // Parameters for the checkpoint-contention regression test.
 const (
 	// contentionIterations is how many batch writes race the checkpoint loop.
@@ -32,286 +45,98 @@ const (
 	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,
-		},
-	}
+// TestGetIPInfoFindsMostSpecificLiveRoute stores nested live prefixes for both
+// families and checks that a lookup returns the most specific one covering the
+// address, ErrNoRoute when none covers it, and the next less specific prefix
+// once the only route of the most specific one is withdrawn.
+func TestGetIPInfoFindsMostSpecificLiveRoute(t *testing.T) {
+	cfg := &config.Config{StateDir: t.TempDir()}
 
-	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)
+	db, err := New(cfg, logger.New())
+	if err != nil {
+		t.Fatalf("failed to create database: %v", err)
 	}
+	defer func() { _ = db.Close() }()
 
-			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)
-				}
+	// Nested live prefixes, each originated by its own AS.
+	origins := map[string]int{
+		"10.0.0.0/8":        64500,
+		"10.1.0.0/16":       64501,
+		"10.1.2.0/24":       64502,
+		"2001:db8::/32":     64500,
+		"2001:db8:1::/48":   64501,
+		"2001:db8:1:2::/64": 64502,
 	}
+	ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
+	routes := make([]*LiveRoute, 0, len(origins))
+	for prefix, asn := range origins {
+		routes = append(routes, &LiveRoute{
+			ID:          uuid.New(),
+			Prefix:      prefix,
+			MaskLength:  netip.MustParsePrefix(prefix).Bits(),
+			IPVersion:   detectIPVersion(prefix),
+			OriginASN:   asn,
+			PeerIP:      testPeerIP,
+			ASPath:      []int{asn},
+			NextHop:     testPeerIP,
+			LastUpdated: ts,
 		})
 	}
-}
-
-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,
-		},
+	if err := db.UpsertLiveRouteBatch(routes); err != nil {
+		t.Fatalf("UpsertLiveRouteBatch: %v", err)
 	}
 
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			start, end, err := CalculateIPv4Range(tt.cidr)
+	// lookup checks that ip resolves to the live prefix want, or to ErrNoRoute
+	// when want is empty.
+	lookup := func(ip, want string) {
+		t.Helper()
 
-			if tt.wantErr {
-				if err == nil {
-					t.Errorf("CalculateIPv4Range(%s) expected error, got nil", tt.cidr)
+		info, err := db.GetIPInfo(ip)
+		if want == "" {
+			if !errors.Is(err, ErrNoRoute) {
+				t.Errorf("GetIPInfo(%s) = %+v, %v; want ErrNoRoute", ip, info, err)
 			}
+
 			return
 		}
-
 		if err != nil {
-				t.Errorf("CalculateIPv4Range(%s) unexpected error: %v", tt.cidr, err)
+			t.Errorf("GetIPInfo(%s): %v", ip, err)
+
 			return
 		}
-
-			if start != tt.wantStart {
-				t.Errorf("CalculateIPv4Range(%s) start = %d, want %d", tt.cidr, start, tt.wantStart)
+		if info.Netblock != want || info.MaskLength != netip.MustParsePrefix(want).Bits() ||
+			info.ASN != origins[want] {
+			t.Errorf("GetIPInfo(%s) = %s (mask %d) AS%d, want %s AS%d",
+				ip, info.Netblock, info.MaskLength, info.ASN, want, origins[want])
+		}
 	}
 
-			if end != tt.wantEnd {
-				t.Errorf("CalculateIPv4Range(%s) end = %d, want %d", tt.cidr, end, tt.wantEnd)
-			}
+	lookup("10.1.2.3", "10.1.2.0/24")
+	lookup("::ffff:10.1.2.3", "10.1.2.0/24")
+	lookup("10.1.3.4", "10.1.0.0/16")
+	lookup("10.2.0.1", "10.0.0.0/8")
+	lookup("192.0.2.1", "")
+	lookup("2001:db8:1:2::3", "2001:db8:1:2::/64")
+	lookup("2001:db8:1:3::4", "2001:db8:1::/48")
+	lookup("2001:db8:2::1", "2001:db8::/32")
+	lookup("2001:db9::1", "")
 
-			// 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)
-					}
-				}
-			}
+	err = db.DeleteLiveRouteBatch([]LiveRouteDeletion{
+		{Prefix: "10.1.2.0/24", OriginASN: 64502, PeerIP: testPeerIP, IPVersion: ipVersionV4},
+		{Prefix: "2001:db8:1:2::/64", OriginASN: 64502, PeerIP: testPeerIP, IPVersion: ipVersionV6},
 	})
-	}
-}
-
-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)
+		t.Fatalf("DeleteLiveRouteBatch: %v", 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])
-				}
-			}
-		})
-	}
+	lookup("10.1.2.3", "10.1.0.0/16")
+	lookup("2001:db8:1:2::3", "2001:db8:1::/48")
 }
 
 // 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.
+// process-wide hard heap limit, and sees the new file with auto_vacuum
+// incremental.
 func TestConnectionPoolPragmas(t *testing.T) {
 	cfg := &config.Config{StateDir: t.TempDir()}
 
@@ -372,6 +197,74 @@ func TestConnectionPoolPragmas(t *testing.T) {
 		if hardHeapLimit != sqliteHardHeapLimitBytes {
 			t.Errorf("conn %d: hard_heap_limit = %d, want %d", i, hardHeapLimit, sqliteHardHeapLimitBytes)
 		}
+
+		var autoVacuum int
+		if err := c.QueryRowContext(ctx, "PRAGMA auto_vacuum").Scan(&autoVacuum); err != nil {
+			t.Fatalf("conn %d: failed to read auto_vacuum: %v", i, err)
+		}
+		if autoVacuum != autoVacuumIncremental {
+			t.Errorf("conn %d: auto_vacuum = %d, want %d (incremental)", i, autoVacuum, autoVacuumIncremental)
+		}
+	}
+}
+
+// TestVacuumReturnsFreePages checks that after routes are deleted from a new
+// database, one Vacuum call returns every page they used. The deletes leave
+// fewer free pages than the 1000 Vacuum frees per call, so none may remain.
+// With auto_vacuum off (issue https://git.eeqj.de/sneak/routewatch/issues/43)
+// the free pages stayed in the file, and with the PRAGMA run by ExecContext
+// Vacuum freed only one page per call.
+func TestVacuumReturnsFreePages(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()
+	ts := time.Now().UTC()
+	const asn = 64500
+
+	routes := make([]*LiveRoute, 0, vacuumTestRoutes)
+	deletions := make([]LiveRouteDeletion, 0, vacuumTestRoutes)
+	for i := range vacuumTestRoutes {
+		route := mkV4Route(t, fmt.Sprintf("10.%d.%d.0/24", i/256, i%256), asn, ts)
+		routes = append(routes, route)
+		deletions = append(deletions, LiveRouteDeletion{
+			Prefix:    route.Prefix,
+			OriginASN: asn,
+			PeerIP:    route.PeerIP,
+			IPVersion: ipVersionV4,
+		})
+	}
+
+	if err := db.UpsertLiveRouteBatch(routes); err != nil {
+		t.Fatalf("failed to write routes: %v", err)
+	}
+	if err := db.DeleteLiveRouteBatch(deletions); err != nil {
+		t.Fatalf("failed to delete routes: %v", err)
+	}
+
+	var before int
+	if err := db.db.QueryRowContext(ctx, "PRAGMA freelist_count").Scan(&before); err != nil {
+		t.Fatalf("failed to read freelist_count: %v", err)
+	}
+	if before == 0 {
+		t.Fatalf("no free pages after deleting %d routes", vacuumTestRoutes)
+	}
+
+	if err := db.Vacuum(ctx); err != nil {
+		t.Fatalf("Vacuum failed: %v", err)
+	}
+
+	var after int
+	if err := db.db.QueryRowContext(ctx, "PRAGMA freelist_count").Scan(&after); err != nil {
+		t.Fatalf("failed to read freelist_count: %v", err)
+	}
+	if after != 0 {
+		t.Errorf("free pages after Vacuum = %d of %d, want 0", after, before)
 	}
 }
 
@@ -424,21 +317,3 @@ func TestBatchWriteDuringCheckpoint(t *testing.T) {
 	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

@@ -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
@@ -61,8 +63,6 @@ type Store interface {
 	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)
 

internal/database/models.go

@@ -77,9 +77,6 @@ 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
@@ -88,16 +85,6 @@ type PrefixDistribution struct {
 	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
 type IPInfo struct {
 	IP         string   `json:"ip"`

internal/database/schema.sql

@@ -107,9 +107,6 @@ 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)
 );
 
@@ -123,7 +120,6 @@ 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)
 );
 
@@ -132,8 +128,6 @@ CREATE INDEX IF NOT EXISTS idx_live_routes_v4_prefix ON live_routes_v4(prefix);
 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);
 

internal/database/utils.go

@@ -1,6 +1,8 @@
 package database
 
 import (
+	"fmt"
+	"net"
 	"strings"
 
 	"github.com/google/uuid"
@@ -18,3 +20,14 @@ 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/routewatch/app_integration_test.go

@@ -232,25 +232,6 @@ func (m *mockStore) GetLiveRouteCountsContext(ctx context.Context) (ipv4Count, i
 	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()
@@ -450,7 +431,7 @@ func TestRouteWatchLiveFeed(t *testing.T) {
 	}
 
 	// Create server
-	srv := server.New(mockDB, s, logger)
+	srv := server.New(mockDB, s, logger, cfg)
 
 	// Create RouteWatch with 5 second limit
 	deps := Dependencies{

internal/routewatch/prefixhandler.go

@@ -2,6 +2,7 @@ package routewatch
 
 import (
 	"net"
+	"net/netip"
 	"strings"
 	"sync"
 	"time"
@@ -108,7 +109,7 @@ func (h *PrefixHandler) HandleMessage(msg *ristypes.RISMessage) {
 	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",
@@ -125,7 +126,7 @@ func (h *PrefixHandler) HandleMessage(msg *ristypes.RISMessage) {
 	// Process withdrawals
 	for _, prefix := range msg.Withdrawals {
 		h.batch = append(h.batch, prefixUpdate{
-			prefix:      prefix,
+			prefix:      canonicalPrefix(prefix),
 			originASN:   originASN, // Use the originASN from path if available
 			peer:        msg.Peer,
 			messageType: "withdrawal",
@@ -264,6 +265,21 @@ func (h *PrefixHandler) flushBatchLocked() {
 	h.lastFlush = time.Now()
 }
 
+// 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 extracts the mask length and IP version from a prefix string
 func parseCIDR(prefix string) (maskLength int, ipVersion int, err error) {
 	_, ipNet, err := net.ParseCIDR(prefix)
@@ -315,20 +331,6 @@ func (h *PrefixHandler) processAnnouncement(_ *database.Prefix, update prefixUpd
 		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,
@@ -372,20 +374,6 @@ 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
 }
 
@@ -425,20 +413,6 @@ func (h *PrefixHandler) processAnnouncementDirect(update prefixUpdate) {
 		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,

internal/routewatch/prefixhandler_test.go

@@ -0,0 +1,74 @@
+package routewatch
+
+import (
+	"errors"
+	"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/ristypes"
+)
+
+const testPeerIP = "2001:db8:ffff::1"
+
+// TestPrefixHandlerStoresPrefixesTheIPLookupFinds runs announcements and
+// withdrawals through the prefix handler into a real database and looks the
+// addresses up. The prefixes are written the way the feed sends them: IPv6
+// announcements compressed, IPv6 withdrawals uncompressed. The withdrawal must
+// remove the route the announcement stored, and the IP lookup must find the
+// stored prefix.
+func TestPrefixHandlerStoresPrefixesTheIPLookupFinds(t *testing.T) {
+	db, err := database.New(&config.Config{StateDir: t.TempDir()}, logger.New())
+	if err != nil {
+		t.Fatalf("failed to create database: %v", err)
+	}
+	defer func() { _ = db.Close() }()
+
+	// Built without NewPrefixHandler's flush timer, so the test flushes itself.
+	h := &PrefixHandler{db: db, logger: logger.New()}
+	flush := func() {
+		h.mu.Lock()
+		defer h.mu.Unlock()
+		h.flushBatchLocked()
+	}
+
+	ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
+	h.HandleMessage(&ristypes.RISMessage{
+		Peer:            testPeerIP,
+		Path:            ristypes.ASPath{testASNA, testASNB},
+		ParsedTimestamp: ts,
+		Announcements: []ristypes.RISAnnouncement{{
+			NextHop:  testPeerIP,
+			Prefixes: []string{"2001:db8:1::/48", "192.0.2.0/24"},
+		}},
+	})
+	flush()
+
+	for ip, want := range map[string]string{
+		"2001:db8:1::1": "2001:db8:1::/48",
+		"192.0.2.1":     "192.0.2.0/24",
+	} {
+		info, err := db.GetIPInfo(ip)
+		if err != nil {
+			t.Fatalf("GetIPInfo(%s) after announcement: %v", ip, err)
+		}
+		if info.Netblock != want || info.ASN != testASNB {
+			t.Errorf("GetIPInfo(%s) = %s AS%d, want %s AS%d", ip, info.Netblock, info.ASN, want, testASNB)
+		}
+	}
+
+	h.HandleMessage(&ristypes.RISMessage{
+		Peer:            testPeerIP,
+		ParsedTimestamp: ts.Add(time.Minute),
+		Withdrawals:     []string{"2001:db8:1:0:0:0:0:0/48", "192.0.2.0/24"},
+	})
+	flush()
+
+	for _, ip := range []string{"2001:db8:1::1", "192.0.2.1"} {
+		if info, err := db.GetIPInfo(ip); !errors.Is(err, database.ErrNoRoute) {
+			t.Errorf("GetIPInfo(%s) after withdrawal = %+v, %v; want ErrNoRoute", ip, info, err)
+		}
+	}
+}

internal/server/handlers.go

@@ -219,13 +219,6 @@ func (s *Server) handleStatusJSON() http.HandlerFunc {
 
 		const bitsPerMegabit = 1000000.0
 
-		// Get route counts from database
-		ipv4Routes, ipv6Routes, err := s.db.GetLiveRouteCountsContext(ctx)
-		if err != nil {
-			s.logger.Warn("Failed to get live route counts", "error", err)
-			// Continue with zero counts
-		}
-
 		// Get route update metrics
 		routeMetrics := s.streamer.GetMetricsTracker().GetRouteMetrics()
 
@@ -259,8 +252,8 @@ func (s *Server) handleStatusJSON() http.HandlerFunc {
 			Peers:                  dbStats.Peers,
 			DatabaseSizeBytes:      dbStats.FileSizeBytes,
 			LiveRoutes:             dbStats.LiveRoutes,
-			IPv4Routes:             ipv4Routes,
-			IPv6Routes:             ipv6Routes,
+			IPv4Routes:             dbStats.IPv4Routes,
+			IPv6Routes:             dbStats.IPv6Routes,
 			OldestRoute:            dbStats.OldestRoute,
 			NewestRoute:            dbStats.NewestRoute,
 			IPv4UpdatesPerSec:      routeMetrics.IPv4UpdatesPerSec,
@@ -439,13 +432,6 @@ func (s *Server) handleStats() http.HandlerFunc {
 
 		const bitsPerMegabit = 1000000.0
 
-		// Get route counts from database
-		ipv4Routes, ipv6Routes, err := s.db.GetLiveRouteCountsContext(ctx)
-		if err != nil {
-			s.logger.Warn("Failed to get live route counts", "error", err)
-			// Continue with zero counts
-		}
-
 		// Get route update metrics
 		routeMetrics := s.streamer.GetMetricsTracker().GetRouteMetrics()
 
@@ -537,8 +523,8 @@ func (s *Server) handleStats() http.HandlerFunc {
 			Peers:                  dbStats.Peers,
 			DatabaseSizeBytes:      dbStats.FileSizeBytes,
 			LiveRoutes:             dbStats.LiveRoutes,
-			IPv4Routes:             ipv4Routes,
-			IPv6Routes:             ipv6Routes,
+			IPv4Routes:             dbStats.IPv4Routes,
+			IPv6Routes:             dbStats.IPv6Routes,
 			OldestRoute:            dbStats.OldestRoute,
 			NewestRoute:            dbStats.NewestRoute,
 			IPv4UpdatesPerSec:      routeMetrics.IPv4UpdatesPerSec,

internal/server/handlers_test.go

@@ -2,16 +2,20 @@ package server
 
 import (
 	"context"
+	"encoding/json"
 	"net/http"
 	"net/http/httptest"
 	"runtime"
+	"slices"
 	"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/streamer"
+	"github.com/google/uuid"
 )
 
 // blockingStatsDB embeds database.Store (left nil) and overrides only
@@ -38,7 +42,7 @@ func (d blockingStatsDB) GetStatsContext(_ context.Context) (database.Stats, err
 func TestStatsHandlersDoNotLeakOnTimeout(t *testing.T) {
 	release := make(chan struct{})
 	db := blockingStatsDB{release: release}
-	s := New(db, streamer.New(logger.New(), metrics.New()), logger.New())
+	s := New(db, streamer.New(logger.New(), metrics.New()), logger.New(), &config.Config{})
 
 	handlers := map[string]http.HandlerFunc{
 		"status.json": s.handleStatusJSON(),
@@ -70,6 +74,80 @@ func TestStatsHandlersDoNotLeakOnTimeout(t *testing.T) {
 	}
 }
 
+// TestStatsHandlersAnswerFromMemory checks that both stats handlers answer 200
+// with the live route counts and the prefix distribution while the database is
+// closed, so that any query would fail: the request path reads them from
+// memory. The prefix distribution query it used to run read every live route
+// and, on a large database, took the whole 4-second deadline, so
+// /api/v1/stats answered 500 (https://git.eeqj.de/sneak/routewatch/issues/30).
+// The oldest and newest route times still come from one-row lookups at the ends
+// of an index; with the database closed they are left out of the answer.
+func TestStatsHandlersAnswerFromMemory(t *testing.T) {
+	db, err := database.New(&config.Config{StateDir: t.TempDir()}, logger.New())
+	if err != nil {
+		t.Fatalf("database.New: %v", err)
+	}
+
+	ts := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
+	if err := db.UpsertLiveRouteBatch([]*database.LiveRoute{
+		{
+			ID: uuid.New(), Prefix: "198.51.100.0/24", MaskLength: 24, IPVersion: 4,
+			OriginASN: 64500, PeerIP: "192.0.2.1", ASPath: []int{64500}, NextHop: "192.0.2.1",
+			LastUpdated: ts,
+		},
+		{
+			ID: uuid.New(), Prefix: "2001:db8::/32", MaskLength: 32, IPVersion: 6,
+			OriginASN: 64501, PeerIP: "2001:db8::1", ASPath: []int{64501}, NextHop: "2001:db8::1",
+			LastUpdated: ts,
+		},
+	}); err != nil {
+		t.Fatalf("UpsertLiveRouteBatch: %v", err)
+	}
+	if err := db.Close(); err != nil {
+		t.Fatalf("Close: %v", err)
+	}
+
+	s := New(db, streamer.New(logger.New(), metrics.New()), logger.New(), &config.Config{})
+	handlers := map[string]http.HandlerFunc{
+		"status.json": s.handleStatusJSON(),
+		"stats":       s.handleStats(),
+	}
+
+	for name, handler := range handlers {
+		rec := httptest.NewRecorder()
+		handler(rec, httptest.NewRequest(http.MethodGet, "/", nil))
+		if rec.Code != http.StatusOK {
+			t.Errorf("%s: status %d, want %d; body %s", name, rec.Code, http.StatusOK, rec.Body)
+
+			continue
+		}
+
+		var body struct {
+			Data struct {
+				IPv4Routes             int                           `json:"ipv4_routes"`
+				IPv6Routes             int                           `json:"ipv6_routes"`
+				IPv4PrefixDistribution []database.PrefixDistribution `json:"ipv4_prefix_distribution"`
+				IPv6PrefixDistribution []database.PrefixDistribution `json:"ipv6_prefix_distribution"`
+			} `json:"data"`
+		}
+		if err := json.Unmarshal(rec.Body.Bytes(), &body); err != nil {
+			t.Fatalf("%s: decoding the answer: %v", name, err)
+		}
+
+		if body.Data.IPv4Routes != 1 || body.Data.IPv6Routes != 1 {
+			t.Errorf("%s: routes = (v4 %d, v6 %d), want (1, 1)", name, body.Data.IPv4Routes, body.Data.IPv6Routes)
+		}
+		wantV4 := []database.PrefixDistribution{{MaskLength: 24, Count: 1}}
+		if !slices.Equal(body.Data.IPv4PrefixDistribution, wantV4) {
+			t.Errorf("%s: IPv4 distribution = %v, want %v", name, body.Data.IPv4PrefixDistribution, wantV4)
+		}
+		wantV6 := []database.PrefixDistribution{{MaskLength: 32, Count: 1}}
+		if !slices.Equal(body.Data.IPv6PrefixDistribution, wantV6) {
+			t.Errorf("%s: IPv6 distribution = %v, want %v", name, body.Data.IPv6PrefixDistribution, wantV6)
+		}
+	}
+}
+
 // settledGoroutineCount lets transient goroutines finish, then reports the
 // current count.
 func settledGoroutineCount() int {

internal/server/server.go

@@ -4,9 +4,10 @@ package server
 import (
 	"context"
 	"net/http"
-	"os"
+	"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"
@@ -33,16 +34,18 @@ type Server struct {
 	db         database.Store
 	streamer   *streamer.Streamer
 	logger     *logger.Logger
+	port       int
 	srv        *http.Server
 	asnFetcher ASNFetcher
 }
 
 // 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,
+		port:     cfg.Port,
 	}
 
 	s.setupRoutes()
@@ -52,11 +55,6 @@ func New(db database.Store, streamer *streamer.Streamer, logger *logger.Logger)
 
 // Start starts the HTTP server
 func (s *Server) Start() error {
-	port := os.Getenv("PORT")
-	if port == "" {
-		port = "8080"
-	}
-
 	const (
 		readHeaderTimeout = 40 * time.Second
 		readTimeout       = 60 * time.Second
@@ -65,7 +63,7 @@ func (s *Server) Start() error {
 	)
 
 	s.srv = &http.Server{
-		Addr:              ":" + port,
+		Addr:              ":" + strconv.Itoa(s.port),
 		Handler:           s.router,
 		ReadHeaderTimeout: readHeaderTimeout,
 		ReadTimeout:       readTimeout,
@@ -73,7 +71,7 @@ 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)
 
 	// Start in goroutine but log when actually listening
 	go func() {

internal/streamer/streamer.go

@@ -210,9 +210,14 @@ func (s *Streamer) Start() error {
 // the connection status in metrics. This method is safe to call multiple times.
 func (s *Streamer) Stop() {
 	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
 	}
+	s.cancel()
+	s.cancel = nil
 	// Close all handler queues to signal workers to stop
 	for _, info := range s.handlers {
 		close(info.queue)
@@ -660,8 +665,15 @@ func (s *Streamer) stream(ctx context.Context) error {
 			continue
 		}
 
-		// Dispatch to interested handlers
+		// Dispatch to interested handlers. 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

internal/streamer/streamer_test.go

@@ -2,6 +2,8 @@ package streamer
 
 import (
 	"context"
+	"errors"
+	"io"
 	"net/http"
 	"net/http/httptest"
 	"runtime"
@@ -10,6 +12,7 @@ import (
 
 	"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) {
@@ -75,6 +78,45 @@ func TestStreamDoesNotLeakTickersAcrossReconnects(t *testing.T) {
 	}
 }
 
+// updateHandler wants UPDATE messages and does nothing with them.
+type updateHandler struct{}
+
+func (updateHandler) WantsMessage(messageType string) bool { return messageType == "UPDATE" }
+func (updateHandler) HandleMessage(*ristypes.RISMessage)   {}
+func (updateHandler) QueueCapacity() int                   { return 10 }
+
+// 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) {
+	const line = `{"type":"ris_message","data":{"type":"UPDATE","peer":"192.0.2.1",` +
+		`"peer_asn":"64496","timestamp":1700000000}}` + "\n"
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+		_, _ = io.WriteString(w, line)
+	}))
+	defer srv.Close()
+
+	s := New(logger.New(), metrics.New())
+	s.url = srv.URL
+	s.RegisterHandler(updateHandler{})
+	s.RegisterRawHandler(func(string) { s.Stop() })
+
+	// 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
+
+	if err := s.stream(ctx); !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()
+}
+
 // settledGoroutineCount lets transient goroutines finish, then reports the
 // current count.
 func settledGoroutineCount() int {

internal/version/version.go

@@ -7,7 +7,8 @@ package version
 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"
 )
 

script/bootstrap

@@ -3,7 +3,8 @@
 # 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).
+# make, or go). The linter and the tests run only in Docker, as phases of
+# the Dockerfile, so nothing installs golangci-lint here.
 set -eu
 
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
@@ -58,11 +59,10 @@ main() {
     # 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
+    # Warn, don't fail: make build works without it.
+    if missing docker; then
+        echo "bootstrap: WARNING: docker not found; install it to run" \
+            "make test, make lint, make check and make docker." >&2
     fi
 
     go mod download

script/check

@@ -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

@@ -1,14 +1,28 @@
 #!/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.
+# --no-cache for the same reason as script/docker: the gate phases the
+# final stage depends on are RUN steps, and a cached one is a check that
+# did not run.
 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"
+    "$SCRIPT_DIR/check"
+    # Own line: a failing command substitution inside an argument does
+    # not trip `set -e`, so the inline form degrades silently to an
+    # empty constant. The VERSION build argument takes precedence over
+    # the version a build stage derives from the .git in the context.
+    version="$(git describe --tags --always --dirty 2>/dev/null || true)"
+    [ -n "$version" ] || version="unknown"
+    docker build --no-cache \
+        --build-arg VERSION="$version" \
+        -t "$("$SCRIPT_DIR/projectname")" .
 }
 
 main "$@"

script/docker

@@ -1,7 +1,8 @@
 #!/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.
+# --no-cache because the gate phases the final stage depends on are RUN
+# steps, and a cached one is a check that did not run.
 set -eu
 
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -9,7 +10,15 @@ ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    docker build -t "$("$SCRIPT_DIR/projectname")" .
+    # Own line: a failing command substitution inside an argument does
+    # not trip `set -e`, so the inline form degrades silently to an
+    # empty constant. The VERSION build argument takes precedence over
+    # the version a build stage derives from the .git in the context.
+    version="$(git describe --tags --always --dirty 2>/dev/null || true)"
+    [ -n "$version" ] || version="unknown"
+    docker build --no-cache \
+        --build-arg VERSION="$version" \
+        -t "$("$SCRIPT_DIR/projectname")" .
 }
 
 main "$@"

script/lint

@@ -1,13 +1,23 @@
 #!/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. --no-cache because a cached lint layer is a lint
+# that did not run. The tag makes each build replace the previous image
+# instead of leaving a dangling one behind.
 set -eu
 
-ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
+ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    go vet ./...
-    golangci-lint run
+    docker build --no-cache \
+        --target lint \
+        -t "$("$SCRIPT_DIR/projectname")-lint" .
 }
 
 main "$@"

script/test

@@ -1,17 +1,19 @@
 #!/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, --no-cache because a cached test layer is a test that did not
+# run, and a tag so each build replaces the previous image.
 set -eu
 
-ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
+ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    go test -short -timeout 30s -race -cover ./... || {
-        echo "--- Rerunning with -v for details ---"
-        go test -short -timeout 30s -race -v ./...
-        exit 1
-    }
+    docker build --no-cache \
+        --target test \
+        -t "$("$SCRIPT_DIR/projectname")-test" .
 }
 
 main "$@"