Merge branch 'main' into next

README.md is rewritten from scratch so that every sentence reads on the first pass, and is cut from about 5,600 words to about 3,500. It leads with what netwatch is, what it is for and how to run it, then the page, reporting and deployment, then every setting in a reference section, with design and development last. Mechanism detail the code and its comments already hold is left out; the PR lists each topic dropped and why. TODO.md records the rewrite of the three documents.

Model: fable-5-1 (rewrite, review); opus-5-5 (merge)
Co-authored-by: clawbot <clawbot@noreply.example.org>

.dockerignore

@@ -73,9 +73,12 @@
 
 # This repository's own entries, after the shared content above: the
 # frontend build, the viewport test's output, the binary `make build`
-# writes and the reports `make run` stores, and log files at the root.
+# writes and the reports `make run` stores, and Go test binaries,
+# coverage output and log files at the root.
 /dist
 /tmp
 /netwatch-server
 /data
+/*.test
+/*.out
 /*.log

.editorconfig

@@ -11,7 +11,8 @@ insert_final_newline = true
 [Makefile]
 indent_style = tab
 
-# This repository's own entries, after the shared content above.
-
 [*.go]
 indent_style = tab
+
+# This repository's own sections, such as one for another language it
+# uses, go below this comment, and a re-vendor keeps them.

.gitea/workflows/check.yml

@@ -1,9 +1,20 @@
 name: check
 on: [push]
+# Free the shared runner: a new push cancels only the same branch's older run.
+concurrency:
+    group: ${{ github.workflow }}-${{ github.ref }}
+    cancel-in-progress: true
 jobs:
     check:
         runs-on: ubuntu-latest
+        # Free the shared runner from a hung build.
+        timeout-minutes: 20
         steps:
             # actions/checkout v4.2.2, 2026-02-22
             - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
+              # script/cibuild needs no token, so none is left in .git/config.
+              with:
+                  persist-credentials: false
+                  # All history and tags, so git describe finds the version tag.
+                  fetch-depth: 0
             - run: script/cibuild

.gitignore

@@ -27,7 +27,7 @@ node_modules/
 # 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`.
+# its own negation at the end of this file, for example `!.env.example`.
 *.[eE][nN][vV]
 .[eE][nN][vV].*
 .[eE][nN][vV][rR][cC]
@@ -46,7 +46,9 @@ node_modules/
 [iI][dD]_[eE][dD]25519
 [iI][dD]_[eE][dD]25519_[sS][kK]
 
-# This repository's own entries, after the shared content above.
+# This repository's own entries, such as its build outputs, go below
+# this comment, and a re-vendor keeps them. Anchor a binary built at the
+# root: `/myapp`, never `myapp`, which also ignores `cmd/myapp/`.
 
 # Build output
 dist/

.golangci.yml

@@ -25,6 +25,8 @@ linters:
     # silenced by disabling that name, not by enabling the successor.
     - wsl              # Deprecated, replaced by wsl_v5
     - gomodguard       # Deprecated, replaced by gomodguard_v2
+    # Misses findings at random in v2.14.0; back once a pinned release fixes it
+    - canonicalheader
   settings:
     lll:
       line-length: 88

Dockerfile

@@ -10,11 +10,22 @@
 # phase covers the frontend as well, through a copy from a node stage.
 # Inside them each tool is invoked directly, never through make or
 # script/, whose lint and test are themselves docker builds.
+#
+# In every stage ARG CACHEBUST comes right after the steps that install
+# system packages and download Go's modules, or right after FROM in a
+# stage with neither, and every step that lints, tests or compiles
+# comes after it, as does yarn install in the node stages. script/
+# passes a new value on every build, which reruns every RUN step after
+# it. Docker keeps the layers above it until the base image, their own
+# lines or go.mod and go.sum change. The steps that compile mount Go's
+# build cache, and golangci-lint run also mounts golangci-lint's; both
+# survive from build to build.
 
 # Frontend lint stage: eslint with the rules in eslint.config.js. The
 # lint phase below runs it.
 # node:22-alpine as of 2026-02-22
 FROM node@sha256:e4bf2a82ad0a4037d28035ae71529873c069b13eb0455466ae0bc13363826e34 AS frontend-lint
+ARG CACHEBUST
 WORKDIR /app
 COPY package.json yarn.lock ./
 RUN yarn install --frozen-lockfile
@@ -29,8 +40,11 @@ FROM golangci/golangci-lint@sha256:ad862ba6b3798cbe0fd9fd7408d498fd74fbd2623a924
 WORKDIR /src
 COPY go.mod go.sum ./
 RUN go mod download
+ARG CACHEBUST
 COPY . .
-RUN golangci-lint run --config .golangci.yml ./...
+RUN --mount=type=cache,target=/root/.cache/go-build \
+    --mount=type=cache,target=/root/.cache/golangci-lint \
+    golangci-lint run --config .golangci.yml ./...
 # Nothing is wanted from frontend-lint; the copy is what makes this
 # phase run it.
 COPY --from=frontend-lint /app/yarn.lock /dev/null
@@ -43,6 +57,7 @@ COPY --from=frontend-lint /app/yarn.lock /dev/null
 # node would take a reporter option for one more file.
 # node:22-alpine as of 2026-02-22
 FROM node@sha256:e4bf2a82ad0a4037d28035ae71529873c069b13eb0455466ae0bc13363826e34 AS frontend-test
+ARG CACHEBUST
 WORKDIR /app
 COPY package.json yarn.lock ./
 RUN yarn install --frozen-lockfile
@@ -58,19 +73,19 @@ RUN NODE_OPTIONS=--test-reporter=dot timeout 90 yarn --silent run test || \
 # ships and the alpine one does not. -timeout 90s is a backstop above
 # the 60-second cap on the suite. The rerun with -v only shows details:
 # the step fails however it ends, because the first run already failed.
-# Go's module and build caches are in memory (tmpfs) for the go test
+# -count=1 because the mounted cache also holds test results, which go
-# step alone, so the image make test tags does not carry them and the
+# test would otherwise replay.
-# build spends no time writing them into it. They start empty on every
-# build, so no stored test result can stand in for a run.
 # golang:1.25.7-trixie, 2026-10-07
 FROM golang@sha256:2b174ffcf56c7ad0c47d30d2630693265639ddf2a5141149c2da34db921791b4 AS test
 WORKDIR /src
+COPY go.mod go.sum ./
+RUN go mod download
+ARG CACHEBUST
 COPY . .
-RUN --mount=type=tmpfs,target=/go/pkg/mod \
+RUN --mount=type=cache,target=/root/.cache/go-build \
-    --mount=type=tmpfs,target=/root/.cache/go-build \
+    go test -count=1 -timeout 90s -race -cover ./... || \
-    go test -timeout 90s -race -cover ./... || \
     { echo "--- Rerunning with -v for details ---"; \
-      go test -timeout 90s -race -v ./...; exit 1; }
+      go test -count=1 -timeout 90s -race -v ./...; exit 1; }
 # Nothing is wanted from the frontend-test stage; the copy is what makes
 # this phase run its tests.
 COPY --from=frontend-test /app/yarn.lock /dev/null
@@ -83,23 +98,28 @@ COPY --from=frontend-test /app/yarn.lock /dev/null
 FROM frontend-test AS frontend-build
 COPY --from=lint /src/go.sum /dev/null
 COPY --from=test /src/go.sum /dev/null
-# vite.config.js reads the commit for the page's footer with git.
+# vite.config.js reads the commit for the page's footer with git. git is
+# installed again on every image build, although it comes above
+# ARG CACHEBUST, because this stage starts from frontend-test, whose
+# steps run again on every build.
 RUN apk add --no-cache git
+ARG CACHEBUST
 RUN yarn build
 
 # Backend build stage. Nothing is wanted from the two phases; the copies
-# are what make BuildKit build them first, so this stage cannot run
+# are what make BuildKit build them first, so nothing after them can
-# unless lint and test passed.
+# run unless lint and test passed.
 # golang:1.25-alpine (2026-02-27)
 FROM golang@sha256:f6751d823c26342f9506c03797d2527668d095b0a15f1862cddb4d927a7a4ced 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
+ARG CACHEBUST
+COPY --from=lint /src/go.sum /dev/null
+COPY --from=test /src/go.sum /dev/null
+# A tar-stream context keeps the sender's file owners, which git refuses.
+RUN git config --system --add safe.directory /src
 COPY . .
 
 # The same go build as in script/build, which builds the binary on the
@@ -112,7 +132,8 @@ COPY . .
 # 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)}"; \
+RUN --mount=type=cache,target=/root/.cache/go-build \
+    version="${VERSION:-$(git describe --tags --always)}"; \
     if [ -e .git ]; then \
         case "$version" in ""|dev|unknown) \
             echo "version is '$version' although .git is present" >&2; \
@@ -127,6 +148,7 @@ RUN version="${VERSION:-$(git describe --tags --always)}"; \
 # and the stages it copies from, the two phases included.
 # nginx:stable-alpine as of 2026-02-22
 FROM nginx@sha256:15e96e59aa3b0aada3a121296e3bce117721f42d88f5f64217ef4b18f458c6ab
+ARG CACHEBUST
 
 # netwatch-server runs as this user, which owns the report directory.
 # nginx keeps the image's own arrangement: its main process runs as

README.md

@@ -1,8 +1,17 @@
-NetWatch is an MIT-licensed JavaScript single-page application by
+NetWatch is an MIT-licensed single-page web application by
-[@sneak](https://sneak.berlin) that provides real-time network latency
+[@sneak](https://sneak.berlin) that shows the latency from your browser to 26
-monitoring to common internet hosts, displayed with color-coded figures and
+well-known internet hosts. The page is JavaScript and does all its measuring in
-sparkline graphs, served from a static bucket or from its Docker image, where a
+the browser. A small Go backend, `netwatch-server`, stores the measurements the
-small Go backend stores the measurements the page reports.
+page reports. The page and the backend ship together in one Docker image.
+
+## Rationale
+
+When a network misbehaves, it helps to see at a glance which hosts answer and
+how fast. NetWatch is one page with a latency figure and a small graph for each
+host, refreshed every few seconds. The page measures from the browser, so it
+shows the path your own traffic takes. It can be served from any static file
+host. The backend only stores what the page reports. Without it, the page works
+the same and nothing is stored.
 
 ## Getting Started
 
@@ -10,10 +19,10 @@ small Go backend stores the measurements the page reports.
 # Install the dependencies and the git pre-commit hook
 make setup
 
-# Run the page on the Vite dev server
+# Serve the page from the Vite dev server
 make dev
 
-# Build netwatch-server and run it on port 8080
+# Build the page and netwatch-server, then run the server on port 8080
 make run
 
 # Run the tests, both linters and the format check
@@ -22,487 +31,417 @@ make check
 # Build the page into dist/ and netwatch-server into the repo root
 make build
 
-# Build the image and run it
+# Build the Docker image and run it
 make docker
 docker run -p 8080:8080 netwatch
 ```
 
-`make check` and `make docker` need Docker. `make dev` passes `/api` to
+`make check` and `make docker` need Docker. `make run` starts `netwatch-server`
-`http://127.0.0.1:8080`, where `make run` starts `netwatch-server` with its
+on port 8080 with its defaults, so the page's reports are stored in
-defaults, so the reports the page posts are stored in `data/reports`.
+`data/reports`. `make dev` passes the page's `/api` requests there.
 
-## Entrypoints
+## The page
 
-This repository adheres to the
+The page shows one row per host: the host's name, its URL, its latest latency in
-[Scripts to Rule Them All](https://github.com/github/scripts-to-rule-them-all)
+milliseconds and a graph of its last 100 checks. Below the figure are the
-standard: normalized scripts in `script/` are the entrypoints for the
+minimum, median, average and maximum of the host's answers in that history, or
-development workflow, and the Makefile targets are thin shims that call them.
+`timeout` or `unreachable` after a failed check. A round of checks runs every 3
-The scripts cover both the frontend and the Go backend, so `make check` fails if
+seconds by default. The header shows the interval, the history's span, and
-either one is broken. We provide:
+whether the page is running or paused.
 
-- `script/bootstrap` — install all dependencies (the pinned node via nvm unless
+### Targets
-  one new enough for the frontend's dependencies is installed, yarn via
-  corepack, `yarn install --frozen-lockfile`, Go `GO_VERSION` unless the `go` on
-  `PATH` is exactly that version, and the Go modules), linking what it installs
-  itself into `~/.local/bin`, which has to be on `PATH`. It installs no Go
-  linter and not Docker: `make test` and `make lint` run in Docker
-- `script/setup` — make a fresh clone ready for development: bootstrap plus the
-  git pre-commit hook
-- `script/dev` — run the Vite dev server, which proxies `/api` to a locally
-  running `netwatch-server`
-- `script/build` — build the frontend for production into `dist/`, then the
-  static `netwatch-server` binary into the repo root with its version stamped
-  in: `VERSION` from the environment, or when that is unset or empty,
-  `git describe` inside a git checkout, then `dev`
-- `script/run` — build with `script/build`, then run `netwatch-server` with its
-  defaults (see [Settings](#settings))
-- `script/clean` — remove what `script/build` writes
-- `script/projectname` — print the project name (used for the Docker image tag)
-- `script/test` — build the `test` phase of `Dockerfile` without the cache: the
-  frontend's unit tests in its `frontend-test` stage, and the backend's Go tests
-  with the race detector and coverage. The frontend's production build is left
-  to the image build, in the `frontend-build` stage
-- `script/lint` — build the `lint` phase of `Dockerfile` without the cache:
-  eslint in its `frontend-lint` stage, and golangci-lint over the Go packages
-- `script/fmt` — format all files (writes): prettier over the JavaScript, CSS,
-  HTML and Markdown, then gofmt over `cmd/` and `internal/`. It runs on the
-  host, as `script/fmt-check` does, with `~/.local/bin` put on `PATH`
-- `script/fmt-check` — check formatting (read-only): prettier, then gofmt
-- `script/check` — run test, lint, and fmt-check
-- `script/add-dependency` — add a frontend package, or move one to another
-  version: `make add-dependency PACKAGE=<name>@<version>` runs `yarn add --dev`,
-  which changes `package.json` and `yarn.lock` together, then
-  `yarn install --frozen-lockfile`
-- `script/tidy` — run `go mod tidy`: to add a Go module, import it and run
-  `make tidy`; to move one to another version, edit its `require` line in
-  `go.mod`, then run `make tidy`
-- `script/frontend-test` — run the unit tests in `test/unit/` on the host with
-  Node's built-in test runner, through the `test` script in `package.json`, and
-  if any fails, run them again listing every test, and fail; then the production
-  build. Each test run has a 90-second timeout. `make test` runs the same tests
-  in Docker, without the production build
-- `script/frontend-fmt` — format the JavaScript, CSS, HTML and Markdown with
-  prettier (writes)
-- `script/frontend-fmt-check` — check prettier formatting (read-only)
-- `script/frontend-viewport-test` — responsive-layout verification of the built
-  frontend in a containerised headless Chrome (see
-  [test/viewport/README.md](test/viewport/README.md)). Not part of
-  `script/check`: it takes minutes.
-- `script/docker` — build the image from `Dockerfile` without the build cache,
-  tagged `netwatch` via `script/projectname`
-- `script/cibuild` — CI entrypoint: runs `script/bootstrap` and `script/check`,
-  then builds the image as `script/docker` does, without the build cache
-- `script/precommit` — run by the git pre-commit hook; runs `script/check`
-- `script/install-precommit` — install the git pre-commit hook
 
-## Responsive layout
+The page checks 26 hosts on the internet, the WAN hosts:
 
-The narrow-viewport layout lives in the `max-width: 768px` media block in
+- `datavi.be`, pinned to the top at start
-`web/src/styles.css`. It is verified automatically by
+- the Anthropic API and the OpenAI API
-`make frontend-viewport-test`, which drives a digest-pinned headless Chrome
+- the AWS Console, the Google Cloud Console and Microsoft Azure
-against the built `dist/` and asserts on computed layout at widths derived from
+- Cloudflare, Fastly, Akamai, Google, GitHub and Backblaze B2
-that CSS — on every breakpoint it declares and one pixel either side of it, plus
+- 8 S3 regional endpoints: Cape Town, London, Bahrain, Tokyo, Singapore, Sydney,
-a 320px floor, a desktop baseline and two landscape sizes. See
+  Oregon and São Paulo
-[test/viewport/README.md](test/viewport/README.md) for what it covers and what
+- 6 Hetzner speed-test servers: Nuremberg, Falkenstein, Helsinki, Ashburn,
-it genuinely cannot.
+  Hillsboro and Singapore
 
-## Rationale
+Two local targets follow under the heading "Local Network", left out of the
+summary and the health box. `Local CPE` is the cable modem at
+`http://192.168.100.1`. `Local Gateway` is the first of `http://192.168.1.1`,
+`http://192.168.0.1`, `http://192.168.8.1` and `http://10.0.0.1` to answer
+within 1.5 seconds of the page loading; with no answer, it has no row. Browsers
+block a page served from a public address from reaching private addresses. So
+the local targets answer only when the page is served from `localhost` or from a
+private address.
 
-When debugging network issues, it's useful to have a persistent at-a-glance view
+The host list is written in `web/src/main.js`; there is no setting for it.
-of latency and reachability to multiple well-known internet endpoints. NetWatch
-provides this as a single page that does all its measuring in the browser, so it
-can be served from anywhere static files are served. The backend in its Docker
-image only stores the measurements the page reports; without it, the page works
-the same and nothing is stored.
 
-## Design
+### What a figure measures
 
-The page's source is in `web/`, the Go backend's in `cmd/` and `internal/`, and
+A check is a request to the target's URL: `GET`, or `HEAD` for the S3 regions.
-nginx's configuration in `configs/`.
+The figure is the time from sending the request until the response headers
+arrive, in milliseconds. Unlike ping, it includes the time the server takes to
+answer. On a new connection it also includes the DNS lookup, the connection
+setup and the TLS handshake. Any answer counts as reachable, an error page
+included. The browser follows redirects, so a target whose home page redirects
+is checked at `/robots.txt` on the same host.
 
-The page is built with Vite and Tailwind CSS v4. Its code is all in
+A check times out after 80% of the interval: 2.4 seconds at the default. A check
-`web/src/main.js`, with a class-based architecture:
+that answers after that limit counts as a timeout too. A check that fails with a
+network error is recorded as unreachable. The first round after the page loads
+is discarded, as its figures include the connection setup. So is the first round
+after an interval change.
 
-- **`CONFIG`**: Configuration object (update interval, timeouts, axis ticks,
+### Health and summary
-  etc.). The interval menu sets `updateInterval`, the one value the page writes
-  into `CONFIG`; the timeouts, the time the history spans and the x-axis ticks
-  are computed from it
-- **`HostState`**: Per-host state management — history buffer, latency tracking,
-  status transitions
-- **`AppState`**: Top-level state container — WAN hosts, local hosts, pause
-  state, aggregate stats
-- **`SparklineRenderer`**: Canvas 2D sparkline drawing with fixed axes,
-  color-coded line segments, error regions, and DPR-aware scaling
-- **UI functions**: `buildUI()` constructs the DOM, `updateHostRow()` /
-  `updateSummary()` / `updateHealthBox()` handle incremental updates
-- **`tick()`**: Main loop — measures all hosts in parallel, pushing each host's
-  sample and redrawing its row as soon as its check ends, then redraws every
-  row, the summary and the health box once the last check ends. The first round,
-  after loading or an interval change, is discarded. The rows are sorted when
-  the last check ends in round 2, the first one kept, and in rounds 11, 21, 31
-  and so on. When paused, pushes blank markers (no probes, no false outage)
-- **`Reporter`**: Posts collected samples to the backend
 
-### Reporting
+Above the rows, a health box sums up the WAN hosts' latest results. A host
+counts as failed when its last check timed out or found it unreachable.
 
-Every `reportInterval` (default 60s) the page POSTs a JSON report to the
+- `OFFLINE` (red): more than 10 hosts failed and at most 4 answered
-same-origin path `/api/v1/reports`: a random per-browser `clientId` kept in
+- `DEGRADED` (orange): otherwise, more than 4 hosts failed
-`localStorage`, `geo` sent as null, and each host's unreported, non-paused
+- `SLOW` (yellow): otherwise, more than 3 hosts took over 1000 ms
-samples (timestamp, latency, error). A per-host high-water mark makes every
+- `HEALTHY` (green): otherwise
-report a delta, so only new samples are sent; the mark advances only on a
-delivered report, and while paused nothing is sent. Changing the interval clears
-each host's history on the page; the samples in it not yet reported still go in
-the next report. Delivery failure is quiet — one debug-log line per outage,
-retried at the next interval, never blocking probing. The report-building step
-is a pure function of host state.
 
-### Backend
+Below it, a summary line covers the WAN hosts only:
 
-`netwatch-server` is a small Go HTTP server that stores the reports the page
+- `Reachable`: how many of the 26 answered their last check
-posts. It keeps them in memory and writes them to `DATA_DIR` as zstd-compressed
+- `Now`: the minimum, median, average and maximum of their latest figures
-files of JSON lines: every minute, whenever 10 MiB are waiting, and when it
+- the history's span, such as `5m`: the minimum and maximum over the whole
-stops. Its routes:
+  history
+- `Checks`: the rounds run since the page loaded or the interval changed
 
-- `POST /api/v1/reports` — takes a report, without credentials; each client
+When no WAN host answers, the page checks 4 of them, picked at random, every
-  address may send a limited number a minute, and the report files are capped in
+half second. As soon as one answers, a new round starts.
-  size, the oldest deleted first
-- `GET /.well-known/healthcheck` — answers 200 with `"status":"ok"`, the
-  server's version and its uptime
-- `GET /metrics` — Prometheus metrics behind basic auth, only when
-  `METRICS_USERNAME` and `METRICS_PASSWORD` are set; each client address may
-  make a limited number of requests to it a minute
 
-It is an `fx`-wired Go application: `cmd/netwatch-server/` starts it, and its
+### Rows, colors and controls
-packages are in `internal/`:
 
-- **`config`**: Loads configuration from environment variables and config files
+A figure and its line in the graph take the color of the legend in the footer,
-  via Viper.
+from green under 50 ms to red from 500 ms up; gray is no answer. The graph's
-- **`handlers`**: HTTP request handlers for the API (health check, report
+axis ends at 1000 ms; a higher latency is drawn at the top, and the figure still
-  ingestion).
+shows its real value.
-- **`reportbuf`**: The in-memory buffer of reports and the files it writes them
-  to.
-- **`server`**: Chi-based HTTP server with middleware wiring and route
-  registration.
-- **`healthcheck`**, **`middleware`**, **`logger`**, **`globals`**: Supporting
-  infrastructure.
 
-In the image, the `test` phase of `Dockerfile` tests it, the `builder` stage
+From time to time, the page sorts the WAN rows by their latest latency: fastest
-builds it, and `bin/entrypoint.sh` runs it as user `netwatch` on
+first, failed last. The pin button at the left of a row keeps the row at the
-`127.0.0.1:8081`, behind nginx (see [Deployment](#deployment)). Outside the
+top; pinned rows are sorted by name. `datavi.be` is pinned at start.
-image, `make run` builds it and runs it on port 8080.
 
-#### Settings
+`Pause` stops the checks: the graphs keep scrolling with a gap for each paused
+round, and the figures turn gray. The `Interval` menu sets the time between
+rounds: 1, 2, 3, 5, 10, 15, 30 or 60 seconds. Changing it clears the history and
+starts a new round at once. The `Debug log` checkbox in the footer opens a panel
+with the page's last 1000 log lines.
 
-| Variable               | Default              | Description                                                                                              |
+## Reporting
-| ---------------------- | -------------------- | -------------------------------------------------------------------------------------------------------- |
-| `BIND_ADDRESS`         | empty                | IP address to listen on; empty listens on every interface                                                |
-| `PORT`                 | `8080`               | HTTP listen port                                                                                         |
-| `DATA_DIR`             | `./data/reports`     | Directory for compressed reports                                                                         |
-| `DATA_DIR_MAX_BYTES`   | `1073741824` (1 GiB) | Most bytes of report files kept in `DATA_DIR`, oldest deleted first; see [Report limits](#report-limits) |
-| `DEBUG`                | `false`              | Enable debug logging                                                                                     |
-| `TRUSTED_PROXIES`      | RFC 1918 ranges      | Comma-separated CIDRs whose `X-Forwarded-For` / `X-Real-IP` headers are trusted for client IP resolution |
-| `REPORTS_PER_MINUTE`   | `60`                 | Reports each client address may send a minute; see [Report limits](#report-limits)                       |
-| `CORS_ALLOWED_ORIGINS` | empty                | Comma-separated origins whose pages may call the API; see [CORS](#cors)                                  |
-| `METRICS_USERNAME`     | empty                | Basic auth user name for `/metrics`; see [Metrics](#metrics)                                             |
-| `METRICS_PASSWORD`     | empty                | Basic auth password for `/metrics`; see [Metrics](#metrics)                                              |
-| `SENTRY_DSN`           | empty                | DSN of the Sentry project to send errors to; see [Sentry](#sentry)                                       |
 
-`TRUSTED_PROXIES` defaults to `10.0.0.0/8,172.16.0.0/12,192.168.0.0/16`. A list
+Every 60 seconds the page posts its undelivered samples as a report to
-replaces the default; set and empty, the server trusts no proxy. Loopback is not
+`/api/v1/reports` on the origin it was served from. The report is JSON with four
-in the default, so a reverse proxy on the same host has to be named, as in
+fields:
-`TRUSTED_PROXIES=127.0.0.1/32`. A request whose direct peer is outside this set
-has its forwarded headers ignored, and the direct peer is logged and
-rate-limited instead. The container image does not use this default; see
-[Deployment](#deployment).
 
-A variable set to a value the server cannot use, such as `PORT=abc`,
+- `clientId`: a random id, made once per browser and kept in `localStorage`, or
-`DEBUG=maybe` or a `BIND_ADDRESS` that is not an IP address, stops it from
+  once per session when the browser has no usable storage
-starting, with an error naming the variable. An empty variable counts as unset,
+- `geo`: always `null`
-except `TRUSTED_PROXIES`.
+- `hosts`: one entry per host with new samples: its `name`, `url` and `status`,
+  and `history`, its samples. A sample has the check's time `t` in Unix
+  milliseconds, its `latency` (`null` for a failed check) and its `error`
+  (`"timeout"`, `"unreachable"` or `null`).
+- `timestamp`: the time of the report
 
-#### Report storage
+Each report carries only the samples no earlier report has delivered. A sample
+waits as long as the history holds it: the last 100 rounds of each host. An
+interval change clears the history. The cleared samples are kept, the last 100
+of each host, until a report delivers them. Nothing is sent while the page is
+paused. A report with no answer after 30 seconds is abandoned, and its samples
+go in the next one. So a server that stored them but answered late gets them
+twice.
 
-Reports are written as `reports-<timestamp>-<number>.jsonl.zst` files in
+A failed delivery is quiet, as on a static file host with no backend: one debug
-`DATA_DIR`. The timestamp is in UTC to the millisecond, so the names sort by
+log line when deliveries start failing, one when they recover. The checks go on.
-time. The number starts at 1 when the server starts and goes up by one for each
-file the server starts to write, so two files written in the same millisecond
-still get different names. A failed write uses up its number and leaves a gap in
-the numbers: its file, if it was created, is removed. The file stays, counted
-toward `DATA_DIR_MAX_BYTES` from the next start, only if removing it fails too.
-Files are created with `O_EXCL` to prevent overwrites.
-
-The reports of a failed write stay in memory, and the next flush, within a
-minute, tries again to write them. From the failed write until a write succeeds,
-`POST /api/v1/reports` refuses every report with 500 and keeps nothing of it, so
-the page keeps its samples and sends them again at its next report. Reports
-already answered 200 are lost only if the server stops before a write succeeds;
-the final flush at stop tries them too, and if it fails, the process exits
-non-zero.
-
-#### Report limits
-
-`POST /api/v1/reports` takes reports from anyone who can reach it, without
-credentials, so it is bounded instead. Each refusal below answers with the same
-`{"status":"error"}` body as any other error.
-
-- **JSON only.** A report must be sent with `Content-Type: application/json`,
-  parameters such as `charset` allowed, as the page sends it; any other is
-  refused with 415, and nothing of it is stored. A page on another origin can
-  send a `text/plain` POST, or a form, without a CORS preflight; refusing them
-  keeps any site from making its visitors' browsers store reports (see
-  [CORS](#cors)).
-- **Rate limit.** Each client address, resolved through `TRUSTED_PROXIES`, may
-  send `REPORTS_PER_MINUTE` reports a minute; past that it gets 429 with
-  `Retry-After: 60`. The minute slides: reports from the minute before still
-  count, fading out over the current one, so an address is sure never to be
-  refused only while it sends at most half of `REPORTS_PER_MINUTE` in any 60
-  seconds. The page sends one report a minute from each open tab, so the default
-  of 60 refuses nothing from up to 30 tabs behind one address, such as a
-  household or an office sharing it, however their reports bunch up. Report
-  responses also carry `X-RateLimit-Limit`, `X-RateLimit-Remaining` and
-  `X-RateLimit-Reset` headers.
-- **Size cap.** The report files in `DATA_DIR` may total at most
-  `DATA_DIR_MAX_BYTES`, counting the files already there at start. Reports
-  waiting in memory count at their uncompressed size until they are written,
-  those of a failed write included; the part of its file written is removed.
-  When a report would take the total past the cap, the oldest report files are
-  deleted to make room, and each deletion is logged with the file's name and
-  size; a file still being written is never deleted. A report is refused with
-  507, and nothing of it is stored, only when the reports waiting to be written
-  fill the cap on their own, and then no file is deleted. At start, report files
-  past the cap, as after lowering it, are deleted the same way. So the cap is
-  how much of the newest reports is kept: the default of 1 GiB is small enough
-  for any host; set it to the space you can give `DATA_DIR`.
-
-#### CORS
-
-The page calls the API from the origin it is served from, so by default the
-server sends no CORS headers, and browsers let no other origin's pages call it.
-To serve the page from elsewhere, list that origin in `CORS_ALLOWED_ORIGINS`
-(for example `https://netwatch.example.com`); pages from a listed origin may
-`GET` and `POST` with a `Content-Type` header. Each entry must be a plain
-origin, `scheme://host` with an optional `:port`, as browsers send it: no path,
-not even a trailing `/`, and no `*`. Any other entry stops the server from
-starting, with an error naming `CORS_ALLOWED_ORIGINS`.
-
-#### Metrics
-
-With both `METRICS_USERNAME` and `METRICS_PASSWORD` set, the server serves
-Prometheus metrics at `GET /metrics` to requests with those as their basic auth
-credentials, and answers any other with 401. For each request that reaches the
-health check or `POST /api/v1/reports`, those the rate limit refuses included,
-the metrics record its duration and response size, labelled with its path,
-method and status; they also count those requests in progress, and include Go's
-runtime and process metrics. No other request is recorded: not those to
-`/metrics` itself, and not those answered before they reach either route, such
-as a CORS preflight, or a request refused with 404 for a path no route has, 405
-for a method its route does not take, or 413 for declaring a body length over
-the 1 MiB limit. A report whose body goes over the limit without declaring its
-length reaches the route, is answered 413 there, and is recorded with that
-status. Clients can make up any number of paths and methods, and each would add
-labels to the metrics for as long as the server runs.
-
-`/metrics` is rate limited, so that its password cannot be guessed quickly: each
-client address, resolved through `TRUSTED_PROXIES`, may make 60 requests to it a
-minute, whatever their credentials. Past that it gets 429 with
-`Retry-After: 60`, and its credentials are not checked. The minute slides as it
-does for reports (see [Report limits](#report-limits)), so a scraper polling
-every 2 seconds or less often is never refused. This allowance is apart from the
-one for reports.
-
-#### Sentry
-
-With `SENTRY_DSN` set, each panic in a handler is reported to that Sentry
-project under the release `netwatch-server-` followed by the server's version.
-On shutdown the server waits up to 2 seconds for Sentry to finish sending.
-
-### Monitoring targets
-
-- **26 WAN hosts**: datavi.be (pinned at start), Anthropic API, OpenAI API, AWS
-  Console, Google Cloud Console, Microsoft Azure, Cloudflare, Fastly CDN,
-  Akamai, Google, GitHub, B2, 8 S3 regional endpoints (Cape Town, London,
-  Bahrain, Tokyo, Singapore, Sydney, Oregon, São Paulo) and 6 Hetzner speed test
-  servers (Nuremberg, Falkenstein, Helsinki, Ashburn, Hillsboro, Singapore)
-- **Local CPE**: Cable modem at 192.168.100.1 (always monitored)
-- **Local Gateway**: Auto-detected on startup by probing common default gateway
-  addresses (192.168.1.1, 192.168.0.1, 192.168.8.1, 10.0.0.1); first responder
-  wins. Note: modern browsers enforce Private Network Access restrictions that
-  block public-origin pages from reaching RFC1918 addresses, so local targets
-  only work when NetWatch is served from localhost or a private address.
-
-Local hosts are tracked separately from WAN stats.
-
-### Latency measurement
-
-GET requests to each target's URL as written, with `mode: 'no-cors'` and
-`cache: 'no-store'`, timed with `performance.now()`. No query string is added:
-the Hetzner speed-test servers close the connection without an answer when the
-URL has one, and `no-store` keeps the browser's cache out of the measurement.
-Each check times out after 80% of the refresh interval (24 seconds at 30
-seconds) and is then recorded as a timeout, so a round's checks have all
-finished before the next round is due. When no WAN host answers, a recovery
-probe checks 4 WAN hosts, picked at random when it starts, every half second,
-giving up the checks it started half a second before. As soon as one answers, a
-new round starts at once, as it does after an interval change. A round started
-early gives up the last round's checks if they are still waiting, and that round
-records nothing more, so rounds never overlap. The browser chooses between IPv4
-and IPv6 for each target, as for any request; the local targets are IPv4
-addresses.
-
-Each recorded check that fails writes one line to the browser console with
-`console.error`, and the same line to the debug log: the target's name and URL,
-the time, whether it timed out, answered over the time limit or hit a network
-error (with the error the browser gives the page, such as
-`TypeError: Failed to fetch`, which does not say why), and how long the request
-took. A target that answers after failed checks writes one `console.info` line
-with its latency and how many checks in a row had failed.
-
-### Color coding
-
-| Latency     | Color  |
-| ----------- | ------ |
-| < 50ms      | Green  |
-| < 100ms     | Lime   |
-| < 200ms     | Yellow |
-| < 500ms     | Orange |
-| >= 500ms    | Red    |
-| Unreachable | Gray   |
-
-### Output structure
-
-```
-dist/
-├── index.html
-└── assets/
-    ├── index-*.css
-    └── index-*.js
-```
-
-## Features
-
-- A round of checks every 3 seconds by default; the interval menu sets 1, 2, 3,
-  5, 10, 15, 30 or 60 seconds and clears the history
-- Sparklines of each target's last 100 rounds: 300 seconds at 3 seconds
-- The first round after loading or an interval change is discarded, as DNS and
-  TLS setup inflate its latencies
-- Health indicator from the WAN hosts' latest results: OFFLINE (red) when more
-  than 10 fail and at most 4 answer, otherwise DEGRADED (orange) when more than
-  4 fail, otherwise SLOW (yellow) when more than 3 take over 1000ms, otherwise
-  HEALTHY (green)
-- Summary stats across WAN hosts only: how many answered, the min, median,
-  average and max of their latest latencies, the min and max over the whole
-  history, and the number of rounds run (`Checks`)
-- Fixed chart axes: Y-axis 0–1000ms, higher latencies drawn at the top; X-axis
-  the time the history spans
-- Color-coded latency figures and sparkline line segments
-- WAN host rows sorted by latest latency, unreachable last; pinned rows stay on
-  top, in name order
-- Play/pause: pause stops probes but history keeps scrolling (blank gaps, no
-  false outage)
-- Debug log panel, behind a checkbox in the footer, with five levels (error,
-  warning, notice, info, debug) and the last 1000 lines
-- Local and UTC clocks
-- Clickable service URLs
-- A footer link to the commit the page was built from
-- Canvas-based sparkline rendering with devicePixelRatio scaling
-- Zero runtime dependencies: all resources bundled into build artifacts
 
 ## Deployment
 
-`make build` writes the page to `dist/`, which any static file host (S3, GCS,
+`make docker` builds the image `netwatch` from `Dockerfile`: the whole service
-Cloudflare Pages, Vercel, Netlify, GitHub Pages) can serve; with no backend
+in one container. nginx serves the page and passes `/api/`,
-there, its reports fail quietly and nothing is stored. Or run the Docker image
+`/.well-known/healthcheck` and `/metrics` to `netwatch-server` in the same
-behind a reverse proxy.
+container. The backend listens on `127.0.0.1:8081` only, as user `netwatch` (uid
+1000). `bin/entrypoint.sh` starts both and watches them.
 
-The Docker image, built from `Dockerfile` by `make docker`, is the whole service
+Before the backend starts, `bin/entrypoint.sh` runs two subcommands of
-in one container: nginx serves the built frontend and passes `/api/`,
+`netwatch-server`. `check-cidr` checks each `TRUSTED_PROXIES` entry (see
-`/.well-known/healthcheck` and `/metrics` to the Go backend, `netwatch-server`.
+[Running under upaas](#running-under-upaas)). `prepare-data-dir` creates the
-`bin/entrypoint.sh` starts it listening only inside the container, on
+report directory and gives `/data`, with the directories, regular files and
-`127.0.0.1:8081`, with `TRUSTED_PROXIES=127.0.0.1/32`, so it takes the client
+symbolic links in it, to user `netwatch`. It changes nothing outside `/data`.
-address nginx passes on and no other. The image:
 
-- Listens on port 8080 by default (override with `PORT` env var)
+The image also:
-- Takes the client address from `X-Forwarded-For` only on requests from the
-  reverse proxies named in `TRUSTED_PROXIES`, by default from any address in
-  `10.0.0.0/8`, `172.16.0.0/12` or `192.168.0.0/16`. `bin/entrypoint.sh` checks
-  each entry with `netwatch-server check-cidr`, which parses it as the backend
-  parses its own `TRUSTED_PROXIES`
-- Sends access logs to stdout
-- Caches static assets with immutable headers
-- Sends the security headers `REPO_POLICIES.md` requires on every response, as
-  `configs/security-headers.conf` sets them, in place of the backend's own
-- Stores reports in `DATA_DIR`, `/data/reports` by default, on the `/data`
-  volume. Before the backend starts, the image creates `DATA_DIR` and gives it
-  and `/data` to user `netwatch` (uid 1000), which the backend runs as, with
-  `netwatch-server prepare-data-dir`, which acts on nothing outside `/data`, so
-  a host directory bind-mounted at `/data` ends up owned by uid 1000
-- Writes buffered reports to disk on `docker stop`, and exits non-zero if nginx
-  or the backend exits on its own, so the platform restarts it
 
-## Running under upaas
+- takes the client address from `X-Forwarded-For` only on a request from a proxy
+  named in `TRUSTED_PROXIES`, and passes that one address to the backend. The
+  backend is started with `TRUSTED_PROXIES=127.0.0.1/32`, so it trusts nginx
+  alone.
+- writes nginx's access log and the backend's JSON log to stdout
+- refuses a request body over 1 MiB with nginx's own 413 error page
+- requests `/.well-known/healthcheck` through nginx every 30 seconds, as its
+  `HEALTHCHECK`. The check fails unless both nginx and the backend answer, and
+  while report files cannot be written.
+- on `docker stop`, writes the waiting reports to disk and exits 0, or non-zero
+  if that write fails. If nginx or the backend exits on its own, the other is
+  stopped and the container exits non-zero.
 
-What the [upaas](https://git.eeqj.de/sneak/upaas) app for netwatch needs:
+### Running under upaas
 
-- **Port:** container port `8080`.
+The [upaas](https://git.eeqj.de/sneak/upaas) app for netwatch needs:
-- **Volume:** container path `/data`; the reports are kept in `/data/reports`.
-- **Environment variables:** none is required. An empty one counts as unset,
-  except `TRUSTED_PROXIES`, and one set to a value netwatch cannot use stops the
-  container at start, with the reason in its log.
-    - `PORT`, default `8080`: the container port, from 1 to 65535. `8081` cannot
-      be used: the backend listens on it inside the container
-    - `REPORTS_PER_MINUTE`, default `60`: reports each client address may send a
-      minute
-    - `DATA_DIR_MAX_BYTES`, default `1073741824` (1 GiB): the most room the
-      report files may take; the oldest are deleted to stay under it
-    - `CORS_ALLOWED_ORIGINS`, default empty: other origins whose pages may call
-      the API
-    - `DEBUG`, default `false`: debug logging
-    - `DATA_DIR`, default `/data/reports`: the directory the reports are kept
-      in: `/data` or a path below it, with no `.` or `..` part and no extra `/`.
-      The container also stops if the path goes through a symbolic link that
-      leads out of `/data` or is written as a full path
-    - `TRUSTED_PROXIES`, default `10.0.0.0/8,172.16.0.0/12,192.168.0.0/16`: the
-      addresses the reverse proxy in front of the container connects from, as IP
-      addresses or CIDRs separated by commas. nginx takes the client address
-      from `X-Forwarded-For` only on a request from one of them, and the rate
-      limit counts that address. A list replaces the default. Set and empty,
-      `X-Forwarded-For` is ignored and every client behind the proxy shares the
-      proxy's one allowance of `REPORTS_PER_MINUTE`. Any client that connects
-      from one of these addresses can write its own `X-Forwarded-For`, and
-      through a port Docker publishes, every client may connect from the Docker
-      network's gateway, such as `172.17.0.1`, which the default includes; where
-      that matters, name only the addresses nothing but the proxy connects from.
-    - `METRICS_USERNAME` and `METRICS_PASSWORD`, default empty: with both set,
-      the backend records Prometheus metrics of its requests and serves them at
-      `/metrics` on the container port, to requests with this user name and
-      password as their basic auth credentials. With neither set, there are no
-      metrics and `/metrics` is not found. One set without the other, or a user
-      name containing `:`, stops the container
-    - `SENTRY_DSN`, default empty: set to a Sentry project's DSN, the backend
-      sends its errors to that Sentry project: each request whose handling
-      crashes, which still gets a 500 response. A value Sentry does not accept
-      stops the container. Empty, the backend sends nothing to Sentry
-- **Health check:** the image's `HEALTHCHECK` requests
-  `/.well-known/healthcheck` through nginx every 30 seconds, so it fails unless
-  both nginx and the backend answer. upaas reads the container's health 60
-  seconds after a deploy and fails the deploy unless it is `healthy`. The
-  container also stops when either process exits.
 
-## Browser Compatibility
+- Port: container port `8080`.
+- Volume: container path `/data`.
+- Environment variables: none is required. Every setting in
+  [Settings](#settings) applies, with four differences in the image:
+    - `BIND_ADDRESS` has no effect. The backend listens on `127.0.0.1` inside
+      the container.
+    - `PORT` is nginx's public port, from 1 to 65535 and not `8081`, where the
+      backend listens inside the container.
+    - `DATA_DIR` must be `/data` or a path below it, written in full, with no
+      `.` or `..` part and no extra `/`. A path through a symbolic link is
+      refused unless the link is relative and stays inside `/data`.
+    - `TRUSTED_PROXIES` names the proxies nginx trusts, as IP addresses or CIDRs
+      separated by commas; the backend's own setting is fixed to `127.0.0.1/32`.
+      `bin/entrypoint.sh` gives a lone address a `/32` or `/128` and checks each
+      entry with `netwatch-server check-cidr`, which refuses what the backend
+      would refuse. Set and empty, nginx trusts no proxy, and every client
+      behind a proxy then shares the proxy's one allowance of
+      `REPORTS_PER_MINUTE`. A client that connects from a trusted address can
+      write its own `X-Forwarded-For`. A port Docker publishes can bring
+      connections from the Docker network's gateway, such as `172.17.0.1`, which
+      the default includes. Where that matters, name only the proxy's own
+      addresses.
+- Health: the image's `HEALTHCHECK`, described above.
 
-Requires a modern browser with ES modules, Fetch API, Canvas API, and CSS custom
+The container stops at start when a variable holds a value it cannot use. The
-properties.
+reason is in its log.
 
-## Limitations
+## Settings
 
-- **CORS**: The checks are cross-origin requests in `no-cors` mode, so the page
+`netwatch-server` reads its settings from environment variables. It also reads a
-  cannot read the answer, only time it: any answer counts as reachable, an error
+`.env` file in the directory it starts in, and a YAML file named
-  page included.
+`netwatch-server.yaml` in `/etc/netwatch-server/` or
-- **Local targets**: The cable modem at 192.168.100.1 and the detected gateway
+`~/.config/netwatch-server/`, when one exists. A variable in the environment
-  answer only on a network that has them, and only when NetWatch is served from
+wins over a file.
-  localhost or a private address (see Monitoring targets).
+
-- **Network conditions**: Measurements reflect browser-to-endpoint latency,
+The server does not start when a variable holds a value it cannot use. Its error
-  which includes your local network, ISP, and internet routing.
+names the variable. An empty variable counts as unset, except `TRUSTED_PROXIES`.
+
+- `BIND_ADDRESS`, default empty: the IP address to listen on. Empty listens on
+  every interface. The image sets it to `127.0.0.1`.
+- `PORT`, default `8080`: the port to listen on, from 1 to 65535. In the image
+  this is nginx's port; the backend listens on `8081`.
+- `DATA_DIR`, default `./data/reports` (`/data/reports` in the image): the
+  directory for the report files. The server creates it if it is missing, and
+  stops at start if it cannot create a file in it.
+- `DATA_DIR_MAX_BYTES`, default `1073741824` (1 GiB): the most bytes of report
+  files kept in `DATA_DIR`, oldest deleted first. A positive whole number. See
+  [Report limits](#report-limits).
+- `REPORTS_PER_MINUTE`, default `60`: how many reports each client address may
+  send a minute. A positive whole number. See [Report limits](#report-limits).
+- `TRUSTED_PROXIES`, default `10.0.0.0/8,172.16.0.0/12,192.168.0.0/16`: the
+  reverse proxies whose `X-Forwarded-For` and `X-Real-IP` headers are trusted,
+  as CIDRs separated by commas. See [Trusted proxies](#trusted-proxies).
+- `CORS_ALLOWED_ORIGINS`, default empty: other origins whose pages may call the
+  API, separated by commas. See [CORS](#cors).
+- `METRICS_USERNAME` and `METRICS_PASSWORD`, default empty: the basic auth
+  credentials for `/metrics`. See [Metrics](#metrics).
+- `SENTRY_DSN`, default empty: the DSN of a Sentry project. Each request whose
+  handling crashes is reported there; the request still gets a 500.
+- `DEBUG`, default empty: any non-empty value turns debug logging on, `false`
+  and `0` included. Every value is accepted.
+
+### Trusted proxies
+
+Each `TRUSTED_PROXIES` entry must be a CIDR, such as `127.0.0.1/32`; a lone
+address is refused. A list replaces the default; set and empty, the server
+trusts no proxy. Loopback is not in the default, so a reverse proxy on the same
+host has to be named.
+
+A request from a peer outside this set has its forwarded headers ignored; the
+peer's own address is logged and rate limited. From a peer inside the set, the
+client address is the right-most `X-Forwarded-For` entry that is not in the set.
+The peer's address is used when every entry is in the set, or when the search
+from the right meets an entry that is not an address, such as `unknown`.
+`X-Real-IP` is read only when `X-Forwarded-For` is missing or empty.
+
+### Routes
+
+- `POST /api/v1/reports` takes a report from the page. It needs no credentials;
+  [Report limits](#report-limits) bound it instead.
+- `GET /.well-known/healthcheck` answers 200 with a JSON object: `appname`,
+  `version`, `now`, `uptime_seconds`, `uptime_human` and `status`, which is
+  `"ok"`. While report files cannot be written, it answers 503 with the same
+  fields and `"status":"error"`.
+- `GET /metrics` serves Prometheus metrics, only when `METRICS_USERNAME` and
+  `METRICS_PASSWORD` are set. See [Metrics](#metrics).
+
+Every request body is capped at 1 MiB, and every request at 60 seconds.
+
+### Report storage
+
+The server keeps the reports it takes in memory and writes them to `DATA_DIR` as
+files of JSON lines, compressed with zstd. It writes a new file every minute,
+whenever 10 MiB of reports are waiting, and when it stops. With none waiting, it
+writes no file. The files are named `reports-<timestamp>-<number>.jsonl.zst`;
+the timestamp is UTC to the millisecond, so the names sort by time.
+
+The reports of a failed write, as on a full disk, stay in memory for the next
+write to try again. Until a write succeeds, `POST /api/v1/reports` refuses every
+report with 500 and the health check answers 503.
+
+### Report limits
+
+`POST /api/v1/reports` takes reports from anyone who can reach it, so it is
+bounded in four ways. Each refusal by the backend answers with the body
+`{"status":"error"}`.
+
+- JSON only. A report must be sent with `Content-Type: application/json`; any
+  other content type is refused with 415, and malformed JSON with 400.
+- Body limit. A report body may be at most 1 MiB. A request that declares a
+  longer body, or a report that runs past that, is refused with 413.
+- Rate limit. Each client address may send `REPORTS_PER_MINUTE` reports a
+  minute; past that it gets 429 with `Retry-After: 60`. The page sends one
+  report a minute from each open tab, so the default of 60 refuses nothing from
+  up to 30 tabs behind one address.
+- Size cap. The report files in `DATA_DIR` may total at most
+  `DATA_DIR_MAX_BYTES`, counting the files already there at start and the
+  reports waiting in memory. When a report would take the total past the cap,
+  the oldest files are deleted to make room. A report is refused with 507 when
+  deleting the oldest files cannot make room.
+
+At the defaults, one address can fill the 1 GiB cap with its own reports in
+about 17 minutes, and every report stored before them is deleted to make room.
+
+### CORS
+
+The page needs no entry here: it calls the API from the origin it is served
+from. By default the server sends no CORS headers, and browsers let no other
+origin's pages call it. `CORS_ALLOWED_ORIGINS` lists other origins, such as
+`https://netwatch.example.com`, whose pages may then call the API with `GET` and
+`POST`. Each entry must be a plain origin as browsers send it: `scheme://host`,
+with an optional `:port`, no path, no trailing `/` and no `*`.
+
+### Metrics
+
+With both `METRICS_USERNAME` and `METRICS_PASSWORD` set, the server serves
+Prometheus metrics at `GET /metrics`. A request with those as its basic auth
+credentials gets them; any other gets 401. With neither set, `/metrics`
+answers 404. The metrics record each request that reaches the health check or
+`POST /api/v1/reports`: its duration and response size, labelled with its path,
+method and status. Each client address may make 60 requests to `/metrics` a
+minute. Past that it gets 429 with `Retry-After: 60`, and its credentials are
+not checked.
+
+### Logging
+
+The backend logs JSON to stdout, or plain text when stdout is a terminal. Each
+request is logged with its method, URL, status, client address and duration.
+Every value taken from a request is cut to 128 bytes before it is logged.
+
+## Design
+
+The repository holds the page, the backend and the image:
+
+- `web/`: the page's source. `index.html` loads `src/main.js` and
+  `src/styles.css`.
+- `cmd/netwatch-server/` and `internal/`: the Go backend. `cmd/` holds only
+  `main`, which calls into `internal/cli`.
+- `configs/`: nginx's configuration for the image, `nginx.conf` and
+  `security-headers.conf`.
+- `bin/entrypoint.sh`: the image's entrypoint.
+- `Dockerfile`: the lint and test phases, the two build stages and the image.
+- `script/`: the development commands; `Makefile` targets call them.
+- `test/unit/`: the page's unit tests. `test/viewport/`: the layout test.
+
+### The page
+
+The page is built with Vite and Tailwind CSS v4. All its code is in
+`web/src/main.js`:
+
+- `CONFIG`: the timing and display constants. The interval menu sets its
+  `updateInterval`; the check timeout and the history span follow from it.
+- `WAN_HOSTS`: the WAN targets, each with its URL and, for the S3 entries, the
+  method `HEAD`.
+- `HostState` and `AppState`: one host's history and status, and the whole
+  page's state.
+- `measureLatency`: one check. `tick`: one round, which checks every host in
+  parallel and draws each row as soon as its check ends.
+- `SparklineRenderer`: the graphs. `buildUI`, `updateHostRow`, `updateSummary`
+  and `updateHealthBox`: the page and its parts.
+- `Reporter`: builds and posts the reports. `buildReport` is a pure function of
+  the hosts' state.
+
+### The backend
+
+`netwatch-server` is a Go HTTP server. The `fx` library starts and stops its
+parts in dependency order. The backend's packages are in `internal/`:
+
+- `cli`: the command line: the server when given no arguments, or the
+  `check-cidr` and `prepare-data-dir` subcommands that `bin/entrypoint.sh` runs.
+- `config`: reads the settings and refuses values the server cannot use.
+- `server`: the router, the routes and the HTTP server.
+- `middleware`: the request log, the security headers, CORS, the body limit, the
+  rate limit, the client address resolution and the metrics.
+- `handlers`: the health check and the report route.
+- `reportbuf`: the in-memory buffer of reports and the files it writes them to.
+  It also holds `PrepareDataDir`, which `prepare-data-dir` runs.
+- `healthcheck`, `logger` and `globals`: the health status, the logger, and the
+  name and version the build stamps in.
+
+## Development
+
+### Entrypoints
+
+This repository adheres to the
+[Scripts to Rule Them All](https://github.com/github/scripts-to-rule-them-all)
+standard. The scripts in `script/` are the entrypoints for the development
+workflow, and the `Makefile` targets are thin shims that call them.
+
+- `script/bootstrap`: install the dependencies. It installs the pinned node,
+  yarn and Go when the ones on `PATH` do not pass its checks, and links them
+  into `~/.local/bin`, which has to be on `PATH`. It installs no Go linter and
+  no Docker.
+- `script/setup`: `script/bootstrap`, then the git pre-commit hook.
+- `script/dev`: run the Vite dev server.
+- `script/build`: build the page into `dist/`, then the static `netwatch-server`
+  binary into the repo root. Its version is `VERSION` from the environment, else
+  `git describe --tags --always --dirty`, else `dev`.
+- `script/run`: `script/build`, then run `netwatch-server` with its defaults.
+- `script/clean`: remove what `script/build` writes.
+- `script/projectname`: print `netwatch`, the image tag `script/docker` uses.
+- `script/yarn`: run yarn under the node `script/bootstrap` would use.
+- `script/test`: build the `test` phase of `Dockerfile`, writing no image: the
+  Go tests with the race detector and coverage, and the page's unit tests.
+- `script/lint`: build the `lint` phase of `Dockerfile`, writing no image:
+  golangci-lint over the Go packages and eslint over the JavaScript.
+- `script/fmt`: run the formatters: prettier over the JavaScript, CSS, HTML and
+  Markdown, then gofmt over `cmd/` and `internal/`. They write the files in
+  place.
+- `script/fmt-check`: check the formatting (read-only), with the same scope as
+  `script/fmt`.
+- `script/check`: `script/test`, `script/lint` and `script/fmt-check`.
+- `script/add-dependency`: add a page dependency, or move one to another
+  version: `make add-dependency PACKAGE=<name>@<version>`.
+- `script/tidy`: run `go mod tidy`.
+- `script/frontend-test`: run the unit tests in `test/unit/` on the host, with
+  Node's built-in test runner, then the production build.
+- `script/frontend-fmt` and `script/frontend-fmt-check`: the prettier half of
+  `script/fmt` and `script/fmt-check`.
+- `script/frontend-viewport-test`: check the built page's layout in a
+  containerised headless Chrome, not as part of `script/check`; see
+  [test/viewport/README.md](test/viewport/README.md).
+- `script/docker`: build the image from `Dockerfile`, tagged `netwatch`.
+- `script/cibuild`: `script/bootstrap`, `script/check`, then the image build.
+  The Gitea workflow `.gitea/workflows/check.yml` runs it on every push.
+- `script/precommit`: run by the git pre-commit hook; it runs `script/check`.
+- `script/install-precommit`: install the git pre-commit hook. `make hooks`
+  calls it.
+
+### Tests
+
+`make test` runs the backend's Go tests and the page's unit tests in Docker. The
+unit tests import `web/src/main.js` in Node and call the functions it exports;
+no browser is involved. `make frontend-viewport-test` checks the built page's
+layout in a headless Chrome; [test/viewport/README.md](test/viewport/README.md)
+says what it checks and what it cannot.
 
 ## TODO
 

REPO_POLICIES.md

@@ -1,6 +1,6 @@
 ---
 title: Repository Policies
-last_modified: 2026-10-04
+last_modified: 2026-10-18
 ---
 
 This document covers repository structure, tooling, and workflow standards. Code
@@ -54,34 +54,39 @@ style conventions are in separate documents:
   `cibuild`. `script/bootstrap` installs all dependencies idempotently and
   assumes nothing is present: base tools come from nix, apt, brew, or apk
   (detected in that order; apt runs noninteractive). For node it uses the
-  installed node if present; otherwise it installs a PINNED node version via
+  installed node only when its major version is the pinned one; otherwise (node
-  nvm, first installing nvm itself if missing — from a hash-verified GitHub
+  missing, or another major, such as the node 24 the CI runner image ships) it
-  release archive (never `curl | sh`), with bash installed as an explicit
+  installs the PINNED node version via nvm and installs yarn and the packages
-  prerequisite since nvm requires bash. yarn is then pinned via
+  under it. nvm itself is installed first if missing, from a hash-verified
-  `corepack prepare yarn@<version> --activate`. Never install "latest" or "lts";
+  GitHub release archive (never `curl | sh`), with bash installed as an explicit
-  always exact versions. `script/cibuild` runs the CI build: it changes to the
+  prerequisite since nvm requires bash. Only the major version is compared
-  repo root, runs `script/bootstrap`, runs `script/check`, and builds the image
+  because the `Dockerfile` stages start from a node image whose exact version is
-  with the version; the Gitea workflow calls it. **`script/cibuild` runs
+  not the pin, and nvm cannot install a prebuilt node on alpine. yarn is pinned
+  via `corepack prepare yarn@<version> --activate`. Never install "latest" or
+  "lts"; always exact versions. `script/cibuild` runs the CI build: it changes
+  to the repo root, runs `script/bootstrap`, runs `script/check`, and builds the
+  image with the version; the Gitea workflow calls it. **`script/cibuild` runs
   `script/bootstrap` first**, because the workflow checks out the repo and runs
   nothing else, while `script/fmt-check` runs the formatter on the host: on a
   pristine checkout with nothing installed the run dies there, after the
-  containerised gates have passed. **The bootstrap alone is not enough**:
+  containerised gates have passed. **The bootstrap alone is not enough**: when
-  `script/bootstrap` installs node and yarn under nvm and leaves neither on the
+  `script/bootstrap` installs node and yarn under nvm it leaves neither on the
-  `PATH` of the shell that called it, so a bare `yarn` still exits 127. The host
+  `PATH` of the shell that called it, so a bare `yarn` either exits 127 or runs
-  entrypoints that need yarn — `script/fmt` and `script/fmt-check` — therefore
+  under another node. The host entrypoints that need yarn — `script/fmt` and
-  source nvm for the pinned node version before invoking it, exactly as
+  `script/fmt-check` — therefore use the same test as `script/bootstrap`: when
-  `script/bootstrap`'s own install step does. A runner carrying nothing but
+  the node on `PATH` has the pinned major version they run the `yarn` on `PATH`,
-  docker and git then gets through `script/check`. Four further scripts are our
+  and otherwise they load nvm through `$HOME/.nvm/nvm.sh` and run yarn under the
-  own extensions to the standard: `script/check` runs `script/test`,
+  pinned node. A runner carrying nothing but docker and git then gets through
-  `script/lint` and `script/fmt-check`; `script/precommit` is what the git
+  `script/check`. Four further scripts are our own extensions to the standard:
-  pre-commit hook runs, and it calls `script/check`; `script/install-precommit`
+  `script/check` runs `script/test`, `script/lint` and `script/fmt-check`;
-  installs the git pre-commit hook (the `make hooks` target shims to it); and
+  `script/precommit` is what the git pre-commit hook runs, and it calls
-  `script/projectname` (literally that filename) simply outputs the project's
+  `script/check`; `script/install-precommit` installs the git pre-commit hook
-  name. Scripts that need the name call `script/projectname` — e.g.
+  (the `make hooks` target shims to it); and `script/projectname` (literally
-  `script/docker` assembles its image tag from it — so those scripts stay
+  that filename) simply outputs the project's name. Scripts that need the name
-  byte-identical across all repos. Repo-type-specific pre-commit extras (e.g.
+  call `script/projectname` — e.g. `script/docker` assembles its image tag from
-  `go mod tidy` verification in Go repos) belong in `script/precommit`, not in
+  it — so those scripts stay byte-identical across all repos. Repo-type-specific
-  the hook itself. Model scripts are at
+  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).
@@ -118,8 +123,10 @@ style conventions are in separate documents:
   and nothing else:
 
     ```sh
-    docker build --no-cache --target lint -t "$(script/projectname)-lint" .
+    docker build --build-arg CACHEBUST="$(date +%s)-$$" \
-    docker build --no-cache --target test -t "$(script/projectname)-test" .
+        --target lint --output type=cacheonly .
+    docker build --build-arg CACHEBUST="$(date +%s)-$$" \
+        --target test --output type=cacheonly .
     ```
 
     **A stage that is not the last one in the file is built only when the final
@@ -129,10 +136,15 @@ style conventions are in separate documents:
     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
+    **The gate builds write no image.** With `--output type=cacheonly` the phase
-    `script/cibuild` and `script/docker`. An untagged build leaves a dangling
+    runs and a failing step fails the build, but the result is not exported.
-    image behind on every invocation, on every developer host and every CI
+    Nothing uses those images, and writing one out is slow: a Go test phase's
-    runner; a tagged one replaces the previous image.
+    image holds the toolchain. A build given neither `--output` nor `-t` writes
+    an untagged image and leaves it dangling, on every developer host and every
+    CI runner. `script/cibuild` and `script/docker` build the image that ships
+    and tag it, so each build replaces the previous image; each assigns the tag
+    on its own line before the build, so `set -e` stops it where
+    `script/projectname` fails.
 
     Inside a phase the tool is invoked directly — `golangci-lint`, `go test`,
     `eslint`, `prettier` — never through `make lint` or `script/test`, which are
@@ -147,22 +159,42 @@ style conventions are in separate documents:
     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
+    org, in both directions. A container has its own `TMPDIR`, so the lock is
-    a digest-pinned binary, so neither is reachable.
+    out of reach. The lint phase does keep golangci-lint's cache between check
+    builds (below), but it lints every checkout at `/src`, and a finding is
+    taken from that cache only for the same files, settings and golangci-lint
+    version.
 
-- **Any build that runs checks is built with `--no-cache`.** Docker invalidates
+- **Any build that runs checks is given a new `CACHEBUST` value.** Docker
-  a `COPY` layer only when the copied content changes, so on an unchanged tree
+  invalidates a `COPY` layer only when the copied content changes, so on an
-  the check `RUN` is served from cache, nothing executes, and the build still
+  unchanged tree the check `RUN` is served from cache, nothing executes, and the
-  exits 0. Every `docker build` in `script/` therefore passes `--no-cache`:
+  build still exits 0. Every `docker build` in `script/` therefore passes
-  `script/lint`, `script/test`, `script/cibuild` and `script/docker` are the
+  `--build-arg CACHEBUST="$(date +%s)-$$"`, the time and the script's process
-  four, and there is no fifth — `script/check` runs the two gate phases and
+  id, so no two builds share a value, not even the three in one
-  `script/fmt-check`, and builds no image of its own. A bare `docker build .` is
+  `script/cibuild`. `script/lint`, `script/test`, `script/cibuild` and
-  not evidence that anything ran: a sub-second build reporting success is a
+  `script/docker` are the four, and there is no fifth — `script/check` runs the
-  cache hit, not a result. Never invalidate by pruning — `docker builder prune`
+  two gate phases and `script/fmt-check`, and builds no image of its own. In
-  and friends destroy a build cache shared with every other build on the host.
+  every stage of the `Dockerfile`, `ARG CACHEBUST` comes right after the steps
-  When a check is added or changed, prove it works by planting a defect it must
+  that install system packages and download Go's modules, or right after `FROM`
-  catch and watching the run fail on it, then revert the defect. A green run
+  in a stage with neither, and every step that lints, tests or compiles comes
-  alone shows neither that the check ran nor that it covers what it should.
+  after it. A step that installs anything else as well, such as a
+  `script/bootstrap` that also installs node packages, comes after it too. A new
+  value reruns every `RUN` step after that line; `COPY` steps are compared by
+  content and need nothing. Docker keeps the layers above the line until the
+  base image, their own lines or the files they copy change. A stage without the
+  `ARG` line is served from cache on an unchanged tree, so a repository
+  re-vendors the scripts and adds the `ARG` lines in one commit. The scripts do
+  not pass `--no-cache`: it reruns every step too, but it also builds the layers
+  above the line again and empties every cache mount, so Go's build cache and
+  golangci-lint's, which the canonical Go `Dockerfile` below mounts so that they
+  survive between check builds, would start empty every time. A bare
+  `docker build .` is not evidence that anything ran: a sub-second build
+  reporting success is a cache hit, not a result. Never invalidate by pruning —
+  `docker builder prune` and friends destroy a build cache shared with every
+  other build on the host. When a check is added or changed, prove it works by
+  planting a defect it must catch and watching the run fail on it, then revert
+  the defect. A green run alone shows neither that the check ran nor that it
+  covers what it should.
 
 - **The gate phases are separate stages, and the build stage depends on both.**
   The lint phase is based on the `golangci/golangci-lint` image (pinned by
@@ -171,40 +203,56 @@ style conventions are in separate documents:
   `Dockerfile`:
 
     ```dockerfile
+    # In every stage ARG CACHEBUST comes right after the steps that install
+    # system packages and download Go's modules, or right after FROM in a
+    # stage with neither, and every step that lints, tests or compiles
+    # comes after it. script/ passes a new value on every build, which
+    # reruns every RUN step after it. Docker keeps the layers above it
+    # until the base image, their own lines or go.mod and go.sum change.
+    # The steps that compile mount Go's build cache, and golangci-lint run
+    # also mounts golangci-lint's; both survive from build to build.
+
     # Lint phase
     # golangci/golangci-lint:v2.x.x, YYYY-MM-DD
     FROM golangci/golangci-lint@sha256:... AS lint
     WORKDIR /src
     COPY go.mod go.sum ./
     RUN go mod download
+    ARG CACHEBUST
     COPY . .
-    RUN golangci-lint run --config .golangci.yml ./...
+    RUN --mount=type=cache,target=/root/.cache/go-build \
+        --mount=type=cache,target=/root/.cache/golangci-lint \
+        golangci-lint run --config .golangci.yml ./...
 
     # Test phase. -race needs cgo and so a C compiler, which the Debian Go
-    # image ships and the alpine one does not.
+    # image ships and the alpine one does not. -count=1 because the mounted
+    # cache also holds test results, which go test would otherwise replay.
     # golang:1.x, YYYY-MM-DD
     FROM golang@sha256:... AS test
     WORKDIR /src
     COPY go.mod go.sum ./
     RUN go mod download
+    ARG CACHEBUST
     COPY . .
-    RUN go test -timeout 90s -race -cover ./... || \
+    RUN --mount=type=cache,target=/root/.cache/go-build \
+        go test -count=1 -timeout 90s -race -cover ./... || \
         { echo "--- Rerunning with -v for details ---"; \
-          go test -timeout 90s -race -v ./...; exit 1; }
+          go test -count=1 -timeout 90s -race -v ./...; exit 1; }
 
     # Build stage. Nothing is wanted from either phase above; the copies
-    # are what make BuildKit build them first, so this stage cannot run
+    # are what make BuildKit build them first, so nothing after them can
-    # unless lint and test passed.
+    # 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
+    ARG CACHEBUST
+    COPY --from=lint /src/go.sum /dev/null
+    COPY --from=test /src/go.sum /dev/null
+    # A tar-stream context keeps the sender's file owners, which git refuses.
+    RUN git config --system --add safe.directory /src
     COPY . .
 
     # The VERSION build arg when one is given, otherwise
@@ -212,7 +260,8 @@ style conventions are in separate documents:
     # .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)}"; \
+    RUN --mount=type=cache,target=/root/.cache/go-build \
+        VERSION="${VERSION:-$(git describe --tags --always)}"; \
         if [ -e .git ]; then \
             case "$VERSION" in ""|dev|unknown) \
                 echo "version is '$VERSION' although .git is present" >&2; \
@@ -225,6 +274,7 @@ style conventions are in separate documents:
 
     # Runtime stage, and the last one
     FROM alpine@sha256:...
+    ARG CACHEBUST
     COPY --from=builder /app /usr/local/bin/app
     ENTRYPOINT ["app"]
     ```
@@ -239,6 +289,26 @@ style conventions are in separate documents:
     - Keep the runtime stage last, and if you add a stage after it, give it the
       same two copies. A plain `docker build .` builds the last stage's chain
       and nothing else.
+    - Check builds keep these, and none of them can hand back a stale result:
+        - The layers above `ARG CACHEBUST`: the system packages and Go's
+          modules. Docker installs the packages again when the base image or the
+          install line changes, and downloads the modules again when `go.mod` or
+          `go.sum` changes. Go checks every module against `go.sum` as it
+          downloads it, and no later build can change a kept layer. The modules
+          are kept in a layer rather than a cache mount because a cache mount is
+          shared by every build on the host, and Go builds from whatever files
+          it holds without checking them again (only `go mod verify` does), so a
+          file changed there would go unnoticed.
+        - Go's build cache (`/root/.cache/go-build`, where Go keeps it for
+          root), mounted on the steps that compile: `golangci-lint run`, both
+          `go test` runs and `go build`. Go reuses a compiled package only when
+          all its inputs are unchanged. The cache also holds test results, which
+          is why both `go test` runs carry `-count=1`.
+        - golangci-lint's cache (`/root/.cache/golangci-lint`, where the pinned
+          image keeps it), mounted on `golangci-lint run`. golangci-lint reuses
+          a package's findings only when the package's files and those of every
+          package it imports, `go.mod`, the linters and their settings in
+          `.golangci.yml`, and the golangci-lint version are all unchanged.
     - If the project uses `//go:embed` directives that reference build artifacts
       (e.g. a web frontend compiled in a separate stage), the lint phase must
       create placeholder files so the embed directives resolve. Example:
@@ -247,12 +317,20 @@ style conventions are in separate documents:
       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:
+      lists in the same `RUN`, so the layer does not keep them. The test phase,
+      on the Debian Go image, installs the libraries its tests need the same
+      way. Either install goes right after `FROM`, above `ARG CACHEBUST`, so
+      Docker keeps it until the base image or the install line changes:
 
         ```dockerfile
+        FROM golangci/golangci-lint@sha256:... AS lint
         RUN apt-get update \
             && apt-get install -y --no-install-recommends libvips-dev \
             && rm -rf /var/lib/apt/lists/*
+        WORKDIR /src
+        COPY go.mod go.sum ./
+        RUN go mod download
+        ARG CACHEBUST
         ```
 
     - `.dockerignore` lets `.git` into the build context. It keeps out every git
@@ -278,20 +356,39 @@ style conventions are in separate documents:
       `.git` and the version still comes out empty, `dev` or `unknown`. A plain
       `docker build .` with no build arguments must succeed; a Dockerfile that
       refuses an empty build argument drops that refusal and keeps the argument.
+      A checkout whose `.git` is a file (a linked worktree, or a repository
+      checked out as a submodule) is the exception: that file points to a git
+      directory outside the build context, so the build cannot read the version
+      and a plain `docker build .` fails; pass the version with
+      `--build-arg VERSION=...`, as `script/docker` and `script/cibuild` already
+      do.
 
 - Every repo should have a Gitea Actions workflow (`.gitea/workflows/`) that
-  runs `script/cibuild` on push, and checks out the repo as its only other step.
+  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
+  with `persist-credentials: false`: `script/cibuild` needs no token, and
-  successful run means every check passed; a bare `docker build .` does not
+  without it the checkout leaves the job's token in `.git/config` for every
+  later step. The checkout step also sets `fetch-depth: 0`, which fetches the
+  tags `git describe` needs: by default it clones shallow with no tags, and a
+  tagged repository's CI build would stamp a bare short commit id. The
+  workflow's `concurrency` block groups runs by workflow and branch
+  (`${{ github.workflow }}-${{ github.ref }}`) with `cancel-in-progress: true`,
+  so a new push cancels the older run on the same branch, queued or running, and
+  no other: runs for replaced commits do not hold up the shared runner.
+  `script/cibuild` bootstraps, runs the gate phases, and then builds the image,
+  so a successful run means every check passed; a bare `docker build .` does not
   carry the same guarantee, because its gate phases may come from the cache. The
-  image build is uncached and so runs the gate phases a second time. That is the
+  image build gets a `CACHEBUST` value of its own and so runs the gate phases a
-  price of the rule above, and it is worth paying: the image that ships is built
+  second time. That is the price of the rule above, and it is worth paying: the
-  from a run of its own gates rather than from a cache entry. A separate
+  image that ships is built from a run of its own gates rather than from a cache
-  workflow limited to `main` by a `branches` list under `on: push` cannot be
+  entry. The `check` job sets `timeout-minutes: 20`, so a hung build frees the
-  checked by review: to try a change to it, add the feature branch to that list
+  shared runner after 20 minutes. That allows for the three Docker builds
-  and push, then remove the branch from the list again before merging. Keep any
+  described above (the test phase, the lint phase, then the image), each held to
-  job in it that publishes behind `if: github.ref_name == 'main'`, so the run
+  the 5-minute Docker build limit below, plus the bootstrap. A separate workflow
-  from the feature branch publishes nothing.
+  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
@@ -351,12 +448,11 @@ style conventions are in separate documents:
     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
+    without running them. The directory lasts from one run to the next on a
-    target runs and the directory lasts from one run to the next. The `test`
+    developer's machine, where this target runs, and in the `test` phase of the
-    phase of the `Dockerfile` needs no `-count=1`: its base image holds no
+    `Dockerfile`, which mounts it to keep compiled packages between check
-    result for this repo's tests and nothing before its `go test` step runs a
+    builds, so the phase's `go test` runs carry `-count=1` too. A new
-    test, so there is nothing to replay. `--no-cache` (above) is what makes that
+    `CACHEBUST` value (above) is what makes that step run on an unchanged tree.
-    step run on an unchanged tree.
 
     Python example:
 
@@ -384,9 +480,12 @@ style conventions are in separate documents:
 
 - `.gitignore` should be comprehensive from the start: OS files (`.DS_Store`),
   editor files (`.swp`, `*~`), in-repo agent scratch directories (`.claude/`),
-  language build artifacts, and `node_modules/`. Fetch the standard `.gitignore`
+  `node_modules/`, and the repo's own build outputs. Fetch the standard
-  from `https://git.eeqj.de/sneak/prompts/raw/branch/main/.gitignore` when
+  `.gitignore` from
-  setting up a new repo. These patterns are written to `.gitignore`'s own
+  `https://git.eeqj.de/sneak/prompts/raw/branch/main/.gitignore` when setting up
+  a new repo. A repo's `.gitignore` is the standard file followed by the repo's
+  own entries, such as its binaries; a re-vendor replaces the standard part and
+  keeps those entries. These patterns are written to `.gitignore`'s own
   semantics, in which an unanchored pattern already matches at every depth; they
   are not a `.dockerignore` and must not be transplanted into one unmodified.
 
@@ -434,13 +533,15 @@ style conventions are in separate documents:
   byte-identically across repos:
 
     ```sh
-    # Own line: a failing command substitution inside an argument does not
+    # The version and the tag each get their own line: a failing command
-    # trip `set -e`, so the inline form degrades to an empty constant.
+    # 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 \
+    tag="$(script/projectname)"
+    docker build --build-arg CACHEBUST="$(date +%s)-$$" \
         --build-arg VERSION="$version" \
-        -t "$(script/projectname)" .
+        -t "$tag" .
     ```
 
     `--always` makes an untagged repo yield an abbreviated commit hash rather
@@ -452,8 +553,8 @@ style conventions are in separate documents:
     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
+    checkout action clones shallow and fetches no tags, so the canonical
-    tag-derived version must set `fetch-depth: 0` on its checkout step.
+    `.gitea/workflows/check.yml` sets `fetch-depth: 0` on its checkout step.
 
 - **Verify `.dockerignore` by enumerating the image, not by reading the
   patterns.** Plant files at the root _and_ at least two directories deep, build
@@ -520,6 +621,11 @@ style conventions are in separate documents:
       function defined and never invoked has the same exit status and the same
       empty output as one that worked.
 
+    Node is the exception to the whole-token comparison: the canonical
+    `script/bootstrap` compares only its major version, because the node 22
+    alpine image the `Dockerfile` stages start from is not the exact pin, and
+    nvm cannot install a prebuilt node on alpine.
+
     Keep it POSIX sh: no arrays, no `[[`, no `grep -P`.
 
     A Go tool a repo needs on the host is installed with `go install` pinned to
@@ -636,7 +742,11 @@ style conventions are in separate documents:
       Never edit existing migrations after release.
 
 - All repos should have an `.editorconfig` enforcing the project's indentation
-  settings.
+  settings: the standard file from
+  `https://git.eeqj.de/sneak/prompts/raw/branch/main/.editorconfig`, which sets
+  tabs for `Makefile` and Go files, followed by the repo's own sections, such as
+  one for another language it uses. A re-vendor replaces the standard part and
+  keeps those sections.
 
 - Avoid putting files in the repo root unless necessary. Root should contain
   only project-level config files (`README.md`, `AGENTS.md`, `Makefile`,

bin/entrypoint.sh

@@ -13,24 +13,27 @@ set -u
 # address and start anyway, and reports a bad port without naming
 # PORT, so a value that is not a usable port stops the container here,
 # before either process starts.
-export PORT="${PORT:-8080}"
+check_port() {
-case "$PORT" in
+    export PORT="${PORT:-8080}"
-    *[!0-9]*)
+    case "$PORT" in
-        echo "entrypoint: PORT must be a port number, not '$PORT'" >&2
+        *[!0-9]*)
+            echo "entrypoint: PORT must be a port number, not '$PORT'" >&2
+            exit 1
+            ;;
+    esac
+    # The length is checked first because, for a number too big for it,
+    # the shell's test prints an error and is false, so the range checks
+    # alone would let it through.
+    if [ "${#PORT}" -gt 5 ] || [ "$PORT" -lt 1 ] || [ "$PORT" -gt 65535 ]; then
+        echo "entrypoint: PORT must be from 1 to 65535, not '$PORT'" >&2
         exit 1
-        ;;
+    fi
-esac
+    if [ "$PORT" -eq 8081 ]; then
-# The length is checked first because, for a number too big for it,
+        echo "entrypoint: PORT cannot be 8081, netwatch-server" \
-# the shell's test prints an error and is false, so the range checks
+            "listens there" >&2
-# alone would let it through.
+        exit 1
-if [ "${#PORT}" -gt 5 ] || [ "$PORT" -lt 1 ] || [ "$PORT" -gt 65535 ]; then
+    fi
-    echo "entrypoint: PORT must be from 1 to 65535, not '$PORT'" >&2
+}
-    exit 1
-fi
-if [ "$PORT" -eq 8081 ]; then
-    echo "entrypoint: PORT cannot be 8081, netwatch-server listens there" >&2
-    exit 1
-fi
 
 # TRUSTED_PROXIES names the reverse proxies in front of the container,
 # as IP addresses or CIDRs separated by commas. nginx takes the client
@@ -47,21 +50,23 @@ fi
 # TRUSTED_PROXIES. Its error, naming the CIDR, is dropped for the one
 # below, naming the entry as written. set -f keeps a * in an entry from
 # becoming a list of file names.
-TRUSTED_PROXIES="${TRUSTED_PROXIES-10.0.0.0/8,172.16.0.0/12,192.168.0.0/16}"
+write_trusted_proxies() {
-set -f
+    TRUSTED_PROXIES="${TRUSTED_PROXIES-10.0.0.0/8,172.16.0.0/12,192.168.0.0/16}"
-for proxy in $(printf '%s' "$TRUSTED_PROXIES" | tr ',' ' '); do
+    set -f
-    case "$proxy" in
+    for proxy in $(printf '%s' "$TRUSTED_PROXIES" | tr ',' ' '); do
-        */*) cidr="$proxy" ;;
+        case "$proxy" in
-        *:*) cidr="$proxy/128" ;;
+            */*) cidr="$proxy" ;;
-        *) cidr="$proxy/32" ;;
+            *:*) cidr="$proxy/128" ;;
-    esac
+            *) cidr="$proxy/32" ;;
-    if ! netwatch-server check-cidr "$cidr" 2> /dev/null; then
+        esac
-        echo "entrypoint: TRUSTED_PROXIES must be IP addresses or CIDRs" \
+        if ! netwatch-server check-cidr "$cidr" 2> /dev/null; then
-            "separated by commas; '$proxy' is neither" >&2
+            echo "entrypoint: TRUSTED_PROXIES must be IP addresses or CIDRs" \
-        exit 1
+                "separated by commas; '$proxy' is neither" >&2
-    fi
+            exit 1
-    echo "set_real_ip_from $cidr;"
+        fi
-done > /etc/nginx/trusted-proxies.conf
+        echo "set_real_ip_from $cidr;"
+    done > /etc/nginx/trusted-proxies.conf
+}
 
 # netwatch-server keeps its report files in DATA_DIR, on the /data
 # volume, which may be a host directory owned by root or by another
@@ -73,12 +78,10 @@ done > /etc/nginx/trusted-proxies.conf
 # naming DATA_DIR, unless DATA_DIR is /data or a path below it, and it
 # acts on nothing outside /data, whatever symbolic or hard links it
 # meets there.
-export DATA_DIR="${DATA_DIR:-/data/reports}"
+prepare_data_dir() {
-netwatch-server prepare-data-dir "$DATA_DIR" || exit 1
+    export DATA_DIR="${DATA_DIR:-/data/reports}"
-
+    netwatch-server prepare-data-dir "$DATA_DIR" || exit 1
-# A stop signal is only noted here; the loop below acts on it.
+}
-stop_requested=""
-trap 'stop_requested=yes' TERM INT
 
 # netwatch-server runs as the netwatch user and listens on loopback
 # only, on a port other than the public one; configs/nginx.conf proxies
@@ -87,9 +90,11 @@ trap 'stop_requested=yes' TERM INT
 # the container has. The netwatch user has no login shell, hence -s
 # /bin/sh. busybox su replaces itself with the command instead of
 # staying on as its parent, so $! is the server's own PID.
-BIND_ADDRESS=127.0.0.1 PORT=8081 TRUSTED_PROXIES=127.0.0.1/32 \
+start_backend() {
-    su -s /bin/sh netwatch -c 'exec netwatch-server' &
+    BIND_ADDRESS=127.0.0.1 PORT=8081 TRUSTED_PROXIES=127.0.0.1/32 \
-backend=$!
+        su -s /bin/sh netwatch -c 'exec netwatch-server' &
+    backend=$!
+}
 
 # nginx starts through the nginx image's own entrypoint, which applies
 # the image's start-up configuration and then replaces itself with
@@ -98,9 +103,11 @@ backend=$!
 # that rendering to PORT: a variable nginx itself uses, such as $uri,
 # would otherwise be replaced by an environment variable of the same
 # name.
-NGINX_ENVSUBST_FILTER='^PORT$' \
+start_nginx() {
-    /docker-entrypoint.sh nginx -g 'daemon off;' &
+    NGINX_ENVSUBST_FILTER='^PORT$' \
-nginx=$!
+        /docker-entrypoint.sh nginx -g 'daemon off;' &
+    nginx=$!
+}
 
 running() {
     kill -0 "$1" 2>/dev/null
@@ -109,26 +116,47 @@ running() {
 # POSIX sh cannot wait for whichever of two children exits first, so
 # look once a second. The shell collects a child that has exited while
 # it runs sleep, and running() is false for that child from then on.
-while [ -z "$stop_requested" ] && running "$backend" && running "$nginx"; do
+wait_for_stop_or_exit() {
-    sleep 1
+    while [ -z "$stop_requested" ] && running "$backend" && running "$nginx"; do
-done
+        sleep 1
+    done
+}
 
 # Stop both, then wait until neither is left.
-kill -TERM "$backend" "$nginx" 2>/dev/null
+stop_both() {
-while running "$backend" || running "$nginx"; do
+    kill -TERM "$backend" "$nginx" 2>/dev/null
-    sleep 1
+    while running "$backend" || running "$nginx"; do
-done
+        sleep 1
-
+    done
-wait "$backend"
+}
-backend_status=$?
-wait "$nginx"
-nginx_status=$?
-echo "entrypoint: netwatch-server exited $backend_status," \
-    "nginx exited $nginx_status"
 
 # Success is a requested stop that both processes exited cleanly from.
-if [ -n "$stop_requested" ] && [ "$backend_status" -eq 0 ] &&
+exit_with_status() {
-    [ "$nginx_status" -eq 0 ]; then
+    wait "$backend"
-    exit 0
+    backend_status=$?
-fi
+    wait "$nginx"
-exit 1
+    nginx_status=$?
+    echo "entrypoint: netwatch-server exited $backend_status," \
+        "nginx exited $nginx_status"
+    if [ -n "$stop_requested" ] && [ "$backend_status" -eq 0 ] &&
+        [ "$nginx_status" -eq 0 ]; then
+        exit 0
+    fi
+    exit 1
+}
+
+main() {
+    check_port
+    write_trusted_proxies
+    prepare_data_dir
+    # A stop signal is only noted here; wait_for_stop_or_exit acts on it.
+    stop_requested=""
+    trap 'stop_requested=yes' TERM INT
+    start_backend
+    start_nginx
+    wait_for_stop_or_exit
+    stop_both
+    exit_with_status
+}
+
+main "$@"

cmd/netwatch-server/main.go

@@ -1,22 +1,11 @@
-// Package main is the entry point for netwatch-server.
+// Package main is the entry point for netwatch-server; package cli
+// does the work.
 package main
 
 import (
-	"fmt"
 	"os"
-	"os/user"
 
-	"sneak.berlin/go/netwatch/internal/config"
+	"sneak.berlin/go/netwatch/internal/cli"
-	"sneak.berlin/go/netwatch/internal/globals"
-	"sneak.berlin/go/netwatch/internal/handlers"
-	"sneak.berlin/go/netwatch/internal/healthcheck"
-	"sneak.berlin/go/netwatch/internal/logger"
-	"sneak.berlin/go/netwatch/internal/middleware"
-	"sneak.berlin/go/netwatch/internal/reportbuf"
-	"sneak.berlin/go/netwatch/internal/server"
-
-	"go.uber.org/fx"
-	"go.uber.org/fx/fxevent"
 )
 
 //nolint:gochecknoglobals // set via ldflags at build time
@@ -26,64 +15,5 @@ var (
 )
 
 func main() {
-	// "netwatch-server check-cidr CIDR" exits 1, with the error, if
+	os.Exit(cli.Main(os.Stderr, Appname, Version, os.Args[1:], cli.RunServer))
-	// this server would refuse CIDR in its TRUSTED_PROXIES.
-	// bin/entrypoint.sh runs it on each entry it gives nginx.
-	if len(os.Args) == 3 && os.Args[1] == "check-cidr" {
-		_, err := middleware.ParseTrustedProxies(os.Args[2:])
-		if err != nil {
-			fmt.Fprintln(os.Stderr, err)
-			os.Exit(1)
-		}
-
-		return
-	}
-
-	// "netwatch-server prepare-data-dir DATA_DIR" gets DATA_DIR ready
-	// for the netwatch user, or exits 1 with the error; see
-	// reportbuf.PrepareDataDir. bin/entrypoint.sh runs it as root
-	// before it starts this server as that user.
-	if len(os.Args) == 3 && os.Args[1] == "prepare-data-dir" {
-		netwatch, err := user.Lookup("netwatch")
-		if err != nil {
-			fmt.Fprintln(os.Stderr, err)
-			os.Exit(1)
-		}
-
-		err = reportbuf.PrepareDataDir("/data", os.Args[2], netwatch)
-		if err != nil {
-			fmt.Fprintf(os.Stderr, "DATA_DIR '%s': %v\n", os.Args[2], err)
-			os.Exit(1)
-		}
-
-		return
-	}
-
-	globals.Appname = Appname
-	globals.Version = Version
-
-	fx.New(
-		// fx logs each step of starting and stopping through the
-		// server's own logger, so off a terminal those lines are
-		// JSON like every other line.
-		fx.WithLogger(func(log *logger.Logger) fxevent.Logger {
-			return &fxevent.SlogLogger{Logger: log.Get()}
-		}),
-		fx.Provide(
-			config.New,
-			globals.New,
-			handlers.New,
-			healthcheck.New,
-			logger.New,
-			middleware.New,
-			reportbuf.New,
-			server.New,
-		),
-		fx.Invoke(
-			// First, so the name and version are logged even when
-			// a setting stops the start.
-			func(log *logger.Logger) { log.Identify() },
-			func(*server.Server) {},
-		),
-	).Run()
 }

configs/nginx.conf

@@ -30,8 +30,13 @@ server {
     access_log /dev/stdout combined;
     error_log  /dev/stderr warn;
 
+    # With no-cache, a browser never uses a copy of the page that names
+    # assets a newer build no longer has without first checking with the
+    # server that it is current.
     location / {
         try_files $uri $uri/ /index.html;
+        add_header Cache-Control no-cache;
+        include /etc/nginx/security-headers.conf;
     }
 
     # Cache static assets aggressively
@@ -46,7 +51,9 @@ server {
     # 127.0.0.1:8081. These headers go with every request passed to it.
     # X-Forwarded-For carries only the client address, as resolved by
     # the real IP settings above, and not the chain the request came
-    # with: the backend takes the first entry, which a client can write.
+    # with. bin/entrypoint.sh starts the backend with
+    # TRUSTED_PROXIES=127.0.0.1/32, the address nginx connects from, so
+    # the backend takes that one entry as the client address.
     proxy_set_header Host $host;
     proxy_set_header X-Real-IP $remote_addr;
     proxy_set_header X-Forwarded-For $remote_addr;

configs/security-headers.conf

@@ -8,10 +8,9 @@ add_header Strict-Transport-Security "max-age=31536000; includeSubDomains" alway
 # Scripts and styles load only from the page's own origin. Inline ones
 # are blocked, style attributes in markup included, so style elements
 # through classes or element.style. data: images are for the favicon
-# in web/index.html. connect-src is * because the browser checks each
+# in web/index.html. connect-src is * because the page checks every
-# probe in web/src/main.js against it, and also every redirect the probe
+# target in web/src/main.js. A list of hosts here would have to change
-# follows, and several of those hosts redirect to others; a list of
+# with every change to those targets. It also covers the reports the
-# hosts here would block those probes. It also covers the reports the
 # page sends to its own origin.
 add_header Content-Security-Policy "default-src 'self'; connect-src *; img-src 'self' data:; object-src 'none'; base-uri 'none'; form-action 'none'; frame-ancestors 'none'" always;
 

internal/cli/cli.go

@@ -0,0 +1,143 @@
+// Package cli runs netwatch-server from its command line: the
+// check-cidr and prepare-data-dir subcommands, which bin/entrypoint.sh
+// runs before it starts the server, and, given no arguments, the
+// server, with the fx wiring.
+package cli
+
+import (
+	"fmt"
+	"io"
+	"os/user"
+
+	"sneak.berlin/go/netwatch/internal/config"
+	"sneak.berlin/go/netwatch/internal/globals"
+	"sneak.berlin/go/netwatch/internal/handlers"
+	"sneak.berlin/go/netwatch/internal/healthcheck"
+	"sneak.berlin/go/netwatch/internal/logger"
+	"sneak.berlin/go/netwatch/internal/middleware"
+	"sneak.berlin/go/netwatch/internal/reportbuf"
+	"sneak.berlin/go/netwatch/internal/server"
+
+	"go.uber.org/fx"
+	"go.uber.org/fx/fxevent"
+)
+
+// usageExitCode is what Main returns for a command line it does not
+// understand, the usual exit code for a usage error.
+const usageExitCode = 2
+
+// Main runs netwatch-server with args, its command line after the
+// program's name, and returns the exit code. With no args, Main runs
+// the server by calling serve with appname and version, the name and
+// version the server logs and reports. With "check-cidr CIDR" or
+// "prepare-data-dir DATA_DIR" it runs that subcommand, which writes its
+// errors to stderr. Any other args get a one-line usage message on
+// stderr and exit code 2. main passes os.Stderr and RunServer; the
+// tests pass a buffer and a function of their own, so that no test
+// starts the server. When serve is RunServer and the server fails to
+// start, to listen or to stop, fx's Run exits the process with status 1
+// and Main never returns.
+func Main(
+	stderr io.Writer,
+	appname, version string,
+	args []string,
+	serve func(appname, version string),
+) int {
+	// "netwatch-server check-cidr CIDR" exits 1, with the error, if
+	// this server would refuse CIDR in its TRUSTED_PROXIES.
+	// bin/entrypoint.sh runs it on each entry it gives nginx.
+	if len(args) == 2 && args[0] == "check-cidr" {
+		return checkCIDR(stderr, args[1])
+	}
+
+	// "netwatch-server prepare-data-dir DATA_DIR" gets DATA_DIR ready
+	// for the netwatch user, or exits 1 with the error; see
+	// reportbuf.PrepareDataDir. bin/entrypoint.sh runs it as root
+	// before it starts this server as that user.
+	if len(args) == 2 && args[0] == "prepare-data-dir" {
+		return prepareDataDir(stderr, "/data", args[1], "netwatch")
+	}
+
+	// Any other command line is refused, so that a mistyped one does
+	// not start the server.
+	if len(args) > 0 {
+		_, _ = fmt.Fprintln(stderr,
+			"usage: netwatch-server [check-cidr CIDR | prepare-data-dir DATA_DIR]")
+
+		return usageExitCode
+	}
+
+	serve(appname, version)
+
+	return 0
+}
+
+// checkCIDR returns 0 if the server would take cidr in its
+// TRUSTED_PROXIES, and otherwise writes the error to stderr and
+// returns 1.
+func checkCIDR(stderr io.Writer, cidr string) int {
+	_, err := middleware.ParseTrustedProxies([]string{cidr})
+	if err != nil {
+		_, _ = fmt.Fprintln(stderr, err)
+
+		return 1
+	}
+
+	return 0
+}
+
+// prepareDataDir gets dataDir ready for the user named username with
+// reportbuf.PrepareDataDir, volume being where the data volume is
+// mounted, and returns 0. Otherwise it writes the error to stderr and
+// returns 1.
+func prepareDataDir(stderr io.Writer, volume, dataDir, username string) int {
+	owner, err := user.Lookup(username)
+	if err != nil {
+		_, _ = fmt.Fprintln(stderr, err)
+
+		return 1
+	}
+
+	err = reportbuf.PrepareDataDir(volume, dataDir, owner)
+	if err != nil {
+		_, _ = fmt.Fprintf(stderr, "DATA_DIR '%s': %v\n", dataDir, err)
+
+		return 1
+	}
+
+	return 0
+}
+
+// RunServer runs the server until it is told to stop, with appname and
+// version as the name and version it logs and reports. If the server
+// fails to start, to listen or to stop, fx's Run exits the process
+// with status 1 itself.
+func RunServer(appname, version string) {
+	globals.Appname = appname
+	globals.Version = version
+
+	fx.New(
+		// fx logs each step of starting and stopping through the
+		// server's own logger, so off a terminal those lines are
+		// JSON like every other line.
+		fx.WithLogger(func(log *logger.Logger) fxevent.Logger {
+			return &fxevent.SlogLogger{Logger: log.Get()}
+		}),
+		fx.Provide(
+			config.New,
+			globals.New,
+			handlers.New,
+			healthcheck.New,
+			logger.New,
+			middleware.New,
+			reportbuf.New,
+			server.New,
+		),
+		fx.Invoke(
+			// First, so the name and version are logged even when
+			// a setting stops the start.
+			func(log *logger.Logger) { log.Identify() },
+			func(*server.Server) {},
+		),
+	).Run()
+}

internal/cli/cli_test.go

@@ -0,0 +1,219 @@
+package cli_test
+
+import (
+	"bytes"
+	"os"
+	"os/user"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"testing"
+
+	"sneak.berlin/go/netwatch/internal/cli"
+	"sneak.berlin/go/netwatch/internal/reportbuf"
+)
+
+// TestMainRunsCheckCIDR: Main given check-cidr returns 0, writing
+// nothing, for a CIDR the server takes in its TRUSTED_PROXIES, and 1,
+// with an error naming TRUSTED_PROXIES and the CIDR, for one it
+// refuses, and does not start the server.
+func TestMainRunsCheckCIDR(t *testing.T) {
+	t.Parallel()
+
+	var stderr bytes.Buffer
+
+	code := cli.Main(&stderr, "netwatch", "test",
+		[]string{"check-cidr", "10.0.0.0/8"}, failIfServerStarts(t))
+	if code != 0 || stderr.Len() != 0 {
+		t.Errorf("check-cidr 10.0.0.0/8: exit %d, stderr %q; want exit 0 and no output",
+			code, stderr.String())
+	}
+
+	stderr.Reset()
+
+	code = cli.Main(&stderr, "netwatch", "test",
+		[]string{"check-cidr", "1.2.3/32"}, failIfServerStarts(t))
+
+	want := `TRUSTED_PROXIES "1.2.3/32": `
+	if code != 1 || !strings.HasPrefix(stderr.String(), want) {
+		t.Errorf("check-cidr 1.2.3/32: exit %d, stderr %q; want exit 1 and %q...",
+			code, stderr.String(), want)
+	}
+}
+
+// TestMainRunsPrepareDataDir: Main given prepare-data-dir runs the
+// prepare-data-dir subcommand for the netwatch user, and does not start
+// the server. The test image has no netwatch user, so that subcommand
+// exits 1 with the error from looking the user up;
+// TestPrepareDataDirRefusesPathOutsideVolume checks the path.
+func TestMainRunsPrepareDataDir(t *testing.T) {
+	t.Parallel()
+
+	var stderr bytes.Buffer
+
+	code := cli.Main(&stderr, "netwatch", "test",
+		[]string{"prepare-data-dir", "/etc"}, failIfServerStarts(t))
+
+	want := user.UnknownUserError("netwatch").Error() + "\n"
+	if code != 1 || stderr.String() != want {
+		t.Errorf("prepare-data-dir /etc: exit %d, stderr %q; want exit 1 and %q",
+			code, stderr.String(), want)
+	}
+}
+
+// TestMainRunsServerWithNoArguments: Main given no arguments starts the
+// server with the name and version it was given, writes nothing and
+// returns 0.
+func TestMainRunsServerWithNoArguments(t *testing.T) {
+	t.Parallel()
+
+	var (
+		stderr                 bytes.Buffer
+		gotAppname, gotVersion string
+	)
+
+	code := cli.Main(&stderr, "netwatch", "test", []string{},
+		func(appname, version string) {
+			gotAppname, gotVersion = appname, version
+		})
+	if code != 0 || stderr.Len() != 0 {
+		t.Errorf("exit %d, stderr %q; want exit 0 and no output",
+			code, stderr.String())
+	}
+
+	if gotAppname != "netwatch" || gotVersion != "test" {
+		t.Errorf("server started with %q %q; want %q %q",
+			gotAppname, gotVersion, "netwatch", "test")
+	}
+}
+
+// TestMainRefusesOtherCommandLines: Main given a subcommand with too
+// few or too many values, or a misspelt subcommand, writes the usage
+// message to stderr, returns 2 and does not start the server.
+func TestMainRefusesOtherCommandLines(t *testing.T) {
+	t.Parallel()
+
+	want := "usage: netwatch-server " +
+		"[check-cidr CIDR | prepare-data-dir DATA_DIR]\n"
+
+	for _, commandLine := range []string{
+		"netwatch-server check-cidr",
+		"netwatch-server check-cidr 192.0.2.0/24 198.51.100.0/24",
+		"netwatch-server prepare-data-dir",
+		"netwatch-server prepare-data-dir /data/reports /data/other",
+		"netwatch-server check-cdir 192.0.2.0/24",
+	} {
+		var stderr bytes.Buffer
+
+		// Main takes the command line after the program's name.
+		args := strings.Fields(commandLine)[1:]
+
+		code := cli.Main(&stderr, "netwatch", "test", args, failIfServerStarts(t))
+		if code != 2 || stderr.String() != want {
+			t.Errorf("%q: exit %d, stderr %q; want exit 2 and %q",
+				commandLine, code, stderr.String(), want)
+		}
+	}
+}
+
+// TestCheckCIDRTakesCIDR: check-cidr exits 0, writing nothing, for a
+// CIDR the server takes in its TRUSTED_PROXIES.
+func TestCheckCIDRTakesCIDR(t *testing.T) {
+	t.Parallel()
+
+	for _, cidr := range []string{"10.0.0.0/8", "192.0.2.1/32", "2001:db8::/32"} {
+		var stderr bytes.Buffer
+
+		code := cli.CheckCIDR(&stderr, cidr)
+		if code != 0 || stderr.Len() != 0 {
+			t.Errorf("%q: exit %d, stderr %q; want exit 0 and no output",
+				cidr, code, stderr.String())
+		}
+	}
+}
+
+// TestCheckCIDRRefusesOtherInput: check-cidr exits 1, with an error
+// naming TRUSTED_PROXIES and the input, for anything the server would
+// refuse in its TRUSTED_PROXIES, such as 1.2.3/32, which nginx would
+// look up as a hostname.
+func TestCheckCIDRRefusesOtherInput(t *testing.T) {
+	t.Parallel()
+
+	for _, cidr := range []string{
+		"1.2.3/32", "example.com/32", "10.0.0.0/33", "10.0.0.1", "",
+	} {
+		var stderr bytes.Buffer
+
+		code := cli.CheckCIDR(&stderr, cidr)
+
+		want := "TRUSTED_PROXIES " + strconv.Quote(cidr) + ": "
+		if code != 1 || !strings.HasPrefix(stderr.String(), want) {
+			t.Errorf("%q: exit %d, stderr %q; want exit 1 and %q...",
+				cidr, code, stderr.String(), want)
+		}
+	}
+}
+
+// TestPrepareDataDirTakesPathInVolume: prepare-data-dir exits 0,
+// writing nothing, for a DATA_DIR below the volume, and creates it.
+func TestPrepareDataDirTakesPathInVolume(t *testing.T) {
+	t.Parallel()
+
+	volume := t.TempDir()
+	dataDir := filepath.Join(volume, "reports")
+
+	var stderr bytes.Buffer
+
+	code := cli.PrepareDataDir(&stderr, volume, dataDir, currentUsername(t))
+	if code != 0 || stderr.Len() != 0 {
+		t.Errorf("exit %d, stderr %q; want exit 0 and no output",
+			code, stderr.String())
+	}
+
+	info, err := os.Stat(dataDir)
+	if err != nil || !info.IsDir() {
+		t.Errorf("DATA_DIR not created: %v", err)
+	}
+}
+
+// TestPrepareDataDirRefusesPathOutsideVolume: prepare-data-dir exits
+// 1, with an error naming DATA_DIR and its value, for a DATA_DIR
+// outside the volume.
+func TestPrepareDataDirRefusesPathOutsideVolume(t *testing.T) {
+	t.Parallel()
+
+	volume := t.TempDir()
+
+	var stderr bytes.Buffer
+
+	code := cli.PrepareDataDir(&stderr, volume, "/etc", currentUsername(t))
+
+	want := "DATA_DIR '/etc': " + reportbuf.ErrDataDirOutsideVolume.Error() + "\n"
+	if code != 1 || stderr.String() != want {
+		t.Errorf("exit %d, stderr %q; want exit 1 and %q",
+			code, stderr.String(), want)
+	}
+}
+
+// currentUsername is the name of the user the test runs as, the only
+// owner a test not run as root can give files to.
+func currentUsername(t *testing.T) string {
+	t.Helper()
+
+	current, err := user.Current()
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	return current.Username
+}
+
+// failIfServerStarts is what the tests give Main in place of
+// cli.RunServer: it fails the test instead of starting the server.
+func failIfServerStarts(t *testing.T) func(appname, version string) {
+	t.Helper()
+
+	return func(string, string) {
+		t.Error("Main started the server")
+	}
+}

internal/cli/export_test.go

@@ -0,0 +1,16 @@
+package cli
+
+import "io"
+
+// CheckCIDR exposes the check-cidr subcommand to the external tests,
+// which read what it writes to stderr.
+func CheckCIDR(stderr io.Writer, cidr string) int {
+	return checkCIDR(stderr, cidr)
+}
+
+// PrepareDataDir exposes the prepare-data-dir subcommand to the
+// external tests, which give it a volume and a user of their own in
+// place of /data and netwatch.
+func PrepareDataDir(stderr io.Writer, volume, dataDir, username string) int {
+	return prepareDataDir(stderr, volume, dataDir, username)
+}

internal/config/config.go

@@ -41,7 +41,6 @@ var (
 		"must be an origin, scheme://host with an optional port",
 	)
 	errNotPort = errors.New("must be a port number, 1 to 65535")
-	errNotBool = errors.New("must be true or false")
 	errNotIP   = errors.New("must be an IP address, or empty")
 
 	errMetricsCredentials = errors.New(
@@ -101,7 +100,7 @@ func New(
 	viper.SetDefault("CORS_ALLOWED_ORIGINS", "")
 	viper.SetDefault("DATA_DIR", "./data/reports")
 	viper.SetDefault("DATA_DIR_MAX_BYTES", defaultDataDirMaxBytes)
-	viper.SetDefault("DEBUG", "false")
+	viper.SetDefault("DEBUG", "")
 	// An empty BIND_ADDRESS listens on every interface.
 	viper.SetDefault("BIND_ADDRESS", "")
 	viper.SetDefault("PORT", "8080")
@@ -120,19 +119,17 @@ func New(
 		}
 	}
 
-	// Read with strconv: viper's GetInt and GetBool would read a value
+	// Read with strconv: viper's GetInt would read a value it cannot
-	// they cannot parse as 0 or false instead of failing.
+	// parse as 0 instead of failing.
 	port, err := strconv.Atoi(viper.GetString("PORT"))
 	if err != nil || port < 1 || port > math.MaxUint16 {
 		return nil, fmt.Errorf("PORT %q: %w",
 			viper.GetString("PORT"), errNotPort)
 	}
 
-	debug, err := strconv.ParseBool(viper.GetString("DEBUG"))
+	// Any non-empty value turns debug logging on, "false" and "0"
-	if err != nil {
+	// included, as the code style guide says.
-		return nil, fmt.Errorf("DEBUG %q: %w",
+	debug := viper.GetString("DEBUG") != ""
-			viper.GetString("DEBUG"), errNotBool)
-	}
 
 	// An empty TRUSTED_PROXIES trusts no proxy, but viper reads an
 	// empty environment variable as unset and would give the default,
@@ -172,8 +169,8 @@ func New(
 }
 
 // check fails with an error naming the first setting here whose value
-// the server cannot use. New checks PORT and DEBUG as it reads them,
+// the server cannot use. New checks PORT as it reads it, and the
-// and the middleware checks TRUSTED_PROXIES as it parses it.
+// middleware checks TRUSTED_PROXIES as it parses it.
 func (s *Config) check() error {
 	// viper reads a value that is not a number as 0, so this also
 	// catches a mistyped setting.

internal/config/config_test.go

@@ -13,8 +13,8 @@ import (
 	"go.uber.org/fx"
 )
 
-// requireConfigError builds the config as main does and fails the
+// requireConfigError builds the config as cli.RunServer does and fails
-// test unless that fails with an error naming setting. It uses
+// the test unless that fails with an error naming setting. It uses
 // fx.New, because fxtest.New fails the test itself on an error.
 func requireConfigError(t *testing.T, setting string) {
 	t.Helper()
@@ -31,8 +31,8 @@ func requireConfigError(t *testing.T, setting string) {
 	}
 }
 
-// requireTrustedProxies builds the config as main does and fails the
+// requireTrustedProxies builds the config as cli.RunServer does and
-// test unless its TrustedProxies are want.
+// fails the test unless its TrustedProxies are want.
 func requireTrustedProxies(t *testing.T, want []string) {
 	t.Helper()
 
@@ -127,12 +127,52 @@ func TestPortMustBeAPortNumber(t *testing.T) {
 	}
 }
 
-// TestDebugMustBeTrueOrFalse: viper reads any other value, such as
+// TestDebugIsOnForAnyNonEmptyValue: unset or empty, debug logging is
-// "yes", as false.
+// off; any other value turns it on, "false" and "0" included, and none
-func TestDebugMustBeTrueOrFalse(t *testing.T) {
+// stops the start.
-	t.Setenv("DEBUG", "yes")
+func TestDebugIsOnForAnyNonEmptyValue(t *testing.T) {
+	for _, tc := range []struct {
+		name  string
+		value string
+		unset bool
+		want  bool
+	}{
+		{name: "unset", unset: true, want: false},
+		{name: "empty", value: "", want: false},
+		{name: "false", value: "false", want: true},
+		{name: "0", value: "0", want: true},
+		{name: "yes", value: "yes", want: true},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			// t.Setenv first, so the variable is put back as it was
+			// after the test.
+			t.Setenv("DEBUG", tc.value)
 
-	requireConfigError(t, "DEBUG")
+			if tc.unset {
+				err := os.Unsetenv("DEBUG")
+				if err != nil {
+					t.Fatal(err)
+				}
+			}
+
+			var cfg *config.Config
+
+			app := fx.New(
+				fx.NopLogger,
+				fx.Provide(globals.New, logger.New, config.New),
+				fx.Populate(&cfg),
+			)
+
+			err := app.Err()
+			if err != nil {
+				t.Fatalf("config error = %v", err)
+			}
+
+			if cfg.Debug != tc.want {
+				t.Fatalf("Debug = %t, want %t", cfg.Debug, tc.want)
+			}
+		})
+	}
 }
 
 // TestBindAddressMustBeAnIPAddress: a host name would be looked up

internal/globals/globals.go

@@ -4,7 +4,7 @@ package globals
 
 import "go.uber.org/fx"
 
-//nolint:gochecknoglobals // set from main before fx starts
+//nolint:gochecknoglobals // set by cli.RunServer before fx starts
 var (
 	// Appname is the application name.
 	Appname string

internal/handlers/healthcheck.go

@@ -3,9 +3,17 @@ package handlers
 import "net/http"
 
 // HandleHealthCheck returns a handler for the health check
-// endpoint.
+// endpoint. It answers 200 while the status is "ok" and 503 otherwise,
+// so that the image's HEALTHCHECK fails.
 func (s *Handlers) HandleHealthCheck() http.HandlerFunc {
 	return func(w http.ResponseWriter, r *http.Request) {
-		s.respondJSON(w, r, s.hc.Healthcheck(), http.StatusOK)
+		resp := s.hc.Healthcheck()
+
+		code := http.StatusOK
+		if resp.Status != "ok" {
+			code = http.StatusServiceUnavailable
+		}
+
+		s.respondJSON(w, r, resp, code)
 	}
 }

internal/handlers/healthcheck_test.go

@@ -9,17 +9,19 @@ import (
 	"testing"
 	"time"
 
+	"sneak.berlin/go/netwatch/internal/config"
 	"sneak.berlin/go/netwatch/internal/globals"
 	"sneak.berlin/go/netwatch/internal/handlers"
 	"sneak.berlin/go/netwatch/internal/healthcheck"
 	"sneak.berlin/go/netwatch/internal/logger"
+	"sneak.berlin/go/netwatch/internal/reportbuf"
 
 	"go.uber.org/fx/fxtest"
 )
 
 // newStartedHandlers builds Handlers with a real health check for the
-// server named in g, and starts them, which records the time the
+// server named in g, and a report buffer in a temporary directory, and
-// uptime counts from.
+// starts them, which records the time the uptime counts from.
 func newStartedHandlers(t *testing.T, g *globals.Globals) *handlers.Handlers {
 	t.Helper()
 
@@ -30,14 +32,23 @@ func newStartedHandlers(t *testing.T, g *globals.Globals) *handlers.Handlers {
 		t.Fatalf("logger: %v", err)
 	}
 
+	buf, err := reportbuf.New(lc, reportbuf.Params{
+		Config: &config.Config{DataDir: t.TempDir(), DataDirMaxBytes: 1 << 20},
+		Logger: log,
+	})
+	if err != nil {
+		t.Fatalf("report buffer: %v", err)
+	}
+
 	hc, err := healthcheck.New(lc,
-		healthcheck.Params{Globals: g, Logger: log})
+		healthcheck.Params{Buffer: buf, Globals: g, Logger: log})
 	if err != nil {
 		t.Fatalf("health check: %v", err)
 	}
 
-	h, err := handlers.New(lc,
+	h, err := handlers.New(lc, handlers.Params{
-		handlers.Params{Globals: g, Healthcheck: hc, Logger: log})
+		Buffer: buf, Globals: g, Healthcheck: hc, Logger: log,
+	})
 	if err != nil {
 		t.Fatalf("handlers: %v", err)
 	}
@@ -48,10 +59,12 @@ func newStartedHandlers(t *testing.T, g *globals.Globals) *handlers.Handlers {
 	return h
 }
 
-// TestHandleHealthCheck checks the health check's answer: 200, a JSON
+// TestHandleHealthCheck checks the health check's answer at start, when
-// content type, and a JSON object with exactly the fields of
+// no write of a report file has failed: 200, a JSON content type, and a
-// healthcheck.HealthcheckResponse, carrying this server's name and
+// JSON object with exactly the fields of healthcheck.HealthcheckResponse,
-// version and an uptime counted from its start.
+// carrying "status":"ok", this server's name and version and an uptime
+// counted from its start. The answer while writes fail is checked in
+// internal/reportbuf, whose tests can make a write fail.
 func TestHandleHealthCheck(t *testing.T) {
 	t.Parallel()
 

internal/healthcheck/healthcheck.go

@@ -10,6 +10,7 @@ import (
 	"sneak.berlin/go/netwatch/internal/config"
 	"sneak.berlin/go/netwatch/internal/globals"
 	"sneak.berlin/go/netwatch/internal/logger"
+	"sneak.berlin/go/netwatch/internal/reportbuf"
 
 	"go.uber.org/fx"
 )
@@ -18,6 +19,7 @@ import (
 type Params struct {
 	fx.In
 
+	Buffer  *reportbuf.Buffer
 	Config  *config.Config
 	Globals *globals.Globals
 	Logger  *logger.Logger
@@ -68,12 +70,18 @@ func New(
 	return s, nil
 }
 
-// Healthcheck returns the current health status of the application.
+// Healthcheck returns the current health status of the application:
+// "ok", or "error" while report files cannot be written.
 func (s *Healthcheck) Healthcheck() *HealthcheckResponse {
+	status := "ok"
+	if s.params.Buffer.LastWriteFailed() {
+		status = "error"
+	}
+
 	return &HealthcheckResponse{
 		Appname:       s.params.Globals.Appname,
 		Now:           time.Now().UTC().Format(time.RFC3339Nano),
-		Status:        "ok",
+		Status:        status,
 		UptimeHuman:   s.uptime().String(),
 		UptimeSeconds: int64(s.uptime().Seconds()),
 		Version:       s.params.Globals.Version,

internal/middleware/middleware.go

@@ -11,6 +11,7 @@ import (
 	"net/http"
 	"net/netip"
 	"runtime/debug"
+	"slices"
 	"strings"
 	"time"
 
@@ -133,6 +134,17 @@ func ipFromHostPort(hostPort string) string {
 // X-Real-IP are honoured only when the direct peer is a
 // trusted proxy; otherwise the direct peer is returned so a
 // spoofed header cannot forge the logged address.
+//
+// Each proxy appends to X-Forwarded-For the address it took the
+// request from, so the entries on the right are written by the
+// trusted proxies and those further left by the client. The
+// client's address is therefore the right-most entry that is not
+// a trusted proxy. An entry written as an address with a port is
+// read as that address, both for the trust check and as the
+// result. When every entry is a trusted proxy, or the walk from
+// the right meets an entry that is not an address at all (such as
+// "unknown"), the peer is used. X-Real-IP is read only when there
+// is no X-Forwarded-For entry at all.
 func clientIP(
 	remoteAddr string,
 	header http.Header,
@@ -144,28 +156,49 @@ func clientIP(
 		return peer
 	}
 
-	if xff := firstForwardedFor(header.Get("X-Forwarded-For")); xff != "" {
+	forwarded := forwardedFor(header)
-		return xff
+
+	if len(forwarded) == 0 {
+		if xr := strings.TrimSpace(header.Get("X-Real-IP")); validIP(xr) {
+			return xr
+		}
+
+		return peer
 	}
 
-	if xr := strings.TrimSpace(header.Get("X-Real-IP")); validIP(xr) {
+	for _, entry := range slices.Backward(forwarded) {
-		return xr
+		if !validIP(entry) {
+			return peer
+		}
+
+		if !addrInAny(entry, trusted) {
+			return entry
+		}
 	}
 
 	return peer
 }
 
-// firstForwardedFor returns the left-most valid address in an
+// forwardedFor returns the entries of X-Forwarded-For in order,
-// X-Forwarded-For list (the original client), or "" if none.
+// leaving out blank ones. A proxy may add its entry as a header
-func firstForwardedFor(value string) string {
+// line of its own rather than append it to the line the client
+// sent, so every line is read, not only the first. Some proxies
+// write an entry as an address with a port, such as
+// 198.51.100.7:5000 or [2001:db8::1]:443; the port is removed so
+// that the entry is read as its address.
+func forwardedFor(header http.Header) []string {
+	value := strings.Join(header.Values("X-Forwarded-For"), ",")
+
+	var entries []string
+
 	for part := range strings.SplitSeq(value, ",") {
-		candidate := strings.TrimSpace(part)
+		entry := strings.TrimSpace(part)
-		if validIP(candidate) {
+		if entry != "" {
-			return candidate
+			entries = append(entries, ipFromHostPort(entry))
 		}
 	}
 
-	return ""
+	return entries
 }
 
 func validIP(s string) bool {

internal/middleware/middleware_test.go

@@ -25,10 +25,16 @@ import (
 const (
 	// loopbackPeer is a remote address inside the trusted-proxy allowlist.
 	loopbackPeer = "127.0.0.1:5000"
+	// loopbackIP is the address of loopbackPeer, which ClientIP returns
+	// when it uses the peer.
+	loopbackIP = "127.0.0.1"
 	// forwardedIP is the client address presented via X-Forwarded-For.
 	forwardedIP = "203.0.113.7"
 	// realIP is the client address presented via X-Real-IP.
 	realIP = "203.0.113.9"
+	// appendedIP is the X-Forwarded-For entry a trusted proxy appended
+	// after the entries the client wrote.
+	appendedIP = "198.51.100.7"
 )
 
 func mustPrefixes(t *testing.T, cidrs ...string) []netip.Prefix {
@@ -85,10 +91,35 @@ func clientIPCases() []clientIPCase {
 			want:       forwardedIP,
 		},
 		{
-			name:       "trusted proxy uses left-most of chain",
+			name:       "trusted proxy uses right-most of chain",
-			remoteAddr: "10.1.2.3:5000",
+			remoteAddr: loopbackPeer,
-			xff:        forwardedIP + ", 10.1.2.3",
+			xff:        "203.0.113.9, " + appendedIP,
-			want:       forwardedIP,
+			want:       appendedIP,
+		},
+		{
+			name:       "trusted proxy entries are skipped from the right",
+			remoteAddr: loopbackPeer,
+			xff:        "203.0.113.9, 10.0.0.5",
+			want:       "203.0.113.9",
+		},
+		{
+			name:       "chain of only trusted proxies uses peer",
+			remoteAddr: loopbackPeer,
+			xff:        "10.0.0.5, 10.1.2.3",
+			want:       loopbackIP,
+		},
+		{
+			name:       "entry that is not an address uses peer",
+			remoteAddr: loopbackPeer,
+			xff:        "203.0.113.9, unknown",
+			want:       loopbackIP,
+		},
+		{
+			name:       "x-real-ip ignored when forwarded-for has entries",
+			remoteAddr: loopbackPeer,
+			xff:        "10.0.0.5",
+			xRealIP:    realIP,
+			want:       loopbackIP,
 		},
 		{
 			name:       "trusted proxy falls back to x-real-ip",
@@ -117,7 +148,7 @@ func clientIPCases() []clientIPCase {
 			name:       "trusted proxy with garbage header uses peer",
 			remoteAddr: loopbackPeer,
 			xff:        "not-an-ip",
-			want:       "127.0.0.1",
+			want:       loopbackIP,
 		},
 	}
 }
@@ -148,6 +179,47 @@ func TestClientIP(t *testing.T) {
 	}
 }
 
+// TestClientIPReadsEveryForwardedForLine checks that an entry a proxy
+// adds as an X-Forwarded-For line of its own is read, and the line the
+// client sent before it is not taken.
+func TestClientIPReadsEveryForwardedForLine(t *testing.T) {
+	t.Parallel()
+
+	header := http.Header{}
+	header.Add("X-Forwarded-For", "203.0.113.9")
+	header.Add("X-Forwarded-For", appendedIP)
+
+	got := middleware.ClientIP(loopbackPeer, header,
+		mustPrefixes(t, "127.0.0.1/32"))
+	if got != appendedIP {
+		t.Errorf("ClientIP() = %q, want %q", got, appendedIP)
+	}
+}
+
+// TestClientIPReadsAddressWithPort checks that an X-Forwarded-For entry
+// written as an address with a port is read as that address, both for
+// the trust check and as the result.
+func TestClientIPReadsAddressWithPort(t *testing.T) {
+	t.Parallel()
+
+	trusted := mustPrefixes(t, "127.0.0.1/32", "10.0.0.0/8")
+
+	for xff, want := range map[string]string{
+		"203.0.113.9, " + appendedIP + ":5000": appendedIP,
+		"203.0.113.9, [2001:db8::1]:443":       "2001:db8::1",
+		"203.0.113.9, 10.0.0.5:5000":           "203.0.113.9",
+	} {
+		header := http.Header{}
+		header.Set("X-Forwarded-For", xff)
+
+		got := middleware.ClientIP(loopbackPeer, header, trusted)
+		if got != want {
+			t.Errorf("X-Forwarded-For %q: ClientIP() = %q, want %q",
+				xff, got, want)
+		}
+	}
+}
+
 func TestSecurityHeaders(t *testing.T) {
 	t.Parallel()
 

internal/reportbuf/datadir_linux.go

@@ -24,9 +24,10 @@ import (
 // its number of names from the open file and gives that file to them,
 // so a hard link swapped in at name is caught too.
 //
-// This file is built only on Linux, because O_PATH and AT_EMPTY_PATH
+// This file is built only on Linux: prepare-data-dir runs only in the
-// exist only there; prepare-data-dir runs only in the Linux image. On
+// Linux image, and macOS, where the backend is also built for
-// any other system chownFile returns an error (see datadir_other.go).
+// development, has neither O_PATH nor AT_EMPTY_PATH. On any other
+// system chownFile returns an error (see datadir_other.go).
 func chownFile(root *os.Root, name string, uid, gid int) error {
 	file, err := root.OpenFile(name, unix.O_PATH, 0)
 	if err != nil {

internal/reportbuf/datadir_other.go

@@ -10,11 +10,12 @@ import (
 
 // chownFile returns an error on every system but Linux. Opening a file
 // that is not a directory without reading it or following it, and
-// giving it to a user on the open file, never by name, needs O_PATH
+// giving it to a user on the open file, never by name, is built only
-// and AT_EMPTY_PATH, which only Linux has (see datadir_linux.go), and
+// for Linux (see datadir_linux.go): prepare-data-dir runs only in the
-// prepare-data-dir runs only in the Linux image. This file lets the
+// Linux image, and macOS, where the backend is also built for
-// package build elsewhere, such as on macOS, where a developer runs
+// development, has neither O_PATH nor AT_EMPTY_PATH. This file lets
-// make build and make run.
+// the package build elsewhere, such as on macOS, where a developer
+// runs make build and make run.
 func chownFile(_ *os.Root, name string, _, _ int) error {
 	return &fs.PathError{Op: "chown", Path: name, Err: errors.ErrUnsupported}
 }

internal/reportbuf/reportbuf.go

@@ -16,7 +16,6 @@ import (
 	"log/slog"
 	"os"
 	"path/filepath"
-	"slices"
 	"strings"
 	"sync"
 	"sync/atomic"
@@ -73,11 +72,9 @@ type Buffer struct {
 	// except in tests that hold the write open or make it fail.
 	fileCreated func(f *os.File)
 	// files are the report files that may be deleted to make room,
-	// in name order, which is oldest first: those in dataDir at
+	// oldest first: those in dataDir at start, in name order, then
-	// start, and each one this buffer writes, put in at its place by
+	// each one this buffer writes, once it is complete. A file still
-	// name once it is complete, even when an older file's write
+	// being written is not among them. filesBytes is their total size.
-	// completes after a newer one's. A file still being written is
-	// not among them. filesBytes is their total size.
 	files      []reportFile
 	filesBytes int64
 	log        *slog.Logger
@@ -88,8 +85,12 @@ type Buffer struct {
 	now func() time.Time
 	// seq numbers the report files, so that two named in the same
 	// millisecond still get different names.
-	seq      atomic.Uint64
+	seq      atomic.Int64
 	stopOnce sync.Once
+	// thresholdWriteWaiting is set while a write Append started at
+	// flushSizeThreshold waits for writeMu; Append starts no other
+	// while it is set.
+	thresholdWriteWaiting atomic.Bool
 	// usedBytes is what Append checks against maxBytes: the size
 	// of the report files in dataDir, plus the reports waiting to
 	// be written to one at their uncompressed size.
@@ -99,6 +100,11 @@ type Buffer struct {
 	// refuses reports, and the reports of that write wait in buf for
 	// the next flush.
 	writeErr error
+	// writeMu is held across each write of a report file, so writes
+	// never overlap: a failed write puts its reports back in buf before
+	// the next write takes them, which keeps them in the order they
+	// were appended.
+	writeMu sync.Mutex
 	// writers are the goroutines that write report files: flushLoop,
 	// and each one Append starts at flushSizeThreshold. OnStop waits
 	// for them before the final flush.
@@ -127,16 +133,33 @@ func New(
 
 	lc.Append(fx.Hook{
 		OnStart: func(_ context.Context) error {
+			// The errors name DATA_DIR, as config.New's do for the
+			// settings it checks.
 			err := os.MkdirAll(b.dataDir, dirPerms)
 			if err != nil {
-				return fmt.Errorf("create data dir: %w", err)
+				return fmt.Errorf("DATA_DIR %q: %w", b.dataDir, err)
+			}
+
+			// Creating and removing a file stops the start when no
+			// file can be created in DATA_DIR, instead of the first
+			// write of a report file failing after reports were
+			// answered 200. A full disk is not caught: the empty
+			// file is still created.
+			f, err := os.CreateTemp(b.dataDir, "")
+			if err != nil {
+				return fmt.Errorf("DATA_DIR %q: %w", b.dataDir, err)
+			}
+
+			err = errors.Join(f.Close(), os.Remove(f.Name()))
+			if err != nil {
+				return fmt.Errorf("DATA_DIR %q: %w", b.dataDir, err)
 			}
 
 			// Report files left by earlier runs count too, and are
 			// the first to be deleted to make room.
 			files, err := reportFiles(b.dataDir)
 			if err != nil {
-				return err
+				return fmt.Errorf("DATA_DIR %q: %w", b.dataDir, err)
 			}
 
 			b.mu.Lock()
@@ -190,8 +213,9 @@ func New(
 // write of a report file failed, until a write succeeds again. If the
 // line would take usedBytes past maxBytes, the oldest report files are
 // deleted to make room; it stores nothing and returns ErrFull if that
-// cannot make room. If the buffer reaches the size threshold, it is
+// cannot make room. If the buffer reaches the size threshold, it starts
-// drained and written to disk asynchronously.
+// a write of the reports in the background (see flushAtThreshold),
+// unless one it started is still waiting.
 func (b *Buffer) Append(v any) error {
 	line, err := json.Marshal(v)
 	if err != nil {
@@ -221,18 +245,9 @@ func (b *Buffer) Append(v any) error {
 	b.buf.Write(line)
 	b.buf.WriteByte('\n')
 
-	if b.buf.Len() >= flushSizeThreshold {
+	if b.buf.Len() >= flushSizeThreshold &&
-		data := b.drainBuf()
+		b.thresholdWriteWaiting.CompareAndSwap(false, true) {
-		b.mu.Unlock()
+		b.writers.Go(b.flushAtThreshold)
-
-		b.writers.Go(func() {
-			writeErr := b.writeFile(data)
-			if writeErr != nil {
-				b.log.Error("flush reports failed", "error", writeErr)
-			}
-		})
-
-		return nil
 	}
 
 	b.mu.Unlock()
@@ -240,6 +255,16 @@ func (b *Buffer) Append(v any) error {
 	return nil
 }
 
+// LastWriteFailed reports whether the last write of a report file to
+// finish failed: false at start until a write fails, and false again
+// once a write succeeds. While it is true, Append refuses reports.
+func (b *Buffer) LastWriteFailed() bool {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+
+	return b.writeErr != nil
+}
+
 // deleteOldestFiles deletes report files, oldest first, until n more
 // bytes fit under maxBytes. It deletes none when the reports waiting
 // to be written leave no room for n even with every file gone, since
@@ -290,21 +315,28 @@ func (b *Buffer) flushLoop() {
 	}
 }
 
-// flushLocked acquires the lock, drains the buffer, and
+// flushLocked waits for any write of a report file under way, then
-// writes the data to a compressed file.
+// writes the reports waiting in the buffer to a new one.
 func (b *Buffer) flushLocked() error {
-	b.mu.Lock()
+	b.writeMu.Lock()
+	defer b.writeMu.Unlock()
 
-	if b.buf.Len() == 0 {
+	return b.writeFile()
-		b.mu.Unlock()
+}
 
-		return nil
+// flushAtThreshold is the write Append starts when the buffer reaches
+// flushSizeThreshold. Once any write under way is done, it lets Append
+// start another and writes the reports waiting in the buffer.
+func (b *Buffer) flushAtThreshold() {
+	b.writeMu.Lock()
+	defer b.writeMu.Unlock()
+
+	b.thresholdWriteWaiting.Store(false)
+
+	err := b.writeFile()
+	if err != nil {
+		b.log.Error("flush reports failed", "error", err)
 	}
-
-	data := b.drainBuf()
-	b.mu.Unlock()
-
-	return b.writeFile(data)
 }
 
 // drainBuf copies the buffer contents and resets it.
@@ -317,10 +349,22 @@ func (b *Buffer) drainBuf() []byte {
 	return data
 }
 
-// writeFile writes data, reports drained from the buffer, to a new
+// writeFile takes the reports waiting in the buffer, if any, and writes
-// timestamped zstd-compressed JSONL file in the data directory. If the
+// them to a new timestamped zstd-compressed JSONL file in the data
-// write fails, it puts the reports back in the buffer.
+// directory. If the write fails, it puts the reports back in the
-func (b *Buffer) writeFile(data []byte) error {
+// buffer. The caller must hold b.writeMu.
+func (b *Buffer) writeFile() error {
+	b.mu.Lock()
+
+	if b.buf.Len() == 0 {
+		b.mu.Unlock()
+
+		return nil
+	}
+
+	data := b.drainBuf()
+	b.mu.Unlock()
+
 	// The timestamp comes first, so the names sort by time; the number
 	// after it tells apart files named in the same millisecond.
 	ts := b.now().UTC().Format("2006-01-02T15-04-05.000Z")
@@ -351,14 +395,7 @@ func (b *Buffer) writeFile(data []byte) error {
 	b.usedBytes -= int64(len(data))
 	b.usedBytes += size
 	b.filesBytes += size
-
+	b.files = append(b.files, reportFile{name: name, size: size})
-	// At its place by name, not at the end: another write, of a newer
-	// file, may have completed while this one was being written.
-	i, _ := slices.BinarySearchFunc(b.files, name,
-		func(f reportFile, target string) int {
-			return strings.Compare(f.name, target)
-		})
-	b.files = slices.Insert(b.files, i, reportFile{name: name, size: size})
 
 	return nil
 }

internal/reportbuf/reportbuf_test.go

@@ -108,6 +108,68 @@ func TestFailedFinalFlushFailsStop(t *testing.T) {
 	}
 }
 
+// TestUncreatableDataDirStopsStart sets DATA_DIR below a regular file,
+// where no directory can be created, even by root, as the tests run in
+// the backend image. The start must fail with an error naming DATA_DIR
+// and its value, as the errors for the other settings do.
+func TestUncreatableDataDirStopsStart(t *testing.T) {
+	file := filepath.Join(t.TempDir(), "file")
+	writeBytes(t, file, 1)
+
+	dir := filepath.Join(file, "reports")
+	t.Setenv("DATA_DIR", dir)
+
+	app := fxtest.New(t,
+		fx.Provide(
+			globals.New,
+			logger.New,
+			config.New,
+			reportbuf.New,
+		),
+		fx.Invoke(func(*reportbuf.Buffer) {}),
+	)
+
+	err := app.Start(t.Context())
+	if err == nil {
+		t.Fatal("start succeeded with DATA_DIR below a regular file")
+	}
+
+	want := fmt.Sprintf("DATA_DIR %q: ", dir)
+	if !strings.Contains(err.Error(), want) {
+		t.Fatalf("start error = %q, want it to contain %q", err, want)
+	}
+}
+
+// TestUnwritableDataDirStopsStart sets DATA_DIR to /proc, which exists
+// and can be read, but where no file can be created, even by root, as
+// the tests run in the backend image: root ignores a directory the test
+// makes read-only. The start must fail with an error naming DATA_DIR and
+// its value, rather than the first write of a report file.
+func TestUnwritableDataDirStopsStart(t *testing.T) {
+	dir := "/proc"
+	t.Setenv("DATA_DIR", dir)
+
+	app := fxtest.New(t,
+		fx.Provide(
+			globals.New,
+			logger.New,
+			config.New,
+			reportbuf.New,
+		),
+		fx.Invoke(func(*reportbuf.Buffer) {}),
+	)
+
+	err := app.Start(t.Context())
+	if err == nil {
+		t.Fatal("start succeeded with DATA_DIR /proc")
+	}
+
+	want := fmt.Sprintf("DATA_DIR %q: ", dir)
+	if !strings.Contains(err.Error(), want) {
+		t.Fatalf("start error = %q, want it to contain %q", err, want)
+	}
+}
+
 // TestStopWaitsForPeriodicWrite holds open the write the periodic flush
 // starts, then stops the app. The stop must wait until the write is
 // released, and the report must then be in its file. The test runs in
@@ -272,9 +334,9 @@ func stopWhileHeld(
 	return <-stopped
 }
 
-// startApp starts a Buffer through fx, as main does, with the DATA_DIR
+// startApp starts a Buffer through fx, as cli.RunServer does, with the
-// and DATA_DIR_MAX_BYTES the calling test has set, and returns it with
+// DATA_DIR and DATA_DIR_MAX_BYTES the calling test has set, and returns
-// the app, for tests that stop the app themselves.
+// it with the app, for tests that stop the app themselves.
 func startApp(t *testing.T) (*reportbuf.Buffer, *fxtest.App) {
 	t.Helper()
 
@@ -296,8 +358,8 @@ func startApp(t *testing.T) (*reportbuf.Buffer, *fxtest.App) {
 	return buf, app
 }
 
-// startBuffer starts a Buffer through fx, as main does, with the
+// startBuffer starts a Buffer through fx, as cli.RunServer does, with
-// DATA_DIR and DATA_DIR_MAX_BYTES the calling test has set.
+// the DATA_DIR and DATA_DIR_MAX_BYTES the calling test has set.
 func startBuffer(t *testing.T) *reportbuf.Buffer {
 	t.Helper()
 
@@ -519,6 +581,62 @@ func TestWrittenFilesDeletedToMakeRoom(t *testing.T) {
 	}
 }
 
+// TestWrittenFilesDeletedOldestFirst writes three report files, then
+// appends a report that fits only once one of them is deleted. The
+// first one written must be the one deleted.
+func TestWrittenFilesDeletedOldestFirst(t *testing.T) {
+	const maxBytes = 1000
+
+	report := map[string]string{"id": "written"}
+
+	t.Setenv("DATA_DIR", t.TempDir())
+	t.Setenv("DATA_DIR_MAX_BYTES", strconv.Itoa(maxBytes))
+
+	buf := startBuffer(t)
+
+	var written []string
+
+	buf.OnFileCreated(func(f *os.File) { written = append(written, f.Name()) })
+
+	for range 3 {
+		err := buf.Append(report)
+		if err != nil {
+			t.Fatalf("append report: %v", err)
+		}
+
+		err = buf.Flush()
+		if err != nil {
+			t.Fatalf("flush: %v", err)
+		}
+	}
+
+	// The three files hold the same report, so they are the same size.
+	info, err := os.Stat(written[0])
+	if err != nil {
+		t.Fatalf("stat %s: %v", written[0], err)
+	}
+
+	// A report that fits beside two of the files, so deleting one makes
+	// exactly the room it needs.
+	pad := maxBytes - 2*int(info.Size()) -
+		lineBytes(t, map[string]string{"id": ""})
+
+	err = buf.Append(map[string]string{"id": strings.Repeat("a", pad)})
+	if err != nil {
+		t.Fatalf("report that needs room: %v", err)
+	}
+
+	if exists(t, written[0]) {
+		t.Fatal("first report file written kept when room was needed")
+	}
+
+	for _, path := range written[1:] {
+		if !exists(t, path) {
+			t.Fatalf("%s deleted before the first report file written", path)
+		}
+	}
+}
+
 // TestFailedDeletionStillCounts makes deleting the oldest report file
 // fail. It is still there, so it still takes room, and the next oldest
 // is deleted in its place.
@@ -604,17 +722,16 @@ func TestFileDeletedByHandFreesRoom(t *testing.T) {
 	}
 }
 
-// TestFileBeingWrittenIsNeverDeleted holds the write of one report file
+// TestFileBeingWrittenIsNeverDeleted holds the write of a report file
-// open while a second write completes, then sends a report that needs
+// open under a cap with room for two reports, then appends a second
-// room. Deleting either file would make it, and the one being written is
+// report and a third that needs room. The file being written is the
-// the older, but only the complete one may be deleted. Once the first
+// only report file, and it is not deleted to make room, so the third
-// write is complete, its file is deleted when room is needed.
+// report is refused. Once the write is complete, its file is deleted
+// when room is needed.
 func TestFileBeingWrittenIsNeverDeleted(t *testing.T) {
 	report := map[string]string{"id": "writing"}
 
 	t.Setenv("DATA_DIR", t.TempDir())
-	// Room for two reports waiting to be written, but not for two
-	// beside a report file.
 	t.Setenv("DATA_DIR_MAX_BYTES", strconv.Itoa(2*lineBytes(t, report)))
 
 	buf := startBuffer(t)
@@ -639,36 +756,25 @@ func TestFileBeingWrittenIsNeverDeleted(t *testing.T) {
 
 	writing := <-created
 
-	// Only the first write is held; the second goes through, and
+	// Errorf, not Fatalf, until the write is released, so that a
-	// so do the writes after it, the final one at stop included.
-	var complete string
-
-	buf.OnFileCreated(func(f *os.File) { complete = f.Name() })
-
-	// Errorf, not Fatalf, until the first write is released, so that a
 	// failure here does not leave it held.
 	err = buf.Append(report)
 	if err != nil {
 		t.Errorf("second report: %v", err)
 	}
 
-	err = buf.Flush()
-	if err != nil {
-		t.Errorf("flush of the second report: %v", err)
-	}
-
 	err = buf.Append(report)
-	if err != nil {
+	if !errors.Is(err, reportbuf.ErrFull) {
-		t.Errorf("report that needs room: %v", err)
+		t.Errorf("report that needs room: error = %v, want ErrFull", err)
 	}
 
 	if !exists(t, writing) {
 		t.Error("report file deleted while it was being written")
 	}
 
-	if exists(t, complete) {
+	// Only this write is held; the writes after it, the final one at
-		t.Error("complete report file kept, though room was needed")
+	// stop included, go through.
-	}
+	buf.OnFileCreated(func(*os.File) {})
 
 	close(release)
 
@@ -677,9 +783,11 @@ func TestFileBeingWrittenIsNeverDeleted(t *testing.T) {
 		t.Fatalf("flush of the first report: %v", err)
 	}
 
+	// The second report still waits to be written, so this one fits
+	// only once the file just written is deleted.
 	err = buf.Append(report)
 	if err != nil {
-		t.Fatalf("report after the first write was complete: %v", err)
+		t.Fatalf("report after the write was complete: %v", err)
 	}
 
 	if exists(t, writing) {
@@ -821,6 +929,82 @@ func TestFailedWriteKeepsReportsAppendedDuringIt(t *testing.T) {
 	}
 }
 
+// TestFailedWritesKeepReportsInOrder holds a write open, then appends a
+// report that takes the buffer to FlushSizeThreshold, so that Append
+// starts a second write. The held write fails first, then the second,
+// which waited for it and took both reports. The next flush writes them
+// in the order they were appended, the held write's first.
+func TestFailedWritesKeepReportsInOrder(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("DATA_DIR", dir)
+
+	buf := startBuffer(t)
+
+	created := make(chan struct{})
+	release := make(chan struct{})
+
+	// Each write is held until release is closed, then fails, as its
+	// file is closed under it.
+	buf.OnFileCreated(func(f *os.File) {
+		created <- struct{}{}
+
+		<-release
+
+		_ = f.Close()
+	})
+
+	err := buf.Append(map[string]string{"id": "held"})
+	if err != nil {
+		t.Fatalf("report of the held write: %v", err)
+	}
+
+	flushed := make(chan error)
+
+	go func() { flushed <- buf.Flush() }()
+
+	<-created
+
+	big := strings.Repeat("a", reportbuf.FlushSizeThreshold)
+
+	// Errorf, not Fatalf, until the held write is released, so that a
+	// failure here does not leave it held.
+	err = buf.Append(map[string]string{"id": big})
+	if err != nil {
+		t.Errorf("report that starts the second write: %v", err)
+	}
+
+	close(release)
+
+	err = <-flushed
+	if err == nil {
+		t.Error("held write succeeded, though its file was closed under it")
+	}
+
+	// The second write has created its file, so it has taken the
+	// reports.
+	<-created
+
+	// Writes from here on, the final one at stop included, succeed.
+	buf.OnFileCreated(func(*os.File) {})
+
+	// Flush waits for the second write to fail before it writes.
+	err = buf.Flush()
+	if err != nil {
+		t.Fatalf("flush after the failed writes: %v", err)
+	}
+
+	files, err := readReportFiles(dir)
+	if err != nil {
+		t.Fatalf("read report files: %v", err)
+	}
+
+	want := `{"id":"held"}` + "\n" + `{"id":"` + big + `"}` + "\n"
+	if len(files) != 1 || files[0] != want {
+		t.Fatalf("%d report files, want one holding the held write's "+
+			"report, then the second one", len(files))
+	}
+}
+
 // TestReportsRefusedWhileWritesFail posts reports through the report
 // handler, as the page does. Once a write has failed, a report is
 // refused with 500 until a write succeeds again; the reports taken
@@ -890,9 +1074,54 @@ func TestReportsRefusedWhileWritesFail(t *testing.T) {
 	}
 }
 
+// TestHealthCheckFailsWhileWritesFail asks the health check handler
+// for its answer around a failed write. It answers 200 with
+// "status":"ok" at start, 503 with "status":"error" from the failed
+// write until a write succeeds, and 200 with "status":"ok" again after.
+func TestHealthCheckFailsWhileWritesFail(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("DATA_DIR", dir)
+
+	buf, h := startBufferWithHandlers(t)
+
+	checkHealthCheck(t, h, "at start", http.StatusOK, "ok")
+
+	// A flush writes a file only when a report is waiting.
+	err := buf.Append(map[string]string{"id": "kept"})
+	if err != nil {
+		t.Fatalf("append: %v", err)
+	}
+
+	// With the data directory gone, the write has nowhere to go.
+	err = os.Remove(dir)
+	if err != nil {
+		t.Fatalf("remove data dir: %v", err)
+	}
+
+	err = buf.Flush()
+	if err == nil {
+		t.Fatal("flush succeeded with the data directory gone")
+	}
+
+	checkHealthCheck(t, h, "after the failed write",
+		http.StatusServiceUnavailable, "error")
+
+	err = os.Mkdir(dir, 0o750)
+	if err != nil {
+		t.Fatalf("make data dir: %v", err)
+	}
+
+	err = buf.Flush()
+	if err != nil {
+		t.Fatalf("flush with the data directory back: %v", err)
+	}
+
+	checkHealthCheck(t, h, "after a write succeeded", http.StatusOK, "ok")
+}
+
 // startBufferWithHandlers starts a Buffer and the Handlers that store
-// reports in it through fx, as main does, with the DATA_DIR the calling
+// reports in it through fx, as cli.RunServer does, with the DATA_DIR
-// test has set.
+// the calling test has set.
 func startBufferWithHandlers(
 	t *testing.T,
 ) (*reportbuf.Buffer, *handlers.Handlers) {
@@ -937,6 +1166,39 @@ func postReport(t *testing.T, h *handlers.Handlers, clientID string) int {
 	return rec.Code
 }
 
+// checkHealthCheck asks the health check handler for its answer, and
+// fails the test, saying when it asked, unless the answer has status
+// code wantCode and a "status" field of wantStatus.
+func checkHealthCheck(
+	t *testing.T,
+	h *handlers.Handlers,
+	when string,
+	wantCode int,
+	wantStatus string,
+) {
+	t.Helper()
+
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequestWithContext(t.Context(),
+		http.MethodGet, "/.well-known/healthcheck", http.NoBody)
+	h.HandleHealthCheck().ServeHTTP(rec, req)
+
+	var body struct {
+		Status string `json:"status"`
+	}
+
+	err := json.Unmarshal(rec.Body.Bytes(), &body)
+	if err != nil {
+		t.Fatalf("health check %s: body not JSON: %v (%q)",
+			when, err, rec.Body.String())
+	}
+
+	if rec.Code != wantCode || body.Status != wantStatus {
+		t.Fatalf("health check %s: %d with status %q, want %d with %q",
+			when, rec.Code, body.Status, wantCode, wantStatus)
+	}
+}
+
 // TestConcurrentAppendsStopAtCap appends from many goroutines at once
 // with room for exactly roomFor reports: exactly that many must be
 // taken, which holds only if Append checks and counts each report
@@ -1088,6 +1350,106 @@ func TestFlushAtSizeThreshold(t *testing.T) {
 	}
 }
 
+// TestEachSizeThresholdStartsAWrite reaches FlushSizeThreshold twice,
+// with no call to Flush and the periodic flush a minute away. Append
+// must start a write each time, so that the reports end up in two
+// report files. The test runs in a synctest bubble, so that
+// synctest.Wait returns once each write is done.
+func TestEachSizeThresholdStartsAWrite(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("DATA_DIR", dir)
+
+	synctest.Test(t, func(t *testing.T) {
+		buf := startBuffer(t)
+
+		var appended string
+
+		for range 2 {
+			appended = appendToThreshold(t, buf)
+
+			synctest.Wait()
+		}
+
+		files, err := readReportFiles(dir)
+		if err != nil {
+			t.Fatalf("read report files: %v", err)
+		}
+
+		if !slices.Equal(files, []string{appended, appended}) {
+			t.Fatalf("%d report files, want two, each holding the %d "+
+				"bytes appended to reach the threshold",
+				len(files), len(appended))
+		}
+	})
+}
+
+// TestAppendStartsOneWaitingWrite holds a write open, then appends two
+// reports of FlushSizeThreshold bytes each. Append starts a write at
+// the first, which waits for the held one, and none at the second,
+// since that write is still waiting. Every write fails, so a write
+// started at the second report would find the reports back in the
+// buffer and try them again: once the held write is released, two
+// writes must have been started, the held one and one more.
+func TestAppendStartsOneWaitingWrite(t *testing.T) {
+	t.Setenv("DATA_DIR", t.TempDir())
+
+	synctest.Test(t, func(t *testing.T) {
+		buf := startBuffer(t)
+
+		var writes atomic.Int64
+
+		release := make(chan struct{})
+
+		// Each write is held until release is closed, then fails, as its
+		// file is closed under it.
+		buf.OnFileCreated(func(f *os.File) {
+			writes.Add(1)
+
+			<-release
+
+			_ = f.Close()
+		})
+
+		err := buf.Append(map[string]string{"id": "one-waiting"})
+		if err != nil {
+			t.Fatalf("report of the held write: %v", err)
+		}
+
+		go func() { _ = buf.Flush() }()
+
+		// Wait returns once the flush's write is held.
+		synctest.Wait()
+
+		big := map[string]string{
+			"id": strings.Repeat("a", reportbuf.FlushSizeThreshold),
+		}
+
+		// Errorf, not Fatalf, until the held write is released, so that a
+		// failure here does not leave it held.
+		for range 2 {
+			err = buf.Append(big)
+			if err != nil {
+				t.Errorf("report of FlushSizeThreshold bytes: %v", err)
+			}
+		}
+
+		close(release)
+
+		// Wait returns once every write started is done.
+		synctest.Wait()
+
+		started := writes.Load()
+
+		// Writes from here on, the final one at stop included, succeed.
+		buf.OnFileCreated(func(*os.File) {})
+
+		if started != 2 {
+			t.Fatalf("%d writes started, want 2: the held one and one "+
+				"at the threshold", started)
+		}
+	})
+}
+
 // appendToThreshold appends reports to buf until it holds
 // FlushSizeThreshold bytes, so that the last append starts writing them
 // to a report file, and returns the lines appended.
@@ -1233,85 +1595,3 @@ func hasReportFile(t *testing.T, dir string) bool {
 
 	return false
 }
-
-// TestFilesDeletedOldestFirstWhenWritesOverlap holds the write of an
-// older report file open until a newer one's write completes, then
-// releases it. When room is needed, the older file is deleted first,
-// though its write was the last to complete.
-func TestFilesDeletedOldestFirstWhenWritesOverlap(t *testing.T) {
-	const maxBytes = 1000
-
-	report := map[string]string{"id": "overlap"}
-
-	t.Setenv("DATA_DIR", t.TempDir())
-	t.Setenv("DATA_DIR_MAX_BYTES", strconv.Itoa(maxBytes))
-
-	buf := startBuffer(t)
-
-	created := make(chan string)
-	release := make(chan struct{})
-
-	buf.OnFileCreated(func(f *os.File) {
-		created <- f.Name()
-
-		<-release
-	})
-
-	err := buf.Append(report)
-	if err != nil {
-		t.Fatalf("older report: %v", err)
-	}
-
-	flushed := make(chan error)
-
-	go func() { flushed <- buf.Flush() }()
-
-	older := <-created
-
-	// Only the older write is held; the newer one goes through, and
-	// so do the writes after it, the final one at stop included.
-	var newer string
-
-	buf.OnFileCreated(func(f *os.File) { newer = f.Name() })
-
-	// Errorf, not Fatalf, until the older write is released, so that a
-	// failure here does not leave it held.
-	err = buf.Append(report)
-	if err != nil {
-		t.Errorf("newer report: %v", err)
-	}
-
-	err = buf.Flush()
-	if err != nil {
-		t.Errorf("flush of the newer report: %v", err)
-	}
-
-	close(release)
-
-	err = <-flushed
-	if err != nil {
-		t.Fatalf("flush of the older report: %v", err)
-	}
-
-	info, err := os.Stat(newer)
-	if err != nil {
-		t.Fatalf("stat %s: %v", newer, err)
-	}
-
-	// A report that fits beside the newer file alone, so deleting the
-	// older one makes exactly the room it needs.
-	pad := maxBytes - int(info.Size()) - lineBytes(t, map[string]string{"id": ""})
-
-	err = buf.Append(map[string]string{"id": strings.Repeat("a", pad)})
-	if err != nil {
-		t.Fatalf("report that needs room: %v", err)
-	}
-
-	if exists(t, older) {
-		t.Fatal("older report file kept when room was needed")
-	}
-
-	if !exists(t, newer) {
-		t.Fatal("newer report file deleted before the older one")
-	}
-}

internal/server/routes_test.go

@@ -22,9 +22,9 @@ import (
 	"go.uber.org/fx/fxtest"
 )
 
-// newServer builds a Server from the same constructors as main,
+// newServer builds a Server from the same constructors as
-// configured from the environment. It is never started, so nothing
+// cli.RunServer, configured from the environment. It is never started,
-// listens.
+// so nothing listens.
 func newServer(t *testing.T) *server.Server {
 	t.Helper()
 

internal/server/server.go

@@ -79,7 +79,7 @@ func New(
 			s.SetupRoutes()
 			s.httpServer = s.newHTTPServer()
 
-			go s.listenAndServe()
+			go func() { s.listenAndServe() }()
 
 			return nil
 		},

package.json

@@ -1,6 +1,6 @@
 {
     "name": "netwatch",
-    "version": "1.0.0",
+    "private": true,
     "description": "Real-time network latency monitor SPA",
     "type": "module",
     "scripts": {

script/add-dependency

@@ -23,8 +23,8 @@ main() {
     "" | -* | *[[:space:]]*) usage ;;
     esac
     cd "$ROOT"
-    yarn add --dev "$1"
+    "$ROOT/script/yarn" add --dev "$1"
-    yarn install --frozen-lockfile
+    "$ROOT/script/yarn" install --frozen-lockfile
 }
 
 main "$@"

script/bootstrap

@@ -3,11 +3,14 @@
 # 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. Node is
-# used directly if it is at least NODE_MIN_VERSION; otherwise it is
+# used directly only if it has the major version of NODE_VERSION and is
-# installed at a pinned version via nvm (installing nvm itself first,
+# at least NODE_MIN_VERSION; otherwise NODE_VERSION is installed via nvm
-# from a hash-verified release archive, never curl | sh). Go, with its
+# (installing nvm itself first, from a hash-verified release archive,
-# gofmt, is used directly only if it is exactly GO_VERSION; otherwise
+# never curl | sh). Yarn is used directly only if it is exactly
-# that release is installed from its hash-verified archive.
+# YARN_VERSION; otherwise that version is prepared through corepack,
+# or installed with npm where there is no corepack.
+# Go, with its gofmt, is used directly only if it is exactly GO_VERSION;
+# otherwise that release is installed from its hash-verified archive.
 #
 # What this script installs outside the system package manager lives
 # under $HOME and is linked into ~/.local/bin, where make and the git
@@ -142,32 +145,63 @@ ensure_nvm() {
     rm -rf "$tmp"
 }
 
-# node_ok: the node on PATH is at least NODE_MIN_VERSION. node itself
+# node_ok: the node on PATH has the major version of NODE_VERSION and is
-# compares the two: major, then minor, then patch.
+# at least NODE_MIN_VERSION. Only the major is compared with the pin,
+# as REPO_POLICIES.md asks. node itself compares the versions: major,
+# then minor, then patch. A node that fails does not pass.
+# script/bootstrap and script/yarn carry this function unchanged.
 node_ok() {
-    if missing node; then return 1; fi
+    if ! command -v node >/dev/null 2>&1; then return 1; fi
     node -e '
         const have = process.versions.node.split(".").map(Number);
         const want = process.argv[1].split(".").map(Number);
+        const pinned = process.argv[2].split(".").map(Number);
+        if (have[0] !== pinned[0]) process.exit(1);
         for (let i = 0; i < 3; i++) {
             if (have[i] !== want[i]) process.exit(have[i] > want[i] ? 0 : 1);
         }
-    ' "$NODE_MIN_VERSION"
+    ' "$NODE_MIN_VERSION" "$NODE_VERSION"
 }
 
 # ensure_node: unless node_ok, install NODE_VERSION and link its node.
+# Afterwards node is looked up through PATH again and must pass node_ok.
 ensure_node() {
-    if node_ok; then return 0; fi
+    if node_ok; then
+        echo "bootstrap: using node $(node --version)"
+        return 0
+    fi
     ensure_nvm
     nvm_sh "nvm install $NODE_VERSION"
     link_bin "$HOME/.nvm/versions/node/v$NODE_VERSION/bin/node" node
+    hash -r
+    if ! node_ok; then
+        echo "bootstrap: after installing node $NODE_VERSION through nvm," >&2
+        echo "  the node on PATH, $(command -v node), is not it" >&2
+        exit 1
+    fi
+    echo "bootstrap: installed node $(node --version)"
 }
 
-# ensure_yarn: corepack writes its shims (pnpm and yarnpkg as well as
+# yarn_ok: yarn --version on PATH prints exactly YARN_VERSION, compared
-# yarn) into $TOOLCHAIN rather than next to itself, and the npm fallback
+# over the whole version. A yarn that fails or prints anything else does
-# installs there too; only yarn is linked.
+# not pass.
+yarn_ok() {
+    if missing yarn; then return 1; fi
+    version="$(yarn --version 2>/dev/null)" || return 1
+    [ "$version" = "$YARN_VERSION" ]
+}
+
+# ensure_yarn: unless yarn_ok, prepare YARN_VERSION and link its yarn.
+# corepack writes its shims (pnpm and yarnpkg as well as yarn) into
+# $TOOLCHAIN rather than next to itself, and the npm fallback installs
+# there too; only yarn is linked. Afterwards yarn is looked up through
+# PATH again and must pass yarn_ok, so another yarn that hides the link
+# stops bootstrap.
 ensure_yarn() {
-    if ! missing yarn; then return 0; fi
+    if yarn_ok; then
+        echo "bootstrap: using yarn $(yarn --version)"
+        return 0
+    fi
     shims="$TOOLCHAIN/corepack-shims"
     mkdir -p "$shims"
     if ! missing corepack; then
@@ -182,6 +216,13 @@ ensure_yarn() {
         shims="$TOOLCHAIN/npm-global/bin"
     fi
     link_bin "$shims/yarn" yarn
+    hash -r
+    if ! yarn_ok; then
+        echo "bootstrap: after installing yarn $YARN_VERSION in $shims," >&2
+        echo "  the yarn on PATH, $(command -v yarn), does not report it" >&2
+        exit 1
+    fi
+    echo "bootstrap: installed yarn $(yarn --version)"
 }
 
 # go_ok: the go on PATH has its gofmt beside it (a Go release ships the

script/build

@@ -8,11 +8,12 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    yarn build
+    "$ROOT/script/yarn" build
 
-    # VERSION comes from the environment. Unset or empty, it is git
+    # VERSION comes from the environment. Unset or empty, it is
-    # describe, or "dev" where there is no git or no repository history.
+    # `git describe --tags --always --dirty`, as in script/docker, or
-    version="${VERSION:-$(git describe --always --dirty 2>/dev/null || echo dev)}"
+    # "dev" where there is no git or no repository history.
+    version="${VERSION:-$(git describe --tags --always --dirty 2>/dev/null || echo dev)}"
 
     CGO_ENABLED=0 go build -trimpath \
         -ldflags "-s -w -X main.Version=$version" \

script/cibuild

@@ -2,9 +2,12 @@
 # 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
+# A new CACHEBUST value reruns every RUN step after the ARG CACHEBUST
-# final stage depends on are RUN steps, and a cached one is a check that
+# line, since a cached gate phase is a check that did not run; unlike
-# did not run.
+# --no-cache, it keeps the layers above that line and the cache mounts.
+# This relies on every stage of the Dockerfile declaring ARG CACHEBUST
+# before every step that lints, tests or compiles; a stage without that
+# line is served from cache on an unchanged tree, with no warning.
 set -eu
 
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -14,15 +17,17 @@ main() {
     cd "$ROOT"
     "$SCRIPT_DIR/bootstrap"
     "$SCRIPT_DIR/check"
-    # Own line: a failing command substitution inside an argument does
+    # The version and the tag each get their own line: a failing
-    # not trip `set -e`, so the inline form degrades silently to an
+    # command substitution inside an argument does not trip `set -e`,
-    # empty constant. The VERSION build argument takes precedence over
+    # so the inline form degrades silently to an empty constant. The
-    # the version a build stage derives from the .git in the context.
+    # 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 \
+    tag="$("$SCRIPT_DIR/projectname")"
+    docker build --build-arg CACHEBUST="$(date +%s)-$$" \
         --build-arg VERSION="$version" \
-        -t "$("$SCRIPT_DIR/projectname")" .
+        -t "$tag" .
 }
 
 main "$@"

script/dev

@@ -7,7 +7,7 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    yarn dev
+    "$ROOT/script/yarn" dev
 }
 
 main "$@"

script/docker

@@ -1,8 +1,12 @@
 #!/bin/sh
 # script/docker: build the Docker image tagged with the project name.
 # Identical in all repos; the tag comes from script/projectname.
-# --no-cache because the gate phases the final stage depends on are RUN
+# A new CACHEBUST value reruns every RUN step after the ARG CACHEBUST
-# steps, and a cached one is a check that did not run.
+# line, since a cached gate phase is a check that did not run; unlike
+# --no-cache, it keeps the layers above that line and the cache mounts.
+# This relies on every stage of the Dockerfile declaring ARG CACHEBUST
+# before every step that lints, tests or compiles; a stage without that
+# line is served from cache on an unchanged tree, with no warning.
 set -eu
 
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -10,15 +14,17 @@ ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    # Own line: a failing command substitution inside an argument does
+    # The version and the tag each get their own line: a failing
-    # not trip `set -e`, so the inline form degrades silently to an
+    # command substitution inside an argument does not trip `set -e`,
-    # empty constant. The VERSION build argument takes precedence over
+    # so the inline form degrades silently to an empty constant. The
-    # the version a build stage derives from the .git in the context.
+    # VERSION build argument takes precedence over the version a build
+    # stage derives from the .git in the context.
     version="$(git describe --tags --always --dirty 2>/dev/null || true)"
     [ -n "$version" ] || version="unknown"
-    docker build --no-cache \
+    tag="$("$SCRIPT_DIR/projectname")"
+    docker build --build-arg CACHEBUST="$(date +%s)-$$" \
         --build-arg VERSION="$version" \
-        -t "$("$SCRIPT_DIR/projectname")" .
+        -t "$tag" .
 }
 
 main "$@"

script/frontend-fmt

@@ -10,7 +10,7 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    yarn prettier --write '**/*.{js,css,html,md}'
+    "$ROOT/script/yarn" prettier --write '**/*.{js,css,html,md}'
 }
 
 main "$@"

script/frontend-fmt-check

@@ -7,7 +7,7 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    yarn prettier --check '**/*.{js,css,html,md}'
+    "$ROOT/script/yarn" prettier --check '**/*.{js,css,html,md}'
 }
 
 main "$@"

script/frontend-test

@@ -14,12 +14,14 @@ ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    NODE_OPTIONS=--test-reporter=dot timeout 90 yarn --silent run test || {
+    NODE_OPTIONS=--test-reporter=dot \
+        timeout 90 "$ROOT/script/yarn" --silent run test || {
         echo "--- Rerunning with every test listed for details ---"
-        NODE_OPTIONS=--test-reporter=spec timeout 90 yarn --silent run test
+        NODE_OPTIONS=--test-reporter=spec \
+            timeout 90 "$ROOT/script/yarn" --silent run test
         exit 1
     }
-    timeout 30 yarn build
+    timeout 30 "$ROOT/script/yarn" build
 }
 
 main "$@"

script/frontend-viewport-test

@@ -7,6 +7,8 @@
 # headless Chrome at it over CDP, and asserts on computed layout at every
 # viewport width derived from the app's own CSS. See
 # test/viewport/README.md for what this covers and what it cannot.
+# Before the browser starts, it checks that the page is served with
+# Cache-Control: no-cache and Content-Security-Policy.
 #
 # Deliberately not part of script/check: it needs Docker and takes far
 # longer than the 20s budget make test has to stay inside.
@@ -75,6 +77,32 @@ main() {
         -v "$ROOT/configs/security-headers.conf:/etc/nginx/security-headers.conf:ro" \
         "$SERVER_IMAGE" > /dev/null
 
+    # The page, and /no-such-page, which location / answers with the
+    # page through its try_files fallback, must go out with
+    # Cache-Control: no-cache. That add_header in its location drops the
+    # server's own unless the location includes the security headers
+    # again; only Content-Security-Policy is checked, as it comes from
+    # the same file as the rest. curl tries up to ten more times, a
+    # second apart, while nginx starts. It runs without -S, which would
+    # print an error for every failed attempt, even when a later one
+    # succeeds.
+    for path in / /no-such-page; do
+        if ! headers="$(docker exec "$SERVER" curl -s -o /dev/null -D - \
+            --retry 10 --retry-connrefused --retry-delay 1 \
+            "http://127.0.0.1:8080$path")"; then
+            echo "frontend-viewport-test: nginx did not answer $path" >&2
+            exit 1
+        fi
+        headers="$(printf '%s\n' "$headers" | tr -d '\r')"
+        if ! printf '%s\n' "$headers" | grep -qix 'Cache-Control: no-cache' ||
+            ! printf '%s\n' "$headers" | grep -qi '^Content-Security-Policy:'; then
+            echo "frontend-viewport-test: $path needs" \
+                "Cache-Control: no-cache and Content-Security-Policy; got:" >&2
+            printf '%s\n' "$headers" >&2
+            exit 1
+        fi
+    done
+
     # The image's own entrypoint already exposes CDP on 9222 and passes
     # --no-sandbox, so only extra flags belong here; re-specifying the
     # debugging port collides with it and leaves the endpoint bound to

script/lint

@@ -5,9 +5,14 @@
 # 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
+# --target names it. A new CACHEBUST value reruns every RUN step after
-# that did not run. The tag makes each build replace the previous image
+# the ARG CACHEBUST line, since a cached lint layer is a lint that did
-# instead of leaving a dangling one behind.
+# not run; unlike --no-cache, it keeps the layers above that line and
+# the cache mounts. This relies on every stage of the Dockerfile
+# declaring ARG CACHEBUST before every step that lints, tests or
+# compiles; a stage without that line is served from cache on an
+# unchanged tree, with no warning.
+# --output type=cacheonly writes no image, since nothing uses one.
 set -eu
 
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -15,9 +20,9 @@ ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    docker build --no-cache \
+    docker build --build-arg CACHEBUST="$(date +%s)-$$" \
         --target lint \
-        -t "$("$SCRIPT_DIR/projectname")-lint" .
+        --output type=cacheonly .
 }
 
 main "$@"

script/test

@@ -2,8 +2,13 @@
 # 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
+# named, and a new CACHEBUST value because a cached test layer is a test
-# run, and a tag so each build replaces the previous image.
+# that did not run; unlike --no-cache, it keeps the layers above the ARG
+# CACHEBUST line and the cache mounts. This relies on every stage of the
+# Dockerfile declaring ARG CACHEBUST before every step that lints, tests
+# or compiles; a stage without that line is served from cache on an
+# unchanged tree, with no warning. --output type=cacheonly writes no
+# image, since nothing uses one.
 set -eu
 
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -11,9 +16,9 @@ ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 
 main() {
     cd "$ROOT"
-    docker build --no-cache \
+    docker build --build-arg CACHEBUST="$(date +%s)-$$" \
         --target test \
-        -t "$("$SCRIPT_DIR/projectname")-test" .
+        --output type=cacheonly .
 }
 
 main "$@"

script/yarn

@@ -0,0 +1,55 @@
+#!/bin/sh
+# script/yarn: run yarn with the given arguments under the node
+# script/bootstrap would use. Every script that runs yarn on the host
+# runs it through this one, except script/bootstrap, which picks its
+# own node and yarn.
+set -eu
+
+ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
+
+# Must match the pins in script/bootstrap.
+NODE_VERSION="22.17.0"
+NODE_MIN_VERSION="22.13.0"
+
+# node_ok: the node on PATH has the major version of NODE_VERSION and is
+# at least NODE_MIN_VERSION. Only the major is compared with the pin,
+# as REPO_POLICIES.md asks. node itself compares the versions: major,
+# then minor, then patch. A node that fails does not pass.
+# script/bootstrap and script/yarn carry this function unchanged.
+node_ok() {
+    if ! command -v node >/dev/null 2>&1; then return 1; fi
+    node -e '
+        const have = process.versions.node.split(".").map(Number);
+        const want = process.argv[1].split(".").map(Number);
+        const pinned = process.argv[2].split(".").map(Number);
+        if (have[0] !== pinned[0]) process.exit(1);
+        for (let i = 0; i < 3; i++) {
+            if (have[i] !== want[i]) process.exit(have[i] > want[i] ? 0 : 1);
+        }
+    ' "$NODE_MIN_VERSION" "$NODE_VERSION"
+}
+
+# Run the yarn script/bootstrap installed: the yarn on PATH when the
+# node on PATH passes node_ok, and otherwise yarn under the pinned node
+# in nvm. nvm is a bash script, hence the subshell.
+run_yarn() {
+    if node_ok && command -v yarn >/dev/null 2>&1; then
+        exec yarn "$@"
+    fi
+    if node_ok || [ ! -s "$HOME/.nvm/nvm.sh" ]; then
+        echo "script/yarn: no yarn; run script/bootstrap first" >&2
+        exit 1
+    fi
+    exec bash -c '. "$HOME/.nvm/nvm.sh" && nvm use "$1" >/dev/null &&
+        shift && exec yarn "$@"' bash "$NODE_VERSION" "$@"
+}
+
+main() {
+    cd "$ROOT"
+    # script/bootstrap links the node and yarn it installs into
+    # ~/.local/bin, which the shell that called it may not have on PATH.
+    PATH="$HOME/.local/bin:$PATH"
+    run_yarn "$@"
+}
+
+main "$@"

test/unit/main.test.js

@@ -6,6 +6,7 @@ import { beforeEach, test } from "node:test";
 import assert from "node:assert/strict";
 import {
     AppState,
+    buildUI,
     CONFIG,
     greyOutUI,
     hostRowHTML,
@@ -17,6 +18,7 @@ import {
     measureLatency,
     Reporter,
     tick,
+    updateClocks,
     updateSummary,
 } from "../../web/src/main.js";
 
@@ -28,8 +30,10 @@ import {
 // selector until the next test starts. As on a page, writing its text
 // replaces its markup; the status dot greyOutUI looks for in it is not
 // there. Drawing a sparkline does nothing; it looks for the pixel ratio on
-// window and finds none. In each test, console.error and console.info
+// window and finds none. The commit the page's footer names, which
-// print nothing and keep what they are given.
+// vite.config.js fills in when it builds the page, is blank. In each test,
+// console.error and console.info print nothing and keep what they are
+// given.
 const doNothing = () => {};
 let elements;
 beforeEach((t) => {
@@ -51,6 +55,8 @@ const canvasContext = {
 globalThis.window = {};
 globalThis.requestAnimationFrame = doNothing;
 globalThis.cancelAnimationFrame = doNothing;
+globalThis.__COMMIT_HASH__ = "";
+globalThis.__COMMIT_FULL__ = "";
 globalThis.document = {
     getElementById: () => null,
     createElement: () => ({}),
@@ -77,6 +83,11 @@ function statusText(state, host) {
     return elements[`.status-text[data-host="${index}"]`]?.innerHTML;
 }
 
+// What this test wrote to the browser console with console[method].
+function consoleLines(method) {
+    return console[method].mock.calls.map((call) => call.arguments[0]);
+}
+
 // Mocks the clock for test t, so that a check lasting seconds takes no real
 // time, and replaces fetch with targets that each answer after
 // answerAfter(url) milliseconds of that clock, or never when that is
@@ -169,6 +180,91 @@ test("a check fetches the target's URL as written, with no query string added",
     );
 });
 
+test("a check sends no Referer header, wherever the page is served from", async (t) => {
+    mockTargets(t, () => 10);
+    const check = measureLatency("https://fsn1-speed.hetzner.com");
+    t.mock.timers.tick(10);
+    assert.notEqual(await settled(check), "still waiting");
+    assert.equal(
+        fetch.mock.calls[0].arguments[1].referrerPolicy,
+        "no-referrer",
+    );
+});
+
+// These targets' home pages answer with a redirect. A check follows it and
+// times the request to the new URL too, often on another host. The Google
+// Cloud Console redirects only requests that do not say they come from
+// another site, such as curl's. Each target is checked at /robots.txt on
+// its own host, which answers without a redirect. An S3 region answers it
+// with 403, and any answer counts as reachable.
+test("a target whose home page redirects is checked at /robots.txt on its own host", () => {
+    const checked = new Map(
+        new AppState([]).wan.map((host) => [host.name, host.url]),
+    );
+    for (const [name, url] of [
+        ["AWS Console", "https://console.aws.amazon.com/robots.txt"],
+        ["Google Cloud Console", "https://console.cloud.google.com/robots.txt"],
+        ["Akamai", "https://www.akamai.com/robots.txt"],
+        ["B2", "https://api.backblazeb2.com/robots.txt"],
+        [
+            "S3 af-south-1 (Cape Town)",
+            "https://s3.af-south-1.amazonaws.com/robots.txt",
+        ],
+        [
+            "S3 eu-west-2 (London)",
+            "https://s3.eu-west-2.amazonaws.com/robots.txt",
+        ],
+        [
+            "S3 me-south-1 (Bahrain)",
+            "https://s3.me-south-1.amazonaws.com/robots.txt",
+        ],
+        [
+            "S3 ap-northeast-1 (Tokyo)",
+            "https://s3.ap-northeast-1.amazonaws.com/robots.txt",
+        ],
+        [
+            "S3 ap-southeast-1 (Singapore)",
+            "https://s3.ap-southeast-1.amazonaws.com/robots.txt",
+        ],
+        [
+            "S3 ap-southeast-2 (Sydney)",
+            "https://s3.ap-southeast-2.amazonaws.com/robots.txt",
+        ],
+        [
+            "S3 us-west-2 (Oregon)",
+            "https://s3.us-west-2.amazonaws.com/robots.txt",
+        ],
+        [
+            "S3 sa-east-1 (São Paulo)",
+            "https://s3.sa-east-1.amazonaws.com/robots.txt",
+        ],
+    ]) {
+        assert.equal(checked.get(name), url);
+    }
+});
+
+// Chrome keeps the connection to an S3 region open after a HEAD check, but
+// closes it after most GET checks. The Hetzner servers do not answer HEAD.
+test("a round checks each S3 region with a HEAD request and every other target with a GET request", async (t) => {
+    const state = new AppState([
+        { name: "Local CPE", url: "http://192.168.100.1" },
+    ]);
+    mockTargets(t, () => 10);
+    const round = tick(state, new AbortController().signal);
+    t.mock.timers.tick(10);
+    assert.notEqual(await settled(round), "still waiting");
+    const methods = new Map(
+        fetch.mock.calls.map((call) => [
+            call.arguments[0],
+            call.arguments[1].method,
+        ]),
+    );
+    for (const host of state.allHosts) {
+        const method = host.name.startsWith("S3 ") ? "HEAD" : "GET";
+        assert.equal(methods.get(host.url), method, host.name);
+    }
+});
+
 test("at a 30000ms interval, a target answering after 1000ms shows in its row while another target's check is still waiting", async (t) => {
     CONFIG.updateInterval = 30000;
     const state = new AppState([
@@ -196,9 +292,10 @@ test("at a 30000ms interval, a target answering after 1000ms shows in its row wh
 });
 
 // In the next three tests, the answering target's check is still waiting
-// when something happens after which its result must not show.
+// when something happens after which its result must not be recorded or
+// show in its row, and the check must write nothing to the console.
 
-test("at a 30000ms interval, a check still waiting when its round is given up does not show in its row", async (t) => {
+test("at a 30000ms interval, a check still waiting when its round is given up is not recorded and does not show in its row", async (t) => {
     CONFIG.updateInterval = 30000;
     const state = new AppState([
         { name: "Answering", url: "https://answering.test" },
@@ -214,9 +311,13 @@ test("at a 30000ms interval, a check still waiting when its round is given up do
     roundChecks.abort();
     assert.notEqual(await settled(round), "still waiting");
     assert.equal(latencyFigure(state, answering), undefined);
+    assert.deepEqual(answering.history, []);
+    assert.equal(answering.status, "pending");
+    assert.deepEqual(consoleLines("error"), []);
+    assert.deepEqual(consoleLines("info"), []);
 });
 
-test("at a 30000ms interval, a check still waiting when the user pauses does not show in its row", async (t) => {
+test("at a 30000ms interval, a check still waiting when the user pauses is not recorded and does not show in its row", async (t) => {
     CONFIG.updateInterval = 30000;
     const state = new AppState([
         { name: "Answering", url: "https://answering.test" },
@@ -231,9 +332,13 @@ test("at a 30000ms interval, a check still waiting when the user pauses does not
     t.mock.timers.tick(500);
     assert.equal(await settled(round), "still waiting");
     assert.equal(latencyFigure(state, answering), undefined);
+    assert.deepEqual(answering.history, []);
+    assert.equal(answering.status, "pending");
+    assert.deepEqual(consoleLines("error"), []);
+    assert.deepEqual(consoleLines("info"), []);
 });
 
-test("at a 30000ms interval, a check in the first round does not show in its row", async (t) => {
+test("at a 30000ms interval, a check in the first round is not recorded and does not show in its row", async (t) => {
     CONFIG.updateInterval = 30000;
     const state = new AppState([
         { name: "Answering", url: "https://answering.test" },
@@ -245,6 +350,10 @@ test("at a 30000ms interval, a check in the first round does not show in its row
     t.mock.timers.tick(1000);
     assert.equal(await settled(round), "still waiting");
     assert.equal(latencyFigure(state, answering), undefined);
+    assert.deepEqual(answering.history, []);
+    assert.equal(answering.status, "pending");
+    assert.deepEqual(consoleLines("error"), []);
+    assert.deepEqual(consoleLines("info"), []);
 });
 
 test("at a 30000ms interval, after the user pauses and resumes during a round, no row reads paused once its last check ends", async (t) => {
@@ -294,11 +403,6 @@ function stateWithOneTarget() {
     return { state, target: state.local[0] };
 }
 
-// What this test wrote to the browser console with console[method].
-function consoleLines(method) {
-    return console[method].mock.calls.map((call) => call.arguments[0]);
-}
-
 test("a check that fails with a network error writes one console line with the target, the time, the error and how long the request took", async (t) => {
     const { state, target } = stateWithOneTarget();
     mockTargets(t, () => 23, target.url);
@@ -579,6 +683,32 @@ test("a check that timed out or found its target unreachable has a grey figure a
     assert.equal(latencyHex(null), "#6b7280");
 });
 
+// buildUI draws the page into #app, here a plain object. Each entry in the
+// footer's color legend is a colored dot and the latencies it stands for,
+// with < and > written as &lt; and &gt;. A latency takes the color of the
+// first entry in CONFIG.latencyColors whose limit it is below, so the
+// legend has, for each of those entries in order, a dot of its color (a
+// bg- class where CONFIG.latencyColors has a text- one) and the latencies
+// below its limit, or, for the last, which has no limit, the latencies
+// from the limit before it up. A gray dot for offline comes after them.
+test("the footer's color legend has each entry in CONFIG.latencyColors, with its color and limit, then offline", (t) => {
+    const app = {};
+    t.mock.method(globalThis.document, "getElementById", () => app);
+    buildUI(new AppState([]));
+
+    const legend = [
+        ...app.innerHTML.matchAll(/(bg-\w+-500) [^"]*"><\/span>(\S+)/g),
+    ].map(([, dot, latencies]) => [dot, latencies]);
+    const colors = CONFIG.latencyColors;
+    const expected = colors.map((color, i) => [
+        color.className.replace("text-", "bg-"),
+        color.below === Infinity
+            ? `&gt;=${colors[i - 1].below}ms`
+            : `&lt;${color.below}ms`,
+    ]);
+    assert.deepEqual(legend, [...expected, ["bg-gray-500", "offline"]]);
+});
+
 // A target whose checks, in turn, answered after each of latencies ms, or,
 // for null, found it unreachable.
 function hostAfter(latencies) {
@@ -701,6 +831,56 @@ test("once every WAN target's check has failed, the summary's Reachable slot rea
     assert.equal(elements["#summary-reachable"].className, "text-red-400");
 });
 
+// In the next two tests each of the header's clocks is a plain object, made
+// the first time updateClocks looks it up and kept in elements under "#" and
+// its id.
+
+// The mocked clock reads 1970-01-01T00:00:00.000Z.
+test("the header's clocks show the local time with its time zone's name, and the UTC time", (t) => {
+    t.mock.method(
+        globalThis.document,
+        "getElementById",
+        (id) => (elements[`#${id}`] ??= {}),
+    );
+    t.mock.timers.enable({ apis: ["Date"] });
+
+    updateClocks();
+
+    assert.match(
+        elements["#clock-local"].textContent,
+        /^Local: \d{4}-\d\d-\d\d \d\d:\d\d:\d\d \S/,
+    );
+    assert.equal(
+        elements["#clock-utc"].textContent,
+        "UTC: 1970-01-01 00:00:00 UTC",
+    );
+});
+
+// Here Intl.DateTimeFormat is as in Chrome before 95, Firefox before 91 or
+// Safari before 15.4: it throws a RangeError when asked for the time zone's
+// "shortGeneric" name.
+test("in a browser that cannot name the time zone, updating the clocks throws nothing and leaves them blank", (t) => {
+    t.mock.method(
+        globalThis.document,
+        "getElementById",
+        (id) => (elements[`#${id}`] ??= {}),
+    );
+    const DateTimeFormat = Intl.DateTimeFormat;
+    t.mock.method(Intl, "DateTimeFormat", function (locales, options) {
+        if (options?.timeZoneName === "shortGeneric") {
+            throw new RangeError(
+                "Value shortGeneric out of range for Intl.DateTimeFormat options property timeZoneName",
+            );
+        }
+        return new DateTimeFormat(locales, options);
+    });
+
+    updateClocks();
+
+    assert.equal(elements["#clock-local"].textContent, "");
+    assert.equal(elements["#clock-utc"].textContent, "");
+});
+
 // An app state in which, of the WAN targets, the first timedOut timed out,
 // the next unreachable were found unreachable, the next answered answered
 // after latency ms, and the rest have not been checked yet.

test/viewport/README.md

@@ -1,114 +1,146 @@
-# Responsive-layout harness
+# The viewport test
 
-Automated verification of the responsive layout that landed in #5. Run it with:
+`make frontend-viewport-test` checks the page's layout at several window widths
+in a real browser:
 
 ```bash
 make frontend-viewport-test
 ```
 
-It builds `dist/`, serves it from the same digest-pinned `nginx` image and the
+First it runs the unit tests. Then it builds the page into `dist/`. It serves
-same `configs/nginx.conf` the shipping container uses, drives a digest-pinned
+`dist/` with the nginx image and the `configs/nginx.conf` that the Docker image
-headless Chrome against it over CDP, and asserts on computed layout at every
+ships. Before the browser starts, it fetches `/` and `/no-such-page`, which
-viewport width derived from the app's own CSS. Screenshots land in
+nginx answers with the page. Both must carry `Cache-Control: no-cache` and a
-`tmp/viewport/` alongside a `results.json`; they are artifacts for a human to
+`Content-Security-Policy` header. Then it starts a headless Chrome in a
-look at when something fails, not the evidence. The assertions are the evidence.
+container and opens the page in it at each width. For each width it measures the
+layout inside the page and checks the measurements in Node. The checks decide
+pass or fail. Screenshots and a `results.json` land in `tmp/viewport/`, for a
+person to look at when a check fails.
 
-The target is deliberately outside `make check`: it takes minutes, and
+The test is not part of `make check`: it takes minutes, and `make test` has to
-`make test` has to stay under 60 seconds.
+finish in under 60 seconds.
 
-## How the widths are chosen
+## The widths
 
-Not from a list of phone models. `viewports.js` parses the `@media` conditions
+The breakpoints are read from the page's own files. `viewports.js` reads the
-out of `web/src/styles.css` and scans `web/src/main.js` and `web/index.html` for
+`min-width` and `max-width` conditions out of the `@media` blocks in
-Tailwind responsive prefixes, then tests every breakpoint it finds at one pixel
+`web/src/styles.css`. It also looks in `web/src/main.js` and `web/index.html`
-below it, exactly on it, and one pixel above it. A generic 375px "phone" test
+for Tailwind's responsive class prefixes, such as `md:`, and takes their
-sails straight past an off-by-one at a media query boundary; `max-width: 768px`
+breakpoints. Each breakpoint is tested at one pixel below it, exactly on it, and
-matches _at_ 768, and the sweep pins down which side of that line each layout is
+one pixel above it. Today the CSS has one breakpoint, `max-width: 768px`, and no
-on.
+responsive prefix is used.
 
-Nothing hardcodes 768. Add a second media block or start using `md:` classes and
+A `max-width: 768px` block matches at 768 pixels. Testing both sides of each
-the new breakpoint is covered without this directory being touched. The app is
+breakpoint catches a layout that switches one pixel too early or too late. A
-desktop-first today (all narrow rules live in `max-width` blocks); a
+second media block, or a first `md:` class, is covered without a change to this
-`min-width`-only, mobile-first set is handled as its inverse, and a set that
+directory.
-mixes the two makes the run fail loudly rather than test the right widths with
-the wrong expectation. Four further viewports are fixed anchors, each with a
-stated reason: a 320px floor, a 1280px desktop baseline, and two phone-landscape
-sizes straddling the breakpoint for the rotation case.
 
-## What it asserts
+The page is desktop-first: the wide layout is the default, and every narrow rule
+is in a `max-width` block. So a width is expected to show the narrow layout when
+one of those blocks matches it. A page with only `min-width` blocks would be
+read the other way round. A page that mixes the two kinds fails the run with a
+message, because the right expectation for each width can no longer be read from
+the breakpoints alone.
 
-- **app-rendered** — enough host rows exist and enough of them show a numeric
+Four more widths are fixed, each for a stated reason:
-  latency. This one exists so the rest cannot pass vacuously against a blank
-  page.
-- **no-horizontal-overflow** — `documentElement.scrollWidth` fits the layout
-  viewport, with the widest offending element named.
-- **nothing-past-viewport-edge** — no visible element's box extends past the
-  viewport edge.
-- **no-clipped-text** — nothing hides text behind `overflow: hidden`. Deliberate
-  ellipsis truncation (Tailwind's `truncate`, used on host names and URLs) is
-  excluded: it is a design choice, not breakage.
-- **tap-targets-44px** — every interactive control is at least 44x44 CSS px on
-  touch viewports, _and_ each selector in the control list matched at least the
-  number of visible elements it declares: one of each single control, and one
-  pin button per WAN host row. The second half is what stops the check passing
-  vacuously: with size alone, a renamed class would take its controls out of the
-  measured set and the check would report "all 0 controls are at least 44x44"
-  and pass. See below.
-- **host-rows-stacked / host-rows-side-by-side** — the rows genuinely reflow.
-  Computed `flex-direction` _and_ the actual geometry are checked, and in the
-  narrow layout the info block and the sparkline must each occupy essentially
-  the full row width. A row that merely shrank its 420px column would fail.
-- **probing-still-runs / gateway-detection-still-runs** — narrow viewports keep
-  probing and keep detecting the gateway. The mobile early-return path proposed
-  in #8 was rejected; this is what would catch it coming back.
 
-### The tap-target threshold
+- 320 by 568 pixels: the narrowest width still in common use
+- 667 by 375 pixels: a phone in landscape, inside the narrow layout
+- 844 by 390 pixels: a large phone in landscape, in the wide layout but still a
+  touch device
+- 1280 by 800 pixels: a desktop window
 
-44x44 CSS px. That is the figure in Apple's Human Interface Guidelines and in
+Every width but 1280 by 800 is tested as a touch device.
-WCAG 2.2 SC 2.5.5 "Target Size (Enhanced)". WCAG 2.2 SC 2.5.8 (level AA) sets a
-lower 24x24 floor, but that floor comes with a spacing exception these controls
-do not qualify for — the pin buttons sit directly against the host name they
-belong to.
 
-## Determinism
+## The checks
 
-The browser container runs on an `--internal` docker network and has no route to
+A failed check does not stop the others.
-the internet, so the app's latency probes cannot reach anything real. The
-harness answers them itself from a fixed delay table, with a deterministic
-fraction failed outright, so the rows render a realistic spread of one-, two-
-and three-digit latencies plus some unreachable rows. That spread is what the
-layout has to survive; a `---` placeholder in every row would not exercise it.
 
-## What this cannot verify
+- `app-rendered`: at least 10 WAN host rows exist, and at least 5 rows show a
+  number as their latency. Without this, the other checks could pass against a
+  blank page.
+- `no-horizontal-overflow`: the document is no wider than the window. On a
+  failure, the widest content is named.
+- `nothing-past-viewport-edge`: no visible element extends past the window's
+  edge.
+- `no-clipped-text`: no element hides its text behind `overflow: hidden` or
+  `overflow: clip`. Text cut with an ellipsis is allowed: that cut is a design
+  choice, not a defect. Tailwind's `truncate` cuts host names and URLs that way.
+- `tap-targets-44px`, on touch widths: each measured control is at least 44 by
+  44 CSS pixels. The controls are `#pause-btn`, `#interval-select`,
+  `#debug-toggle` and the `.pin-btn` in each WAN host row. The check also
+  demands that each selector matches enough visible elements: one of each single
+  control, and one pin button per WAN host row. A renamed class would otherwise
+  take its controls out of the measured set, and the check would pass on
+  nothing.
+- `host-rows-stacked` or `host-rows-side-by-side`: each host row shows the
+  layout expected at that width. In the narrow layout, the row's
+  `flex-direction` is `column`, the graph sits below the host's details, and
+  each takes at least 90% of the row's width. In the wide layout, the direction
+  is `row` and the graph sits to the right of the details. A row that only
+  shrank its fixed 420-pixel column fails.
+- `probing-still-runs`: the page sent at least one check to a target at this
+  width. The narrow layout must not skip the checks.
+- `gateway-detection-still-runs`: the `Local Gateway` row is present at this
+  width, so the gateway was looked for and found. The narrow layout must not
+  skip that search.
 
-Real limits, so nobody re-parks this issue as needing hardware:
+### The tap-target size
 
-- **Non-Chromium engines.** This is Chrome. iOS Safari is WebKit and cannot be
+44 by 44 CSS pixels is the figure in Apple's Human Interface Guidelines and in
-  emulated by it; Safari-specific bugs (viewport units under a collapsing URL
+WCAG 2.2 success criterion 2.5.5, "Target Size (Enhanced)". WCAG 2.2 criterion
-  bar, `-webkit-fill-available`, form control metrics) will not show up here.
+2.5.8 sets a lower minimum of 24 by 24 pixels, with an exception that depends on
-- **Real touch input.** `hasTouch` emulation changes what the page is told, not
+the space around a control. The pin buttons sit against the host name they
-  how a finger behaves. Gesture handling, scroll momentum, double-tap zoom and
+belong to, so the test keeps 44.
-  hover-state fallbacks on touch are out of scope.
-- **Physical pixel density and rendering.** `deviceScaleFactor` is set, but
-  subpixel antialiasing, OLED colour rendering and actual legibility at a given
-  physical size are not measurable here.
-- **Fonts.** The container has DejaVu, not the platform's own UI monospace. Text
-  metrics are therefore close to, but not identical to, a real device — a layout
-  that fits here by a few pixels might not there.
-- **On-device performance.** Canvas sparkline redraw cost, battery, and
-  behaviour on a slow radio are not measured.
-- **Browser chrome.** The address bar, safe-area insets and notch cutouts are
-  not simulated.
 
-Everything else this issue was actually about — does the layout reflow, does
+## The browser's network
-anything overflow, is content clipped, are the controls big enough — is a
+
-function of viewport width and CSS, and is covered above.
+The browser container runs on an internal Docker network with no route to the
+internet, so the page's checks reach no real host. The test answers them itself
+from a fixed table of delays: 2, 45, 123, 456 and 780 milliseconds. Which delay
+a target gets depends on its URL, and about one target in seven fails outright.
+So the rows show one-, two- and three-digit latencies and some unreachable rows,
+and the layout is tested against that spread. The request to
+`http://192.168.1.1` is answered after 5 milliseconds, so the `Local Gateway`
+row appears.
+
+At each width, the test sets the interval to 1 second and waits until at least 5
+rows show a latency before it measures. The page discards its first round, so
+this reaches a filled page without waiting out two rounds at 3 seconds.
+
+## What the test cannot check
+
+- Other browser engines. The test runs Chrome. Safari uses WebKit, and bugs that
+  only WebKit has do not show up here.
+- Real touch input. Touch emulation changes what the page is told, not how a
+  finger behaves. Gestures, scroll momentum, double-tap zoom and hover states on
+  touch are not tested.
+- Physical pixels. The device pixel ratio is set, but antialiasing, colour
+  rendering and legibility at a given physical size are not measured.
+- Fonts. The container has DejaVu, not a phone's own monospace font. Text is
+  about as wide as on a real device, but not exactly. A layout that fits here by
+  a few pixels might not fit there.
+- Performance. The cost of redrawing the graphs, battery use and behaviour on a
+  slow connection are not measured.
+- The browser's own frame. The address bar, safe-area insets and notch cutouts
+  are not simulated.
+
+## The files
+
+- `viewports.js` derives the widths from the sources.
+- `facts.js` is run inside the page and measures the layout. It only measures;
+  it decides nothing.
+- `checks.js` turns the measurements into the pass or fail of each check, with
+  the numbers that decided it.
+- `harness.js` connects to the browser, runs each width, writes the screenshots
+  and `results.json`, and prints the results. `script/frontend-viewport-test`
+  starts it; every input it needs arrives in the environment.
 
 ## Relation to the unit tests
 
-Complementary layers, not two stacks. The unit tests in `test/unit/`, which
+The unit tests in `test/unit/` call the functions that `web/src/main.js`
-`make test` runs with Node's built-in test runner, exercise the functions
+exports, with no browser. `make test` runs them with Node's built-in test
-`web/src/main.js` exports in-process with no browser. This harness exercises
+runner. This test runs the built page in a real browser engine and is the only
-rendered layout in a real engine and is the only thing here that can see a media
+test here that can see a media query. Neither replaces the other. Checks on
-query. Neither replaces the other; assertions about computed styles and element
+computed styles and element geometry belong here; checks on functions belong in
-geometry belong here, assertions about functions belong in `test/unit/`.
+`test/unit/`.

web/src/main.js

@@ -67,51 +67,66 @@ export const CONFIG = {
 
 // WAN endpoints to monitor. These are used for the aggregate health/stats
 // display (min/max/avg). Ordered: personal, LLM APIs, big-3 cloud, then
-// CDN/hosting/other.
+// CDN/hosting/other. A check follows a redirect and times the request to
+// the new URL too. So a target whose home page redirects is checked at
+// /robots.txt on its own host. That URL answers without a redirect.
 const WAN_HOSTS = [
     { name: "datavi.be", url: "https://datavi.be" },
     { name: "Anthropic API", url: "https://api.anthropic.com" },
     { name: "OpenAI API", url: "https://api.openai.com" },
-    { name: "AWS Console", url: "https://console.aws.amazon.com" },
+    { name: "AWS Console", url: "https://console.aws.amazon.com/robots.txt" },
-    { name: "Google Cloud Console", url: "https://console.cloud.google.com" },
+    {
+        name: "Google Cloud Console",
+        url: "https://console.cloud.google.com/robots.txt",
+    },
     { name: "Microsoft Azure", url: "https://portal.azure.com" },
     { name: "Cloudflare", url: "https://www.cloudflare.com" },
     { name: "Fastly CDN", url: "https://www.fastly.com" },
-    { name: "Akamai", url: "https://www.akamai.com" },
+    { name: "Akamai", url: "https://www.akamai.com/robots.txt" },
     { name: "Google", url: "https://www.google.com" },
     { name: "GitHub", url: "https://github.com" },
-    { name: "B2", url: "https://api.backblazeb2.com" },
+    { name: "B2", url: "https://api.backblazeb2.com/robots.txt" },
+    // Chrome keeps the connection to an S3 region open after a HEAD check,
+    // but closes it after most GET checks.
     {
         name: "S3 af-south-1 (Cape Town)",
-        url: "https://s3.af-south-1.amazonaws.com",
+        url: "https://s3.af-south-1.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     {
         name: "S3 eu-west-2 (London)",
-        url: "https://s3.eu-west-2.amazonaws.com",
+        url: "https://s3.eu-west-2.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     {
         name: "S3 me-south-1 (Bahrain)",
-        url: "https://s3.me-south-1.amazonaws.com",
+        url: "https://s3.me-south-1.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     {
         name: "S3 ap-northeast-1 (Tokyo)",
-        url: "https://s3.ap-northeast-1.amazonaws.com",
+        url: "https://s3.ap-northeast-1.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     {
         name: "S3 ap-southeast-1 (Singapore)",
-        url: "https://s3.ap-southeast-1.amazonaws.com",
+        url: "https://s3.ap-southeast-1.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     {
         name: "S3 ap-southeast-2 (Sydney)",
-        url: "https://s3.ap-southeast-2.amazonaws.com",
+        url: "https://s3.ap-southeast-2.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     {
         name: "S3 us-west-2 (Oregon)",
-        url: "https://s3.us-west-2.amazonaws.com",
+        url: "https://s3.us-west-2.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     {
         name: "S3 sa-east-1 (São Paulo)",
-        url: "https://s3.sa-east-1.amazonaws.com",
+        url: "https://s3.sa-east-1.amazonaws.com/robots.txt",
+        method: "HEAD",
     },
     // Hetzner regional speed test servers — genuine per-DC endpoints
     {
@@ -271,6 +286,7 @@ export class HostState {
     constructor(host, pinned = false) {
         this.name = host.name;
         this.url = host.url;
+        this.method = host.method;
         // Each entry is either a check's result, { timestamp, latency,
         // error }, or a round skipped while paused, { timestamp,
         // latency: null, paused: true }.
@@ -567,16 +583,18 @@ export class Reporter {
 
 // --- Latency Measurement -----------------------------------------------------
 
-// Checks one target. The check times out after CONFIG.requestTimeout; the
+// Checks one target with the given HTTP method, GET when none is given. The
-// caller can give it up sooner through the optional signal, which also ends
+// check times out after CONFIG.requestTimeout; the caller can give it up
-// it as a timeout. A failed check's reason says what went wrong and after
+// sooner through the optional signal, which also ends it as a timeout. A
-// how long; a check that answered has none.
+// failed check's reason says what went wrong and after how long; a check
+// that answered has none.
 //
-// The URL is fetched as written, with nothing added to it: the Hetzner
+// The Hetzner speed-test servers close the connection without an answer
-// speed-test servers close the connection without an answer when the URL
+// when the URL has a query string or the request has a Referer header. So
-// has a query string, and cache: "no-store" keeps the browser's cache out
+// the URL is fetched as written, and referrerPolicy: "no-referrer" sends
-// of the measurement.
+// no Referer header, wherever the page is served from. cache: "no-store"
-export async function measureLatency(url, signal) {
+// keeps the browser's cache out of the measurement.
+export async function measureLatency(url, signal, method = "GET") {
     const controller = new AbortController();
     const timeoutId = setTimeout(
         () => controller.abort(),
@@ -588,9 +606,10 @@ export async function measureLatency(url, signal) {
 
     try {
         await fetch(url, {
-            method: "GET",
+            method,
             mode: "no-cors",
             cache: "no-store",
+            referrerPolicy: "no-referrer",
             signal: controller.signal,
         });
         const latency = Math.round(performance.now() - start);
@@ -829,7 +848,7 @@ function wanRowsHTML(wanHosts) {
     return rows.join("");
 }
 
-function buildUI(state) {
+export function buildUI(state) {
     const app = document.getElementById("app");
     app.innerHTML = `
     <div class="mx-auto px-[5%] py-8">
@@ -911,7 +930,7 @@ function buildUI(state) {
           <span class="inline-block w-3 h-3 rounded-full bg-lime-500 mr-1 ml-3 align-middle"></span>&lt;100ms
           <span class="inline-block w-3 h-3 rounded-full bg-yellow-500 mr-1 ml-3 align-middle"></span>&lt;200ms
           <span class="inline-block w-3 h-3 rounded-full bg-orange-500 mr-1 ml-3 align-middle"></span>&lt;500ms
-          <span class="inline-block w-3 h-3 rounded-full bg-red-500 mr-1 ml-3 align-middle"></span>&gt;500ms
+          <span class="inline-block w-3 h-3 rounded-full bg-red-500 mr-1 ml-3 align-middle"></span>&gt;=500ms
           <span class="inline-block w-3 h-3 rounded-full bg-gray-500 mr-1 ml-3 align-middle"></span>offline
         </p>
         <p class="mt-2"><a href="https://git.eeqj.de/sneak/netwatch/commit/${__COMMIT_FULL__}" target="_blank" rel="noopener" class="text-gray-600 hover:text-gray-400">${__COMMIT_HASH__}</a></p>
@@ -1114,6 +1133,33 @@ function updateHealthBox(state) {
     else if (health === "healthy") log.notice(`Health: ${labels[health]}`);
 }
 
+// Writes the local time, with its time zone's name, and the UTC time into
+// the header's clocks. The clocks only decorate the page: in a browser
+// whose Intl.DateTimeFormat rejects the "shortGeneric" time zone name
+// (Chrome before 95, Firefox before 91, Safari before 15.4) they are left
+// blank.
+export function updateClocks() {
+    let local = "";
+    let utc = "";
+    try {
+        const now = new Date();
+        const zone = new Intl.DateTimeFormat("en-US", {
+            timeZoneName: "shortGeneric",
+        })
+            .formatToParts(now)
+            .find((p) => p.type === "timeZoneName").value;
+        const pad = (n) => String(n).padStart(2, "0");
+        local = `Local: ${now.toLocaleString("sv-SE", { hour12: false })} ${zone}`;
+        utc =
+            `UTC: ${now.getUTCFullYear()}-${pad(now.getUTCMonth() + 1)}-${pad(now.getUTCDate())} ` +
+            `${pad(now.getUTCHours())}:${pad(now.getUTCMinutes())}:${pad(now.getUTCSeconds())} UTC`;
+    } catch {
+        // Both clocks are left blank.
+    }
+    document.getElementById("clock-local").textContent = local;
+    document.getElementById("clock-utc").textContent = utc;
+}
+
 // --- Debug Log Renderer ------------------------------------------------------
 
 // One line of the debug log panel, its message set as text.
@@ -1263,7 +1309,7 @@ export async function tick(state, signal, onOffline) {
     // the round started.
     await Promise.all(
         state.allHosts.map(async (host) => {
-            const r = await measureLatency(host.url, signal);
+            const r = await measureLatency(host.url, signal, host.method);
             if (state.paused || signal.aborted || state.tickCount === 0) {
                 return;
             }
@@ -1336,7 +1382,7 @@ function startRecoveryProbe(state, startRounds) {
         const checks = new AbortController();
         state._recoveryProbeChecks = checks;
         for (const host of canaries) {
-            measureLatency(host.url, checks.signal).then((r) => {
+            measureLatency(host.url, checks.signal, host.method).then((r) => {
                 if (r.error !== null || checks.signal.aborted) return;
                 log.notice("Recovery probe: connectivity detected");
                 stopRecoveryProbe(state);
@@ -1509,24 +1555,6 @@ async function init() {
         sortAndRebuildWAN(state);
     });
 
-    function updateClocks() {
-        const now = new Date();
-        const pad = (n) => String(n).padStart(2, "0");
-        const utc =
-            `${now.getUTCFullYear()}-${pad(now.getUTCMonth() + 1)}-${pad(now.getUTCDate())} ` +
-            `${pad(now.getUTCHours())}:${pad(now.getUTCMinutes())}:${pad(now.getUTCSeconds())} UTC`;
-        const local =
-            now.toLocaleString("sv-SE", { hour12: false }) +
-            " " +
-            new Intl.DateTimeFormat("en-US", { timeZoneName: "shortGeneric" })
-                .formatToParts(now)
-                .find((p) => p.type === "timeZoneName").value;
-        document.getElementById("clock-local").textContent = "Local: " + local;
-        document.getElementById("clock-utc").textContent = "UTC: " + utc;
-    }
-    updateClocks();
-    setInterval(updateClocks, 1000);
-
     // Rounds never overlap: a round first gives up the last round's checks
     // if they are still waiting, and the last round then records nothing
     // more. At a steady interval they never are, as they time out at 80% of
@@ -1606,6 +1634,11 @@ async function init() {
 
     window.addEventListener("resize", () => handleResize(state));
     setTimeout(() => handleResize(state), CONFIG.resizeDelay);
+
+    // The clocks start last, after the rounds: they only decorate the page,
+    // so nothing in them may stop the monitor from probing.
+    updateClocks();
+    setInterval(updateClocks, 1000);
 }
 
 // Bootstrap only when loaded as the page: a real DOM containing the #app