The deploy.yml workflow_dispatch input always defaulted to 'v0.1.0',
requiring manual edit every time. Now the version defaults to 'auto',
which fetches all tags, finds the latest v* tag via semver sort, bumps
the patch component, and uses that as the deploy tag.
Changes:
- deploy.yml input: default changed to 'auto', required → false,
description updated to explain both auto and manual modes
- Added 'Resolve version' step: fetches tags, bumps latest semver
tag by patch, validates output format, exports to $VERSION
- 'Tag version' step: substituted ${{ github.event.inputs.version }}
→ ${{ env.VERSION }} to use the resolved/computed version
- 'Print deploy status' step: same substitution
- Semver validation guard rejects malformed tags (auto and manual)
Enable pageview tracking when VITE_UMAMI_WEBSITE_ID is set at frontend
build time (Forgejo secret + deploy workflow), with SPA route updates
and no script in local dev. Document setup in docs/umami-analytics.md,
extend integritetspolicy, and add admin Webbstatistik link in prod builds.
Co-authored-by: Cursor <cursoragent@cursor.com>
Replace the header "Byt lösenord" link with an Inställningar menu for
changing email or password. Email changes are two-step: request with
password, confirmation link to the new address, then password again on
confirm so a wrong inbox cannot take over the account.
- Backend: EmailChangeService, V10 email_change_tokens, confirm API
- Frontend: ChangeEmailPage, ConfirmEmailChangePage, header dropdown
- E2E: account-settings round-trips, Mailpit verification, wrong-password guard
- Flyway: V9 restore for dev DBs, CI migration checks, V10 for email tokens
Co-authored-by: Cursor <cursoragent@cursor.com>
Forgejo workflow_dispatch requires an explicit input type; without it the
UI showed invalidinputtype. Clarify README: workflow ref vs version tag.
Co-authored-by: Cursor <cursoragent@cursor.com>
Deploy workflow now writes MAIL_* and APP_PUBLIC_BASE_URL from Actions
secrets into the server .env so Resend SMTP works after domain verify.
Document Resend-only setup, Forgejo secret names, and prod expose-token off.
Co-authored-by: Cursor <cursoragent@cursor.com>
Production deploy failed with no backend logs before rollback. Print
backend and postgres logs on failure, wait longer for JVM startup, and
probe /api/payment/swish-info instead of vehicle lookup (no external scrape).
- Document proof-first troubleshooting in README
- No volume reset workflow; fix only after reading job logs
Docker Compose interpolates $VAR in .env files. Passwords like ...$A72y...
were truncated and the backend failed health checks, triggering rollback.
- Escape $ as $$ when writing production secrets to .env
- Document that deploy handles literal $ in Forgejo secrets
Production must not ship test users, demo orders, or test1234. Dev and CI
still need seeded users for e2e. Prod creates one admin from deploy secrets.
- Move V2/V4/V6 seed migrations to db/dev-migration
- Add application-prod.yml with schema-only Flyway and ignore-missing for moved seeds
- Add AdminBootstrap to create admin from ADMIN_EMAIL and ADMIN_PASSWORD
- Wire docker,prod profile, deploy secrets, and localhost:5433 for SSH DB access
- Add hashPassword Gradle task for optional manual bcrypt generation
The deploy pipeline had two critical bugs:
1. Health check used /api/vehicles/ZZZ999 with curl -f. This endpoint
returns HTTP 404 for unknown plates (correct behavior), which curl -f
treated as a failure. The backend was actually healthy.
Fix: use /api/vehicles/ABC123 (seeded in V6 migration, always 200)
and remove -f flag from curl.
2. No rollback on failure. If health checks failed, containers stayed
running forever because the pipeline exited 1 without stopping them.
Fix: combine health checks into one step. If either fails, run
'docker compose down' (without -v, so DB volume is preserved) before
exiting with failure.
The production deploy failed because port 3000 was already bound by the
dev frontend container (bilhej-frontend). The prod frontend doesn't need
a host port at all — nginx talks to it via the external 'web' network.
Changes:
- Remove host port binding (3000:80) from prod frontend
- Remove unused 'certs' volume from prod compose
- Use --project-name bilhej-prod in deploy workflow to isolate prod
containers/networks from dev and e2e environments
- Add 'docker compose down' before 'up' for clean deploys
- Update health check network names to bilhej-prod_default
The deploy workflow failed when re-running with the same version tag
because Git rejects pushing a tag that already exists on the remote.
- Delete local tag first (ignore if missing)
- Delete remote tag first (ignore if missing)
- Create and push the tag fresh
This makes deploys idempotent: retrying a failed deploy with the same
version (e.g., v0.1.0) will succeed by moving the tag to the current
commit. For a new deploy, the delete commands silently do nothing.
Add a manually-triggered deploy workflow that builds production Docker
images and starts the stack on the srvr.nu server.
- : workflow_dispatch with version input,
writes production .env from Forgejo secrets, builds and starts the
docker-compose.prod.yml stack, runs health checks via temporary curl
containers on the bilhej_default Docker network, tags the git commit.
- : nginx server block for bilhej.se.
Handles HTTP→HTTPS redirect, SSL termination with Let's Encrypt certs,
and proxies all traffic to the bilhej-frontend-prod container on the
Docker 'web' network. The frontend container handles /api/ proxying
to the backend internally.
To deploy:
1. Add production secrets to Forgejo (Settings → Actions → Secrets)
2. Trigger deploy from Actions → Deploy to Production
3. Run certbot for bilhej.se SSL (one-time setup)
4. Add docker/bilhej.nginx.conf to srvr.nu nginx container
5. Point bilhej.se DNS A record to srvr.nu IP
The 4-column table had a trailing empty column on the right because
the Status column was removed but the right border was still placed
after it. This created a visual glitch in the Forgejo log viewer.
- Convert to a clean 3-column table: Layer | Lines | Branch
- Remove emojis from table cells (they're double-width and break alignment)
- Add a plain-text pass/fail line below the table instead
- Use consistent 7-char padding for all percentage values so % signs align
Result: coverage summary renders cleanly in the CI job log.
Adds a step after backend coverage that parses the JaCoCo XML report
and prints a formatted table showing line and branch coverage with
pass/fail status against thresholds (70% lines, 60% branches).
The frontend coverage is already visible from Vitest's built-in text
reporter which prints during npm run test:coverage.
Both HTML reports remain downloadable as artifacts (backend-coverage
and frontend-coverage ZIPs).
Result: coverage numbers are visible at a glance in the CI job log
without needing to download and unzip artifacts.
actions/upload-artifact@v4 requires GHES features not available in
self-hosted Forgejo, causing artifact upload failures with:
GHESNotSupportedError: upload-artifact@v4+ are not supported on GHES.
- Downgrade both coverage upload steps from v4 to v3
- v3 uses a compatible upload mechanism that works on Forgejo
Keeps the artifact upload functionality so coverage HTML reports remain
downloadable from the workflow run page.
The lint-and-test job was running tests twice:
- 'Backend unit tests' ran tests without coverage
- 'Backend coverage' ran the same tests again with JaCoCo
- 'Frontend unit tests' ran tests without coverage
- 'Frontend coverage' ran the same tests again with v8 coverage
This wasted ~2x test time for no benefit since coverage steps already
run all tests.
- Remove 'Backend unit tests' and 'Frontend unit tests' steps
- Keep only coverage steps (jacocoTestCoverageVerification and test:coverage)
- Add artifact upload steps for both coverage HTML reports:
- backend-coverage: backend/build/reports/jacoco/test/html/
- frontend-coverage: frontend/coverage/
- 7-day retention to avoid storage bloat
Result: lint-and-test job runs faster (no duplicate test runs) and
produces downloadable HTML coverage reports visible in the Forgejo
Actions UI.
The Forgejo runner uses catthehacker/ubuntu:act-latest which does not have
a real GitHub Actions cache backend. actions/setup-node@v4 with cache: npm
spends ~4m44s trying to restore a non-existent cache during setup, and then
~4m40s in the post-job hook trying to save the cache during 'Complete job'.
- Remove cache: npm and cache-dependency-path from setup-node step
- npm ci without cache is fast enough for this project size (~10-20s)
Expected result: lint-and-test job drops from ~11m to ~2m total.
The per-job DinD approach failed because Forgejo Runner's service container
DNS resolution does not work when the runner itself uses DinD
(container.docker_host: tcp://dind:2375). The job container could not resolve
the 'dind' service hostname, causing docker compose to fail immediately.
New approach:
- Runner now uses container.docker_host: 'automount' which mounts the host
Docker socket into job containers. The runner runs as root (user: 0:0)
to access /var/run/docker.sock.
- E2E job no longer uses a 'dind' service. docker compose runs directly
against the host Docker daemon inside the job container.
- docker-compose.e2e.yml gets a custom 'e2e' bridge network. All E2E
containers (postgres, backend, frontend, playwright) attach only to this
network, isolating them from other host containers (Nextcloud, Jellyfin,
etc.). They can still reach the internet for vehicle lookup and npm.
Tradeoff: job containers can see other containers via docker ps, but they
are on an isolated network. For a single-user home server, this is the
simplest reliable configuration.
Implement per-job Docker-in-Docker (DinD) for E2E tests, giving each
job a completely isolated Docker daemon and network. This prevents
leakage to the host Docker or other containers.
The previous E2E approach failed because:
1. The Forgejo runner's container.docker_host was not set, causing
the runner itself to try unix:///var/run/docker.sock and crash-loop.
2. The host DinD daemon had isolated networking — job containers
running docker compose could not resolve 'dind' hostname or access
host filesystem bind mounts (e.g. .:/app).
New approach — zero bind mounts, all COPY-based images:
- docker/backend.e2e.Dockerfile: multi-stage build from repo root.
Copies gradlew + settings.gradle + backend/build.gradle to download
dependencies in a cacheable layer, then copies backend/src and builds
the bootJar. Runs the JAR directly on startup.
- docker/frontend.e2e.Dockerfile: multi-stage Node build → nginx.
Reuses existing docker/nginx.conf for /api proxy to backend service.
No volume mounts, fully self-contained.
- docker/playwright.e2e.Dockerfile: extends official Playwright image.
Installs deps from package-lock.json, copies e2e tests + config.
- docker-compose.e2e.yml: zero bind mounts. Services depend on each
other in order: postgres (healthy) → backend → frontend → playwright.
Playwright waits for backend and frontend via curl loops before
running tests.
- .forgejo/workflows/ci.yml: E2E job adds a 'dind' service container
(docker:28-dind, privileged, no TLS). The job sets DOCKER_HOST to
tcp://dind:2375 so the docker CLI inside the job talks to the
per-job DinD daemon. The compose file is docker-compose.e2e.yml.
- Runner fix on tocke: added container.docker_host: 'tcp://dind:2375'
to runner-config.yaml so the runner's own Docker client connects to
the host DinD container, stopping the crash loop.
Key properties:
- Network isolation: each E2E job gets its own DinD with its own
container network. No host container visibility.
- No bind mount leakage: all images use COPY instead of volume mounts.
The per-job DinD has its own filesystem and can't see host paths.
- Deterministic: builds start from clean state every time. Image cache
exists only within the per-job DinD lifetime.
- Lint-and-test job is untouched and remains green.
- Backend coverage runs from repo root where gradlew lives
- Frontend coverage runs from frontend/ with working-directory
- No cd tricks that break relative paths
- Remove working-directory: frontend from coverage step
- cd back to repo root for ./gradlew command, then cd frontend for npm
- Gradle wrapper lives at repo root, not in frontend/
- Rename FORGEJO_SERVER_URL to GITHUB_SERVER_URL
- The actions/checkout action reads GITHUB_SERVER_URL to construct the
clone URL. The runner was cloning https://srvr.nu/jocke/bilhej/ instead
of https://srvr.nu/git/jocke/bilhej/ because the /git/ subpath was lost
- Add .forgejo/workflows/ci.yml triggering on push/PR to master and develop
- Job lint-and-test: ESLint, vue-tsc type check, Vitest, JUnit, coverage
- Job e2e: Docker compose CI stack with Postgres, backend, frontend, Playwright
- Backend tests use H2 in-memory, no Postgres needed for unit tests
- E2E reuses existing docker-compose.ci.yml orchestration
- Strep env vars use fake test values since Stripe integration is deferred
