Files
API/.gitea/workflows/deploy.yml
T
PaddyandClaude Opus 5.5 ae4f153482
Test and deploy / test-and-deploy (pull_request) Successful in 33s
Run tests and deploy in one Gitea Actions job
Two jobs meant checking out, setting up Node and running npm ci (and in the
frontend the build) twice, about two minutes each, because every job starts
in a fresh container. Now one job tests and, for pushes to master only,
deploys. The push-to-master condition is spelled out on every deploy step
instead of an env flag compared to 'true', which would skip the deploy
silently and stay green if the boolean were rendered differently. A failed
test still stops the deploy (an if: without a status function implies
success()).

Docs: the one-job layout, and that Jenkins deployed as root - files it left
in an app folder must be chowned to the subscription user before the first
Gitea deploy (hit on the API).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 22:51:30 +01:00

107 lines
5.3 KiB
YAML

# Tests every push and pull request; deploys pushes to master to the Plesk server.
# Setup and what the deploy may touch: docs/deployment.md.
#
# Same pattern as the calendar frontend (Calendar_Frontend/.gitea/workflows/deploy.yml): build here,
# deliver over SSH as the Plesk subscription user, restart Passenger. Unlike there, the app folder
# also holds things the deploy must never touch (.env with the secrets, logs/, tmp/), so only the
# app's own folders are synced with --delete and nothing is deleted at the top level.
name: Test and deploy
on:
push:
branches: [master]
pull_request:
jobs:
# One job, not test + deploy: every job starts in a fresh container, so two jobs meant checking out,
# setting up Node and running `npm ci` twice (~2 min each, 2026-11-08). The deploy steps only run
# for pushes to master; on pull requests the job ends after the tests. The condition is spelled out
# on every deploy step rather than stored in an env flag: a flag compared as a string ('true') would
# skip the deploy silently - and green - if the runner ever rendered the boolean differently.
test-and-deploy:
runs-on: ubuntu-latest
env:
DEPLOY_HOST: ${{ vars.DEPLOY_HOST }}
DEPLOY_PORT: ${{ vars.DEPLOY_PORT }}
DEPLOY_USER: ${{ vars.DEPLOY_USER }}
DEPLOY_PATH: ${{ vars.DEPLOY_PATH }}
# Absolute on purpose: rsync passes its -e string to ssh without a shell, and ssh expands "~"
# from the passwd entry, not $HOME (see the calendar frontend's docs/gotchas.md).
SSH_DIR: /tmp/deploy-ssh
steps:
- uses: actions/checkout@v4
- uses: actions/setup-node@v4
with:
node-version-file: .nvmrc
- run: npm ci
- run: npx tsc --noEmit
# Unit tests only. The integration tests (npm run test:integration) start their own MariaDB
# container and aren't run here.
# FEEDBACK_IP_SALT (which the Jenkins job exported) isn't needed: vitest.config.ts sets one.
- run: npm test
- name: Build
if: github.event_name == 'push' && github.ref == 'refs/heads/master'
run: npm run build
# Only runtime dependencies go to the server (no typescript, vitest, ...).
- name: Prune dev dependencies
if: github.event_name == 'push' && github.ref == 'refs/heads/master'
run: npm prune --omit=dev
- name: SSH setup
if: github.event_name == 'push' && github.ref == 'refs/heads/master'
env:
DEPLOY_SSH_KEY: ${{ secrets.DEPLOY_SSH_KEY }}
DEPLOY_KNOWN_HOSTS: ${{ secrets.DEPLOY_KNOWN_HOSTS }}
run: |
test -n "$DEPLOY_HOST" && test -n "$DEPLOY_USER" && test -n "$DEPLOY_PATH"
if ! command -v rsync >/dev/null 2>&1 || ! command -v ssh >/dev/null 2>&1; then
apt-get update && apt-get install -y --no-install-recommends rsync openssh-client
fi
install -m 700 -d "$SSH_DIR"
printf '%s\n' "$DEPLOY_SSH_KEY" > "$SSH_DIR/key"
chmod 600 "$SSH_DIR/key"
# Pinned host key, never trust-on-first-use.
printf '%s\n' "$DEPLOY_KNOWN_HOSTS" > "$SSH_DIR/known_hosts"
- name: Upload
if: github.event_name == 'push' && github.ref == 'refs/heads/master'
run: |
rsh="ssh -p ${DEPLOY_PORT:-22} -i $SSH_DIR/key -o UserKnownHostsFile=$SSH_DIR/known_hosts -o IdentitiesOnly=yes -o BatchMode=yes"
target="$DEPLOY_USER@$DEPLOY_HOST:$DEPLOY_PATH"
# The app's own folders, each mirrored exactly (--delete within the folder only).
# src/ is needed at runtime: swagger-jsdoc reads the route docs from src/models/**/*.ts.
for dir in dist src public node_modules; do
rsync -az --delete -e "$rsh" "$dir/" "$target/$dir/"
done
# Single files at the top level, without --delete: .env, logs/ and tmp/ live there too.
# package.json is what makes Node treat dist/ as ES modules ("type": "module").
rsync -az -e "$rsh" package.json package-lock.json "$target/"
- name: Restart
if: github.event_name == 'push' && github.ref == 'refs/heads/master'
run: |
# Passenger restarts the app on its next request when tmp/restart.txt is touched.
ssh -p "${DEPLOY_PORT:-22}" -i "$SSH_DIR/key" -o "UserKnownHostsFile=$SSH_DIR/known_hosts" \
-o IdentitiesOnly=yes -o BatchMode=yes "$DEPLOY_USER@$DEPLOY_HOST" \
"mkdir -p '$DEPLOY_PATH/tmp' && touch '$DEPLOY_PATH/tmp/restart.txt'"
# Set DEPLOY_SMOKE_URL to an endpoint that must answer 200 without a session, e.g.
# https://api.nachklang.art/calendar/events/public/json (what nachklang.art itself reads).
- name: Smoke test
if: github.event_name == 'push' && github.ref == 'refs/heads/master' && vars.DEPLOY_SMOKE_URL != ''
env:
DEPLOY_SMOKE_URL: ${{ vars.DEPLOY_SMOKE_URL }}
run: |
# Passenger starts the app on the first request after the restart; give it a moment.
for _ in 1 2 3 4 5 6; do
if curl -fsS -o /dev/null "$DEPLOY_SMOKE_URL"; then echo "OK: $DEPLOY_SMOKE_URL"; exit 0; fi
sleep 5
done
echo "Smoke test failed: $DEPLOY_SMOKE_URL"
exit 1
- name: Remove key
if: always() && github.event_name == 'push' && github.ref == 'refs/heads/master'
run: rm -f "$SSH_DIR/key"