trails/.github/workflows/cd-infra.yml
Ullrich Schäfer e60c9d7057 refactor(infra): mount config dirs not single files; reload on deploy
Single-file bind mounts (./foo.yml:/etc/foo.yml) pin to the host file's
inode at container-create time. The CD pipeline scp's a replacement file
(new inode), so the running container keeps reading the OLD inode —
`docker compose up -d` won't recreate on a content-only change, and
neither restart/SIGHUP/`caddy reload` re-reads the new file. Net effect:
config-only infra PRs deployed "successfully" but never took effect
(confirmed with PR #500's prometheus.yml; the WAL retention work only
applied because #498 also changed docker-compose.yml, forcing a
recreate). Caddyfile changes had the same latent gap.

Switch the four single-file config mounts to DIRECTORY mounts, which
resolve children live so a reload/restart picks up the new file:

- prometheus, loki, promtail: mount ./<svc> dir (loki/promtail
  --config.file paths updated to the real filenames).
- caddy: move Caddyfile into caddy/ and mount the dir; ./sites overlays
  /etc/caddy/sites (caddy/sites/.gitkeep keeps the mountpoint). Container
  path /etc/caddy/Caddyfile is unchanged, so every `caddy reload --config
  /etc/caddy/Caddyfile` ref still works. scp source paths in cd-infra and
  cd-apps updated to ship the caddy/ dir.

cd-infra now applies config-only changes after `up -d`: SIGHUP prometheus
(zero downtime), restart loki+promtail (no SIGHUP reload), caddy reload
(graceful). Mirrored prometheus/loki mounts in docker-compose.dev.yml.

Validated: docker compose config (both files), caddy validate from the
new path, read-only-parent + sites-overlay mount mechanics, workflow YAML.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-09 13:10:35 +02:00

136 lines
6.3 KiB
YAML

name: CD Infra
on:
push:
branches: [main]
paths:
- "infrastructure/**"
workflow_dispatch:
inputs:
restart_all:
description: "Restart all containers (not just infra)"
type: boolean
default: false
concurrency:
group: deploy-infra
cancel-in-progress: true
jobs:
deploy:
name: Deploy Infrastructure
runs-on: ubuntu-latest
environment: infra
steps:
- uses: actions/checkout@v6
- name: Decrypt secrets
run: |
curl -sLO https://github.com/getsops/sops/releases/download/v3.9.4/sops-v3.9.4.linux.amd64
chmod +x sops-v3.9.4.linux.amd64
# Merge app + infra secrets into one .env for docker-compose
SOPS_AGE_KEY="${{ secrets.AGE_SECRET_KEY }}" ./sops-v3.9.4.linux.amd64 -d infrastructure/secrets.app.env > infrastructure/.env
SOPS_AGE_KEY="${{ secrets.AGE_SECRET_KEY }}" ./sops-v3.9.4.linux.amd64 -d infrastructure/secrets.infra.env >> infrastructure/.env
echo "DOMAIN=trails.cool" >> infrastructure/.env
# Flagship marker — the Journal home renders the project
# marketing block when this is "true" and the self-host
# footer link when it's unset.
echo "IS_FLAGSHIP=true" >> infrastructure/.env
- name: Copy configs to server
uses: appleboy/scp-action@v1
with:
host: ${{ secrets.DEPLOY_HOST }}
username: root
key: ${{ secrets.DEPLOY_SSH_KEY }}
source: "infrastructure/docker-compose.yml,infrastructure/caddy,infrastructure/prometheus/prometheus.yml,infrastructure/loki/loki-config.yml,infrastructure/promtail/promtail-config.yml,infrastructure/postgres/queries.yml,infrastructure/postgres/init-grafana-user.sql,infrastructure/grafana/provisioning,infrastructure/grafana/dashboards"
target: /opt/trails-cool
strip_components: 1
- name: Copy secrets to server
uses: appleboy/scp-action@v1
with:
host: ${{ secrets.DEPLOY_HOST }}
username: root
key: ${{ secrets.DEPLOY_SSH_KEY }}
source: "infrastructure/.env"
target: /opt/trails-cool
strip_components: 1
- name: Deploy via SSH
uses: appleboy/ssh-action@v1
with:
host: ${{ secrets.DEPLOY_HOST }}
username: root
key: ${{ secrets.DEPLOY_SSH_KEY }}
script: |
# Abort on first failure. The 2026-06-06/07 outage: a network
# recreation stopped postgres, a later step failed, and the
# deploy left production down for ~9h while the job's partial
# progress looked plausible. Fail fast, verify health at the end.
set -euo pipefail
cd /opt/trails-cool
# .env was placed by the SCP step (decrypted app + infra secrets)
# Login to ghcr.io (for pulling monitoring images)
GHCR_TOKEN=$(grep DEPLOY_GHCR_TOKEN .env | cut -d= -f2-)
echo "$GHCR_TOKEN" | docker login ghcr.io -u stigi --password-stdin 2>/dev/null || true
# Setup Grafana read-only DB user (idempotent)
docker compose exec -T postgres psql -U trails -d trails -f /docker-entrypoint-initdb.d/init-grafana-user.sql 2>/dev/null || true
GRAFANA_DB_PW=$(grep GRAFANA_DB_PASSWORD .env | cut -d= -f2-)
if [ -n "$GRAFANA_DB_PW" ]; then
docker compose exec -T postgres psql -U trails -d trails -c "ALTER ROLE grafana_reader PASSWORD '$GRAFANA_DB_PW'" 2>/dev/null || true
fi
# Full restart: gh workflow run cd-infra.yml -f restart_all=true
if [ "${{ github.event.inputs.restart_all }}" = "true" ]; then
docker compose --env-file .env up -d --remove-orphans
else
# Restart infra services (config reloads handled below).
# --remove-orphans cleans up containers whose service was deleted
# from the compose file (e.g., the flagship `brouter` removal in
# PR #297 left an orphan that had to be removed by hand).
docker compose --env-file .env up -d --remove-orphans postgres prometheus loki promtail grafana postgres-exporter node-exporter cadvisor
fi
# Apply config-only changes that `up -d` skips — it recreates a
# container only when its compose *definition* changes, not when a
# mounted config file's content changes. The configs are mounted as
# directories (not single files), so a reload/restart re-reads the
# freshly scp'd file; a single-file mount would have pinned the old
# inode. Prometheus hot-reloads on SIGHUP (zero downtime); Loki and
# Promtail reload their main config only on restart; Caddy reloads
# gracefully (validates, swaps live, no downtime).
docker compose --env-file .env kill -s SIGHUP prometheus
docker compose --env-file .env restart loki promtail
docker compose exec caddy caddy reload --config /etc/caddy/Caddyfile
docker compose ps
# Gate on the stack actually being up: postgres healthy and —
# since an infra restart bounces the app containers' database —
# journal back to healthy too. A deploy that leaves either down
# must fail loudly (see the 2026-06-06/07 outage).
for ctr in trails-cool-postgres-1 trails-cool-journal-1; do
for i in $(seq 1 36); do
status=$(docker inspect -f '{{.State.Health.Status}}' "$ctr" 2>/dev/null || echo missing)
[ "$status" = "healthy" ] && break
sleep 5
done
if [ "$status" != "healthy" ]; then
echo "$ctr did not become healthy (last status: $status)"
exit 1
fi
done
# Annotate deploy in Grafana
GRAFANA_TOKEN=$(grep GRAFANA_SERVICE_TOKEN .env | cut -d= -f2-)
if [ -n "$GRAFANA_TOKEN" ]; then
docker compose exec -T grafana curl -sf -X POST \
-H "Authorization: Bearer $GRAFANA_TOKEN" \
-H "Content-Type: application/json" \
-d '{"text":"Deploy infra ${{ github.sha }}","tags":["deploy","infra"]}' \
http://localhost:3000/api/annotations || true
fi