Files
collective-lists/Justfile
Oier Bravo Urtasun ae4fd45b99 deploy(prod): pre-deploy DB backup + paired code/DB rollback skill
Every `just deploy` now takes a pg_dumpall to
/opt/colectivo/backups/predeploy-<ts>-<sha>.sql.gz BEFORE rebuilding the
app or applying migrations, aborts the deploy if the dump is < 1 KiB,
and (on success) appends `<iso-ts> \t <sha> \t <backup-file>` to
/opt/colectivo/.deploys.log on ambrosio. Keeps the newest 10 backups
(prunes older predeploy-* files).

New `infra/scripts/rollback-erosi.sh` reads .deploys.log and pairs a
code rollback with a DB restore atomically:

  just rollback-list          # show recent deploys
  just rollback               # roll back to N-1
  just rollback-to <sha>      # roll back to a specific deploy
  just rollback-code          # roll back code only, keep current DB

Rollback safety:
- 5-second abort window.
- Verifies the target SHA exists locally + the backup is still on prod
  (warns if it's been pruned past the 10-deploy window).
- Uses a temporary git worktree so the user's working tree isn't
  disturbed.
- Stops app/auth/rest/realtime/storage before the gunzip|psql restore.
- Rebuilds the app image at the rolled-back SHA with the same GIT_SHA
  build-arg path the deploy uses (so __APP_COMMIT__ in the bundle
  matches the running code).
- Does NOT write a new .deploys.log entry — rollback is intentionally
  not a deploy event; the next `just deploy` is from current HEAD.
- Storage volume (/var/lib/storage user uploads) is NOT rolled back.

Justfile `deploy` recipe repointed from the stale `deploy.sh` (which
referenced GHCR pull) to `deploy-erosi.sh` (the active prod path).

New project-scoped skill: `.claude/skills/deploy/SKILL.md` documents the
flow + safety boundaries for Claude-assisted invocations. `.gitignore`
keeps `.claude/settings.local.json` ignored but tracks `.claude/skills/`.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-18 14:57:25 +02:00

232 lines
8.7 KiB
Makefile

# Colectivo — task runner
# Requires: just, docker, pnpm
set dotenv-load := true
# Alias: docker compose with correct project root so .env is always found
dc_dev := "docker compose --env-file .env -f infra/docker-compose.dev.yml"
dc_prod := "docker compose --env-file .env -f infra/docker-compose.prod.yml"
# ── Setup ─────────────────────────────────────────────────────────────────────
# First-time (and idempotent) local environment setup
setup:
bash setup.sh
# ── Dev ───────────────────────────────────────────────────────────────────────
# Start full local stack (Docker services + SvelteKit dev server)
dev:
{{dc_dev}} up -d
pnpm --filter @colectivo/web dev
# Stop all services (Docker stack + SvelteKit dev server)
stop:
{{dc_dev}} down
-pkill -f "vite" 2>/dev/null || true
# Stop local Docker stack
dev-stop:
{{dc_dev}} down
# Stop and remove all volumes (full reset)
dev-clean:
{{dc_dev}} down -v
# ── Database ──────────────────────────────────────────────────────────────────
# Drop, migrate, and seed the dev database
db-reset:
supabase db reset
# Apply pending migrations
db-migrate:
supabase db push
# Apply dev seed data
db-seed:
docker exec -i colectivo-dev-db-1 psql -U postgres < supabase/seed.sql
# Regenerate TypeScript types from Supabase schema.
# Writes to a tempfile first so a missing/failing `supabase` CLI does not
# truncate the existing hand-curated database.ts (which would silently
# break every TS consumer in the monorepo).
db-types:
#!/usr/bin/env bash
set -euo pipefail
if ! command -v supabase >/dev/null 2>&1; then
echo "supabase CLI not installed — database.ts is hand-curated, edit it manually" >&2
exit 1
fi
tmp=$(mktemp)
supabase gen types typescript --local > "$tmp"
mv "$tmp" packages/types/src/database.ts
echo "Types written to packages/types/src/database.ts"
# Open Supabase Studio
db-open:
open http://localhost:54323
# ── Keycloak ──────────────────────────────────────────────────────────────────
# Export Keycloak realm to keycloak/realm-export.json
kc-export:
{{dc_dev}} exec keycloak \
/opt/keycloak/bin/kc.sh export \
--realm colectivo \
--users realm_file \
--file /tmp/realm-export.json
{{dc_dev}} cp keycloak:/tmp/realm-export.json keycloak/realm-export.json
echo "Exported to keycloak/realm-export.json"
# Open Keycloak Admin Console
kc-open:
open http://localhost:8080/admin
# ── Build ─────────────────────────────────────────────────────────────────────
# Build all packages
build:
pnpm turbo run build
# Run the production build locally against the dev Docker stack (port 3000)
serve: build
{{dc_dev}} up -d
PORT=3000 node apps/web/build/index.js
# Type-check all packages
check:
pnpm turbo run check
# Lint all packages
lint:
pnpm turbo run lint
# Run tests
test:
pnpm turbo run test
# ── Testing ───────────────────────────────────────────────────────────────────
# Run every test suite (pgTAP + Vitest integration + apps/web unit + Playwright E2E).
# Requires `just dev` to already be running — the stack must be healthy.
test-all: test-db test-integration test-unit test-e2e
# Run apps/web Vitest unit tests (browser-env + jsdom + fake-indexeddb).
# Currently covers the offline sync queue (Fase 2b.2).
test-unit:
pnpm --filter @colectivo/web test:unit
# Run pgTAP SQL tests directly against the dev database
# `set dotenv-load` at the top of this file sources .env so POSTGRES_PASSWORD
# is available — we forward it to psql via PGPASSWORD.
test-db:
#!/usr/bin/env bash
set -euo pipefail
for f in supabase/tests/*.sql; do
PGPASSWORD="$POSTGRES_PASSWORD" psql -U postgres -h localhost -d postgres \
-v ON_ERROR_STOP=1 -f "$f"
done
# Run Vitest integration tests (RLS + trigger assertions via supabase-js)
test-integration:
pnpm --filter @colectivo/test-utils test
# Install Playwright browsers (chromium + webkit). WebKit is required for
# touch-gesture specs (Fase 9.4 swipe-toggle) — the *.webkit.test.ts files
# are gated to that project in playwright.config.ts.
playwright-install:
pnpm --filter @colectivo/web exec playwright install chromium webkit
# Run Playwright E2E tests (headless). Includes the WebKit project — make
# sure `just playwright-install` has run at least once.
test-e2e:
pnpm --filter @colectivo/web exec playwright test
# Run Playwright E2E tests in headed mode (for debugging)
test-e2e-headed:
pnpm --filter @colectivo/web exec playwright test --headed
# Run only the WebKit (Mobile Safari) touch-gesture specs (Fase 9.4).
# Gated behind RUN_WEBKIT=1 in playwright.config so it stays out of the
# default test-all (WebKit needs system libs not available on every host
# — see `just playwright-install` and `sudo pnpm exec playwright install-deps`).
test-webkit:
RUN_WEBKIT=1 pnpm --filter @colectivo/web exec playwright test --project=webkit
# Run Kong rate-limit verification tests. Restarts Kong first to ensure a
# clean counter state (tests burn through the minute + hour quotas). Not
# part of test-all because the tests share the rate-limit counter with the
# rest of the integration suite.
test-rate-limit:
#!/usr/bin/env bash
set -euo pipefail
{{dc_dev}} up -d --force-recreate kong
sleep 3
export RUN_RATE_LIMIT_TESTS=1
pnpm --filter @colectivo/test-utils test -- rate-limit
# Build a prod bundle, serve it on :3000, run Lighthouse against PWA /
# Accessibility / Best-Practices. Fails if any score drops below 90.
# Must run while `just dev` (docker stack) is up — Lighthouse loads the
# real app which hits Supabase via Kong.
lighthouse:
#!/usr/bin/env bash
set -euo pipefail
pnpm --filter @colectivo/web build
PORT=3000 node apps/web/build/index.js &
SERVER_PID=$!
trap "kill $SERVER_PID 2>/dev/null || true" EXIT
for i in 1 2 3 4 5 6 7 8 9 10; do
if curl -sf http://localhost:3000/ -o /dev/null; then break; fi
sleep 1
done
pnpm --filter @colectivo/web exec lighthouse http://localhost:3000 \
--only-categories=pwa,accessibility,best-practices \
--output=html --output-path=./lighthouse-report.html \
--chrome-flags='--headless --no-sandbox' \
--quiet
# Run the RLS isolation curl audit — complements rls-audit.test.ts.
rls-audit:
infra/scripts/rls-audit.sh
# ── Backup & Restore ──────────────────────────────────────────────────────────
# Take a manual backup of a service (e.g. just backup supabase)
backup service:
infra/scripts/backup.sh {{service}}
# Restore a backup file (e.g. just restore supabase backups/2024-01-01.dump)
restore service file:
infra/scripts/restore.sh {{service}} {{file}}
# ── Production ────────────────────────────────────────────────────────────────
# Deploy to production via SSH (ambrosio / erosi.limonia.net).
# Pre-deploy: dumps the prod DB to /opt/colectivo/backups/predeploy-<ts>-<sha>.sql.gz
# Post-deploy: appends (timestamp, sha, backup) to /opt/colectivo/.deploys.log
deploy:
infra/scripts/deploy-erosi.sh
# List recent prod deploys (timestamp / sha / backup-file).
rollback-list:
infra/scripts/rollback-erosi.sh --list
# Roll back code AND database to the previous deploy.
# For a specific deploy: `just rollback-to <sha>`.
rollback:
infra/scripts/rollback-erosi.sh --previous
# Roll back to a specific deploy by sha (must exist in .deploys.log on prod).
rollback-to sha:
infra/scripts/rollback-erosi.sh --to={{sha}}
# Roll back only code (skip DB restore). Useful if the regression is UI-only.
rollback-code:
infra/scripts/rollback-erosi.sh --previous --code-only
# Stream production Docker logs
logs:
{{dc_prod}} logs -f