erp-nutzfahrzeuge/deploy/rollback.md

# Rollback Procedure — ERP Nutzfahrzeuge

---

## Rollback Strategy

The rollback strategy depends on the deployment method used.

---

## Coolify Rollback

### Option A: Redeploy Previous Image

1. In Coolify, navigate to the resource
2. Go to **Deployments** tab
3. Find the last known-good deployment
4. Click **Redeploy** on that deployment
5. Wait for health check to pass
6. Verify endpoints:
   - `GET /api/v1/health` → 200
   - `GET /` → 200

### Option B: Rollback via Git

1. In the repository, identify the last known-good commit:
   ```bash
   git log --oneline -10
   ```
2. Revert to the previous commit:
   ```bash
   git revert <commit_hash>  # or git reset --hard <commit_hash>
   git push origin main
   ```
3. Trigger a new deployment in Coolify (auto-deploy or manual)
4. Wait for health check to pass

---

## Docker Compose Rollback

### Step 1: Stop Current Deployment

```bash
docker-compose down
```

### Step 2: Revert Code

```bash
git checkout <last-known-good-tag-or-commit>
```

### Step 3: Rebuild and Restart

```bash
docker-compose build
docker-compose up -d
```

### Step 4: Verify

```bash
curl -s http://localhost:8000/api/v1/health
docker-compose logs --tail 50 backend
```

---

## Database Rollback

### If Alembic migrations are used:

```bash
cd backend
/opt/venv/bin/python -m alembic downgrade -1  # rollback one revision
# or
/opt/venv/bin/python -m alembic downgrade <revision_id>  # rollback to specific revision
```

### If no migration tool:

- Database schema changes are applied on app startup via SQLAlchemy `create_all`
- Rollback requires manual SQL or restoring from a database backup
- **Always take a database backup before deployment:**
  ```bash
  pg_dump -U erp_user erp_test > backup_$(date +%Y%m%d_%H%M%S).sql
  ```

### Restore from backup:

```bash
psql -U erp_user erp_test < backup_YYYYMMDD_HHMMSS.sql
```

---

## Rollback Decision Tree

| Situation | Action |
|-----------|--------|
| App won't start after deploy | Redeploy previous image (Coolify) or `git checkout` + rebuild (Docker) |
| Health check fails | Redeploy previous image; check DB connectivity |
| Database migration broke schema | `alembic downgrade` or restore from `pg_dump` backup |
| Frontend broken | Rebuild frontend from previous commit; redeploy |
| Performance regression | Redeploy previous image; investigate in staging |
| Security issue found | Rollback immediately; patch in staging; redeploy |

---

## Post-Rollback Verification

After any rollback:

1. ✅ Health check passes: `GET /api/v1/health` → 200 `{"status":"ok"}`
2. ✅ Root endpoint: `GET /` → 200
3. ✅ Login works: `POST /api/v1/auth/login` → 200 with valid credentials
4. ✅ No errors in logs: `docker logs <container> --tail 50`
5. ✅ Frontend loads at production URL
6. ✅ Database queries return expected data

---

## Emergency Rollback (Quick)

If the deployment is causing immediate issues:

```bash
# Coolify: use the Redeploy button on the last good deployment
# Docker Compose:
docker-compose down
git checkout <last-good-commit>
docker-compose build && docker-compose up -d
curl -s http://localhost:8000/api/v1/health
```

**Time to rollback:** < 5 minutes (if image is cached) or < 15 minutes (if rebuild needed).