Skip to content

Production Checklist

What to check before you open access to real users.


Why you need a checklist

Dev and production are different environments. Django has convenient defaults for dev (DEBUG=True, SQLite, InMemory channels), but they are completely unsafe for production. A checklist helps you not miss a single critical change.

Dev:   DEBUG=True   → stack trace with SECRET_KEY is visible → attacker forges cookies
Prod:  DEBUG=False  → empty 500 error, SECRET_KEY hidden

Dev:   SQLite       → file is destroyed on docker compose down
Prod:  PostgreSQL   → named volume, data persists between restarts

Dev:   InMemory     → WebSocket only within one process
Prod:  Redis        → broadcast between several workers

Dev:   runserver    → WSGI, single-threaded, no static files
Prod:  Uvicorn + nginx → ASGI, several workers, static without Python

Security (mandatory)

  • [ ] SECRET_KEY — unique, 50+ characters, generated with get_random_secret_key()
  • [ ] DEBUG = False — NEVER True in production
  • [ ] ALLOWED_HOSTS — specific domains, not ['*']
  • [ ] CSRF_TRUSTED_ORIGINS — contains the prod domain and the ngrok domain
  • [ ] SESSION_COOKIE_SECURE = True — cookie over HTTPS only
  • [ ] CSRF_COOKIE_SECURE = True
  • [ ] SECURE_SSL_REDIRECT = True — HTTP → HTTPS redirect
  • [ ] SECURE_HSTS_SECONDS = 31536000 — HSTS for 1 year
  • [ ] .env is in .gitignore — check git status before pushing
  • [ ] Passwords are in .env, not in the code or the Dockerfile

HTTPS settings in settings.py

# settings.py — uncomment when you have an SSL certificate:
SESSION_COOKIE_SECURE = True    # cookie over HTTPS only
CSRF_COOKIE_SECURE = True
SECURE_SSL_REDIRECT = True      # HTTP → HTTPS redirect
SECURE_HSTS_SECONDS = 31536000  # the browser remembers HTTPS for 1 year
SECURE_HSTS_INCLUDE_SUBDOMAINS = True
SECURE_HSTS_PRELOAD = True

Performance

  • [ ] Remove --reload from entrypoint.sh (uvicorn)
  • [ ] Add --workers 4 to uvicorn (4 = 2 × CPU cores)
  • [ ] CONN_MAX_AGE = 0 for ASGI (async-compatible)
  • [ ] nginx expires 7d for static files (caching)
  • [ ] PostgreSQL max_connections matches the number of uvicorn workers
  • [ ] Redis maxmemory-policy allkeys-lru (don't let Redis fill up RAM)

Production entrypoint.sh

# entrypoint.sh in production — remove --reload, add --workers:
exec python -m uvicorn notes_project.asgi:application \
    --host 0.0.0.0 \
    --port 8001 \
    --workers 4

Django deployment check

docker compose exec web python manage.py check --deploy

It shows warnings about security settings. In production they should all be green.

# With DEBUG=True (expected in dev):
WARNINGS:
?: (security.W004) You have not set a value for the SECURE_HSTS_SECONDS setting.
?: (security.W008) Your SECRET_KEY has less than 50 characters.
?: (security.W012) SESSION_COOKIE_SECURE is not set to True.
?: (security.W016) You have 'django.middleware.csrf.CsrfViewMiddleware' in your
   MIDDLEWARE, but you have not set CSRF_COOKIE_SECURE to True.

# With DEBUG=False and correct settings:
System check identified no issues (0 silenced).

Generating a SECRET_KEY

# Option 1: Python (via Docker)
docker compose run --rm web python -c \
    "from django.core.management.utils import get_random_secret_key; print(get_random_secret_key())"

# Option 2: Python (locally)
python3 -c "from django.core.management.utils import get_random_secret_key; print(get_random_secret_key())"

# Option 3: OpenSSL
openssl rand -base64 50

# Option 4: Python secrets
python3 -c "import secrets; print(secrets.token_urlsafe(50))"

NEVER use

SECRET_KEY=django-insecure-...   ← generated by Django by default
SECRET_KEY=dev-secret-key-...    ← obviously a test key
Both of these values are public and do not protect cookies or HMAC signatures.


Pre-deploy checks

1. Check the tests

docker compose run --rm web python manage.py test notes_app.tests --failfast
# All tests must be green before deploying

2. Check migrations

docker compose exec web python manage.py showmigrations
# All migrations should have [X]
# If there is [ ] → migrate did not run or there is a new unapplied migration

docker compose exec web python manage.py migrate --check
# Returns exit code 1 if there are unapplied migrations

3. Check static files

docker compose exec nginx curl -I http://localhost/static/admin/css/base.css
# HTTP/1.1 200 OK → nginx serves static files correctly
# HTTP/1.1 404    → collectstatic did not run or the volume is not mounted

docker compose exec web python manage.py collectstatic --dry-run --noinput
# Shows what would be collected without actually copying

4. Check the list of environment variables

docker compose exec web env | sort | grep -vE 'PATH|HOME|USER|SHELL'
# Make sure these exist: SECRET_KEY, DATABASE_URL, REDIS_URL, ALLOWED_HOSTS
# Make sure DEBUG=False
# Make sure SECRET_KEY != 'dev-secret-key-...'

Monitoring and logging

  • [ ] LOGGING in settings.py → files or stdout for Docker
  • [ ] nginx access log → analysis (Grafana Loki or ELK)
  • [ ] Health check endpoints respond
  • [ ] Uptime monitor (UptimeRobot, Better Stack)
# settings.py — basic logging configuration:
LOGGING = {
    'version': 1,
    'disable_existing_loggers': False,
    'formatters': {
        'verbose': {
            'format': '{levelname} {asctime} {module} {message}',
            'style': '{',
        },
    },
    'handlers': {
        'console': {
            'class': 'logging.StreamHandler',
            'formatter': 'verbose',   # stdout → docker logs → centralized logging
        },
    },
    'root': {
        'handlers': ['console'],
        'level': 'INFO',
    },
    'loggers': {
        'django.request': {
            'handlers': ['console'],
            'level': 'WARNING',       # only 4xx and 5xx, not every GET
            'propagate': False,
        },
    },
}
# View logs per service:
docker compose logs --since 1h web      # logs for the last hour
docker compose logs --since 1h nginx    # nginx access log
docker compose logs --since 1h --follow web  # live streaming

Backing up PostgreSQL

Before any deploy that changes the schema — make a backup:

# Back up the DB to a file:
docker compose exec db pg_dump \
    -U notes_user \
    -d notes_db \
    -F c \
    -f /tmp/backup_$(date +%Y%m%d_%H%M%S).dump

# Copy the backup from the container to the host:
docker cp $(docker compose ps -q db):/tmp/backup_*.dump ./backups/

# Restore from the backup:
docker compose exec db pg_restore \
    -U notes_user \
    -d notes_db \
    -F c \
    /tmp/backup.dump

Rollback

If a deploy broke the application — the rollback plan:

1. Roll back the code (previous Docker image)

# Stop the current version:
docker compose down

# Build the previous image (if there is a tag):
git checkout v1.2.3
docker compose up --build

# Or use the cached image:
docker compose up    # no --build, uses the previous build

2. Roll back migrations

# View the list of migrations:
docker compose exec web python manage.py showmigrations notes_app

# Roll back to a specific migration:
docker compose exec web python manage.py migrate notes_app 0005
# ↑ rolls back 0006, 0007, ... to the state after 0005
# WARNING: deletes data from new columns!

Migrations with data

Rolling back a migration that removed a column → impossible (the data is already deleted). Rolling back a migration that added a column → safe (the empty column is simply removed). Always make a backup before a deploy with migrations.


Common mistakes on the first deploy

Error Cause Fix
403 CSRF verification failed Domain not in CSRF_TRUSTED_ORIGINS Add the domain in settings.py + env
DisallowedHost Domain not in ALLOWED_HOSTS Add the domain + restart
WebSocket 403 Auth is not passed through nginx Check proxy_set_header in nginx.conf
Static files 404 collectstatic did not run Check entrypoint.sh + volume mount
ERR_NGROK_8012 ngrok cannot find nginx Check that both are in app-net
TimeoutError reading from redis socket_timeout in RedisChannelLayer socket_timeout=None in CONFIG
CommandError: Demo data disabled seed_demo_data is disabled without DEBUG=True Add --force to the command
SynchronousOnlyOperation in a consumer sync ORM in an async context @database_sync_to_async
Data disappeared after docker compose down down -v was used Back up before down -v
web unhealthy → nginx does not start Migrate fails or uvicorn does not start docker compose logs web
WS connection does not open through nginx Missing WebSocket headers Add proxy_http_version 1.1 in nginx.conf
SECRET_KEY changed → all sessions invalid Changing SECRET_KEY → Django invalidates sessions Don't change SECRET_KEY unless necessary

In the book


Official documentation