Apache2 WSGI部署Flask后，如何将Celery设为守护进程自动运行？

Got it, since you've moved your Flask app to Apache2 WSGI and want to skip manually starting Celery every time, here are the most reliable ways to run Celery as a persistent daemon:

Method 1: Use Systemd (Recommended for Modern Linux Distros) #

Systemd is the standard init system on most current Linux releases (Ubuntu 16.04+, Debian 9+, etc.) and makes managing daemons straightforward.

Step 1: Create a Systemd Service File #

Create a new service file at /etc/systemd/system/celery-worker.service with the following content. Replace placeholders like /path/to/your/project and www-data with your actual project path and the user running your Apache/WSGI process:

[Unit]
Description=Celery Worker for Flask App
After=network.target rabbitmq-server.target
Wants=rabbitmq-server.target

[Service]
Type=forking
User=www-data
Group=www-data
WorkingDirectory=/path/to/your/project
Environment="PATH=/path/to/your/venv/bin"
ExecStart=/path/to/your/venv/bin/celery multi start worker1 -A app.app \
  --loglevel=INFO --concurrency=10 --pidfile=/var/run/celery/%n.pid \
  --logfile=/var/log/celery/%n%I.log
ExecStop=/path/to/your/venv/bin/celery multi stopwait worker1 \
  --pidfile=/var/run/celery/%n.pid
ExecReload=/path/to/your/venv/bin/celery multi restart worker1 -A app.app \
  --loglevel=INFO --concurrency=10 --pidfile=/var/run/celery/%n.pid \
  --logfile=/var/log/celery/%n%I.log

Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

Quick notes on this config:

• After=network.target rabbitmq-server.target ensures Celery starts only after RabbitMQ and network services are ready.
• Environment points to your virtual environment's bin directory (critical to avoid dependency conflicts).
• We use multi start to manage worker processes cleanly, with separate pid and log files.
• First create the required directories with: sudo mkdir -p /var/run/celery /var/log/celery and set permissions: sudo chown www-data:www-data /var/run/celery /var/log/celery.

Step 2: Enable and Start the Service #

Run these commands to activate the daemon:

# Reload systemd to detect the new service
sudo systemctl daemon-reload

# Start the Celery worker
sudo systemctl start celery-worker

# Set it to auto-start on system boot
sudo systemctl enable celery-worker

Step 3: Check Status and Logs #

• Verify the service is running: sudo systemctl status celery-worker
• View real-time logs: journalctl -u celery-worker -f

Method 2: Use Supervisor (Alternative for Older Systems) #

If you're on a Linux distro that doesn't use systemd, Supervisor is a solid option for managing process daemons.

Step 1: Install Supervisor #

sudo apt-get install supervisor  # For Debian/Ubuntu
# OR
sudo yum install supervisor     # For RHEL/CentOS

Step 2: Create a Supervisor Config File #

Create /etc/supervisor/conf.d/celery.conf with this content:

[program:celery]
command=/path/to/your/venv/bin/celery -A app.app worker --loglevel=INFO --concurrency=10 -n worker.%h
directory=/path/to/your/project
user=www-data
numprocs=1
autostart=true
autorestart=true
startsecs=10
stopwaitsecs=600
stdout_logfile=/var/log/celery/celery.log
stderr_logfile=/var/log/celery/celery.err.log

Don't forget to create the log directory if needed: sudo mkdir -p /var/log/celery && sudo chown www-data:www-data /var/log/celery

Step 3: Update Supervisor and Start the Process #

# Reload Supervisor to load the new config
sudo supervisorctl reread
sudo supervisorctl update

# Start the Celery process
sudo supervisorctl start celery

Key Things to Verify #

• Environment Consistency: Ensure Celery uses the same configuration (RabbitMQ connection, environment variables) as your WSGI-deployed Flask app. If you rely on env vars for config, add them to your systemd service or Supervisor config.
• Permissions: The user running Celery (usually www-data) needs read access to your project files and permissions to connect to RabbitMQ.
• Virtual Environment: Always use the celery executable from your project's virtual environment to avoid dependency mismatches.

内容的提问来源于stack exchange，提问作者Avi