Something wrong with the translation?

Help us to improve it on our translation platform

Optimize memory usage

Funkwhale has a reasonable memory footprint. If you’re running Funkwhale on a limited device, you can use these tweaks to reduce the footprint.

Reduce workers concurrency

Funkwhale uses Celery to handle asynchronous tasks. By default, Celery spawns a worker per CPU core. This can lead to higher memory usage.

You can set the number of workers using the CELERYD_CONCURRENCY variable in your .env file. For example, a value of CELERYD_CONCURRENCY=1 spawns a single worker.

Note

Reducing the number of celery workers slows down the handling of asynchronous tasks. On larger instances, this can cause performance problems.

Switch to solo pool execution

Celery uses a prefork pool by default. This enables the server to process many tasks at the same time. You can switch to a solo pool which handles tasks one at a time. This reduces memory overhead but removes the ability to process tasks concurrently.

  1. Open your funkwhale-worker unit file in an editor.

    sudo nano /etc/systemd/system/funkwhale-worker.service

  2. Add the --pool=solo flag to the ExecStart line of your unit file.

    ExecStart=/srv/funkwhale/venv/bin/celery \
   --app funkwhale_api.taskapp \
   --pool solo \
   worker \
   --loglevel INFO \
   --concurrency=${CELERYD_CONCURRENCY}

  3. Restart the Celery service.

    sudo systemctl restart funkwhale-worker.service

  1. Add the --pool=solo flag to the celerybeat command in docker-compose.yml.

    celerybeat:
…
command: celery -A --pool=solo funkwhale_api.taskapp beat --pidfile= -l INFO

  2. Restart Celery.

    sudo docker compose restart celerybeat