Whys is Celery worker using /root as home directory?

Question

We are using Celery and RabbitMQ to process jobs. They are both being run by Supervisor so that they can be alive.

Some jobs expect a file in the home directory of a user A. So we run the celery worker as user A in supervisor. But when the job is being processed it is expanding the '~' as '/root'. Why is this happening? How can we ensure that it is able to find the home directory by expanding '~' as our code expects files following a '~' in a lot of places?

Answer 1

You probably need to use supervisord's user configuration directive to run the celery processes as a dedicated user instead of root.

also you should explicity chdir to a working dir - either in your celery code or use the supervisor directory setting.

If you want to rely on ~ working you should probably set the invironment variable $HOME also. However ~ is a shell expansion which may not always work. The most portable solution is to rely on the directory setting.

/etc/supervisor.d/celery.conf

[program:celery]
command=/path/to/celery worker
user=celery
stdout_logfile=/var/log/supervisor/celery.log
redirect_stderr=true
directory=/path/to/celery/working/dir
environment=HOME=/path/to/celery/home