After power outage, worker containers will failed to start because the name of the worker is already registered in the redis database.
The workaround is to remove de container and re-create it.
The permanent fix would be to have a script that:
- Checks the current workers (Group A)
- Checks the workers on the redis db (Group B)
- Remove (Group B - Group A)
- Move tasks (Group B - Group A) were handling to the failed queue.
Related
rq/rq#787