We use resque as our background jobs systems, and resque-pool to manage multiple resque workers on a jobs server.
Starting and Stopping Workers
On a jobs server
sudo systemctl stop resque-pool.service
sudo systemctl start resque-pool.svc
You might want to stop workers so they stop accepting new work, in preparation to shut down a jobs server.
We are trying to have the stop be “graceful” will it give existing workers some amount of time to finish processing their jobs, but it might end up terminating workers in the middle of a job if they were really taking a long time. Workers terminated in the middle of their task should generate an error, reported from HoneyBadger. Since all of our jobs are idempotent, it should be safe to restart them in resque admin once workers are restarted, perhaps on a new server.
However, this has all been very confusing and hard to get right with resque. It may not be working exactly how we expect, and there is also room for improvement; look for a Github issue on it.