Hello, I have issues with concourse worker.

Sometimes, the worker container stalls (builds are frozen, check-resources hang indefinitely), even though still appearing as "running " when using the fly CLI worker command.

Problem is, this container cannot be stopped.
docker-compose stop, docker-compose kill, docker stop, docker kill, all fail with
either "ERROR: An HTTP request took too long to complete. Retry with --verbose to obtain debug information." or no error msg at all.

Stopping/restarting the docker service does not work, rebooting the OS does not work.

Only hard-rebooting the server makes the problem goes away. This is a huge problem imo, since this means I can't run concourse on any important server (I can't reboot production servers just like that).
Problem is, I have no idea what makes the worker crash in a so nasty way

Bug Report

Bug reports are pretty free-form; just replace this with whatever. You can also help us triage the issue by including steps to reproduce, expected results, and the actual result. Help us help you!

The following can also be handy:

• Concourse version: 3.8.0
• Deployment type (BOSH/Docker/binary): docker-compose
• Infrastructure/IaaS: bare-metal server
• OS: ubuntu 16.04