Mailcow dies on reboot

nfsmith

Hey there,

I run mailcow on a Fedora server and have automated updates running. I find that every time the server reboots to install updates mailcow ends up in a failed state on reboot. Is this expected behaviour? Does anyone have any tips on how to find the root cause or strategies to improve this?

Thanks!

diekuh

Hi, This does normally not happen. Any information would help people to help you. Like what does a died state mean? Which exit codes?

nfsmith

Strangely, the web UI and all services seemed to be down, but the Docker containers all seemed to be running fine. That said I can’t seem to reproduce the issue anymore. Makes me think… Perhaps my own IP got fail2bain-ed? I could SSH into the box, but all other services seemed unreachable

MAGIC

You could check the netfilter logs with docker-compose logs netfilter-mailcow, maybe you will find an entry with your IP

nfsmith

Ha! Found the problem. By default Fedora uses socket activation to start the docker service. My server would reboot, nothing would touch the docker socket and so the docker service wasn’t running. As soon as I ran any docker command the service would come up.

Easy fix: systemctl enable docker.service