Updating Mailcow restarts docker daemon every time

insuna

Hey everyone. When I use the update.sh script mailcow always restarts the docker daemon, killing all of my other containers. Is this intended? If so: Why? This is causing a lot of friction and causes me to delay updating mailcow for as long as possible, especially because not long ago a mailcow update completely destroyed my entire /var/docker directory (every single file was corrupted and needed to replaced from a backup).

I went through the update.sh script and it seems to only restart the docker service in the migrate_docker_nat function, but why does that need to run on every single update?

Thank you for your insights!

pkernstock

It should not restart every time. Only once when confirming migration to native ipv6-nat implementation of docker. Can you post update.sh output to see what it does? Also, this function is planned to be rewritten and improved in a lot of ways.

insuna

It asks me to migrate every single time.
If this is not working as intended then it may be, because my alpine linux VM did not come with an /etc/docker path and update.sh writes into /etc/docker/daemon.json without checking if the path even exists. I’ve since created that directory, maybe that will fix this.
I can’t currently post the output of update.sh because I don’t want to be forced to restart all of my containers again.

Edit: I guess this is particularly annoying because I explicitly have enable_ipv6: false set in my docker-compose.yaml because I don’t want to bother dealing with constantly changing AAAA records due to not having a fixed ipv6 prefix.
One might think that means I don’t need the Mailcow v6 NAT stuff anyway?