Update example docker compose so services restart on crash

[tobiasmuehl]

Restic backup can fail for a number of reasons. If the process crashes once, it should be restarted, so once the underlying failure is resolved (B2 API downtime, internet connectivity issues, etc) backups will resume automatically.

[djmaze]

As far as I can see, the container should never crash because everything is wrapped in a cron scheduler process. Does the container crash for you?

[tobiasmuehl]

This happened a long time ago. I don't recall if the container crashed fully, but I assume so. The failure had to do with a server error at B2

[djmaze]

Would be good if anyone can confirm that the container can crash. (And I would really like to know how that can happen.)

As I don't think the container will crash, the restart option won't help with your problem either way.

[iwa]

either can the container crash or not is not the question here imo, having a restart policy is a really good docker practice, and can be useful in other cases like a restart or else, not only to prevent crashes (and even if they are rare, they can still happen)
but the unless-stopped policy should be a better choice, always would trap people into having restic always on 24/7, and their only way to shutdown it would be to delete the container, so it's not ideal

[djmaze]

Okay, let's do that. I also trapped on restart: always quite a few times, so @tobiasmuehl please change it to unless-stopped.

[tobiasmuehl]

done