What happened:

• infrastructure needed an update to deploy SSH keys, which ran the infrastructure deployment. That is the server configs, all server config settings are applied at that time
• part of server configs, is server updates, those were run
• one server update, is to a system (a service) called 'docker', docker is currently running all of the tripleA applications
• a config setting to automatically restart services when they are updated, or any of their dependencies are updated is live in the system
• docker was updated, per the config setting, it was full restarted.
• as a result, all tripleA systems were rebooted (including bots), which would have ejected players from bots.

With the reboot, some bots may have come back up, I expect for some to crash again over time and would still be interested in error reports.

Since then, docker is now on a blacklist for services to be restarted and a secondary config was added that would allow applications running on the docker service to potential survive a reboot of docker as well. 3rd, there is a PR up to autosave when a bot shuts down.

Historically bot restarts had been for more direct causes. This was more "rare" as there are typically no package updates for docker.

I'm taking a quick look at the autosaves to see what the history is there.