As it is running as

(In reply to comment #1)

As it is running as

I meant to continue "a grid job" and write about continuous jobs on Tools that are all peachy, before I noticed that continuous jobs are only restarted when they exit with failure which may or may not be the case for morebots.

Stashbot runs as a Kubernetes deployment which is similar to a continuous grid job, but more likely to restart if the process dies completely. It also implements a ping/pong check for irc connection health which should prevent most hung irc connection issues (rLTSTe49aed30c052: Ping every 5 minutes and disconnect if pongs aren't received).

It is still possible for the python run loop to hang somehow without the Kubernetes pod dying. This could be mitigated by adding some logic in the bot itself to touch a heartbeat file periodically and adding a matching liveness probe to the Kubernetes manifest.