I've seen a few user server pods in a z2jh base deployment stay running and fail to be culled, but removed from jupyterhubs database - and with that, additional attempts aren't made to stop the user server.
In practice, this may be more of a jupyterhub issue - as I think what idle-culler does is that it asks jupyterhub to stop the server and doesn't do it itself, so maybe this is more of a bug about "failing to stop but still declaring the server stoppped" kind of issue?
I'm not able to dig into this at length and resolve it, but wanted to leave this as a trace for now at least.
I've seen a few user server pods in a z2jh base deployment stay running and fail to be culled, but removed from jupyterhubs database - and with that, additional attempts aren't made to stop the user server.
In practice, this may be more of a jupyterhub issue - as I think what idle-culler does is that it asks jupyterhub to stop the server and doesn't do it itself, so maybe this is more of a bug about "failing to stop but still declaring the server stoppped" kind of issue?
I'm not able to dig into this at length and resolve it, but wanted to leave this as a trace for now at least.