We just did some validation on some new PVE Nodes (and related switching) for how they handle total power loss.
Everything seemed to come up just fine, except on both of the nodes pvescheduler tries to start up, and times out (after 2 minutes?) then never tries to start back up again.
I believe this is probably because the VMs were delayed in their auto-start, or other aspects, due to the related switches still initialising themselves.
I'm thinking pvescheduler would benefit from some configuration changes so it retries multiple times (maybe even until it succeeds?), as telling the daemon on each node to start up manually worked without any issue once all the things were done initialising.
What do you folks think? Good idea? Bad idea?
Everything seemed to come up just fine, except on both of the nodes pvescheduler tries to start up, and times out (after 2 minutes?) then never tries to start back up again.
I believe this is probably because the VMs were delayed in their auto-start, or other aspects, due to the related switches still initialising themselves.
I'm thinking pvescheduler would benefit from some configuration changes so it retries multiple times (maybe even until it succeeds?), as telling the daemon on each node to start up manually worked without any issue once all the things were done initialising.
What do you folks think? Good idea? Bad idea?