apcupsd graceful guest shutdown on HA/CEPH cluster

[b.miller]

Hi all. I am soon deploying a three node cluster with HA guests & CEPH storage. I will be setting up apcupsd to initiate graceful shutdown, but I am unsure of the appropriate shutdown scripting when HA & CEPH are in play. I have a few questions and would greatly appreciate any tips:

1. What command is currently the preferred method to shutdown all guests on a node? Can this be done at the node level, or do individual commands need to be issued per guest?
-Is this still a viable script in the current version of PVE? Will it suit HA? https://forum.proxmox.com/threads/clean-scheduled-reboot.38386/post-267307

2. Will Prox's watchdog trigger a fencing event?

3. With appropriate bios options enable, can Prox auto-start and resume all guest VMs after power is restored without manual intervention?

4. Does CEPH require separate configuration or commands via apcupsd to safely shutdown in tandem with guest shutdown?

Thanks so much.

 

[b.miller]

Update.

I am no longer using APCUPSD - I'm using NUT instead.

I know that Prox is smart enough that a fsd shutdown command will shutdown all guests before the hosts.

But does this apply to a Prox cluster running CEPH or does the shutdown script in NUT need to 'noout' and 'norebalance' every single OSD?