Newly installed hypervisor restart when quorum is lost

[criva]

we had already three instances when a hypervsor installed with proxmox 8.1.0 and corosync 3.1.7 restarted when one of the other hypervisors became unavailable.

Is this something that can be a regression of a behaviour we have seen in the past?


At the moment we are worried of the effect that removing one older element of the cluster could have on the rest of the cluster.

 

[UdoB]

Check and verify the corosync setup and have an eye on pvecm status.

Guessing: the reboot you noticed is called "fencing" and it happens when the Quorum is not reached and HA is enabled.

Several people have been surprised by this behavior, but usually it works as intended. That topic has been discussed several times - search for it ;-)

 

[Maximiliano]

Do note that node fencing only happens when there are HA resources running on the node. As it was already mentioned this is intended behavior. For more info see [1].

How many nodes are in your cluster? Please attach the output of `pvecm status`.

[1] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#ha_manager_fencing

 

[criva]

Indeed it is fencing and I suspected it may be related to the HA, thanks a lot for the reply and the references.