What happens ...

[skraw]

... when I have a cluster with 4 nodes, where one is down for some reason. Then I add a new node which runs perfect. Then I restart the node that was down during adding the new node. Does this work, or is it a problem in some way?
The same btw might happen if you have to restore some node from a backup, where the cluster was in a different node-setup (which means the backup does not reflect the latest nodes). Is this identical to above, or is there some additional problem because now a node that already had the information comes back with some older config.
--
Regards

 

[UdoB]

... when I have a cluster with 4 nodes, where one is down for some reason. Then I add a new node which runs perfect. Then I restart the node that was down during adding the new node. Does this work, or is it a problem in some way?

In my experience it is highly recommended to turn on all nodes before adding a new one. If one node is down then it will not get the new configuration automagically when turned back on. Been there, done that - I had to "repair"/update corosync on that node manually...

(Just my 2¢, not a authoritative answer.)

 

[skraw]

Is there nothing else to update besides the corosync.conf?

 

[UdoB]

Is there nothing else to update besides the corosync.conf?

Sorry, I really can't remember. (Probably "pvecm updatecerts" was required also to distribute fingerprints back to the problematic node.)

 

[skraw]

It just came to my mind that the whole issue is quite simple to solve: all needed is somebody having daily snapshots of the proxmox repositories and one would easily be able to update to any given date (version).
Is there someone out there having such an archive?