Replacing a node after cluster setup

[stuntshell]

Hi all,
I've set up a cluster with 03 servers and setup raid0 with the disks.
On one of the nodes I had a disk failure, no sweat this is just a test setup.
So I replaced the disk and reinstalled, then removed the node from the remaining with

pvecm delnode node3

.

Now the question, I remember that with 1.9 I needed to have the nodes clean, no VMs on both nodes to make a cluster.
In my case I'm running a couple of VMs on the two nodes and I have a clean one I want to add.
Is this still the case, or can I just add it via CLI?

Thanks,

 

[dietmar]

Only the node you want to add needs to be clean.

 

[stuntshell]

I´m getting this error when I try to add:
authentication key already exists
By searching the forums, I see a lot of replies about having to recreate the cluster.
Is there an easy fix?

Thanks,

 

[dietmar]

I´m getting this error when I try to add:
authentication key already exists

You are adding a newly created node (not an old one)?

 

[stuntshell]

Yes, a newly created node, but naming it like the old one. Should this be the problem?

Thanks,

 

[dietmar]

Yes, a newly created node, but naming it like the old one. Should this be the problem?

This error is only shown if the file "/var/lib/pve-cluster/corosync.authkey" exists, and I am 100% sure that file does not exists on a new node.

Try to run 'pvecm add ...' using the '--force' flag. And please run that on the new node!

 

[stuntshell]

Right on!
This has worked for me.
I had to modify the /root/.ssh/known_hosts and /etc/ssh/ssh_known_hosts to make migration work again.

Thanks,