Remove failed node from cluster

[FloridaMan]

Hello I had a node fail and I'm trying to remove it from my cluster. when I look at pvecm however it's not there. I tried removing it from the /etc/pve/nodes directory but the only thing that did was remove the join information from my cluster. (The failed node was the original node the others joined) I have a backup of the nodes directory so I put it back and that seems to have restored the join information. When copying back the information to the nodes directory, I received an error it could not restore /etc/pve/nodes/dchv3/qemu-server/100.conf etc.. because it already exists. So I'm guessing it didn't really delete that directory even though I check to see if was gone with ls.
What would be the proper procedure to remove a failed node that's already missing from pvecm ?

root@dchv1:~# pvecm status
Cluster information
-------------------
Name: PVE
Config Version: 3
Transport: knet
Secure auth: on

Quorum information
------------------
Date: Wed Mar 24 11:16:25 2021
Quorum provider: corosync_votequorum
Nodes: 2
Node ID: 0x00000003
Ring ID: 2.16f
Quorate: Yes

Votequorum information
----------------------
Expected votes: 3
Highest expected: 3
Total votes: 2
Quorum: 2
Flags: Quorate

Membership information
----------------------
Nodeid Votes Name
0x00000002 1 192.168.130.32
0x00000003 1 192.168.130.31 (local)

I have noticed that the node still exists in the /etc/pve/corocsync.conf file

 

[FloridaMan]

I posted too soon. I called pvecm del node (dead node name) even though it was not in the pvecm list and it worked. It removed the node from the corosync file. After that I removed the /etc/pve/nodes/(dead node name) and that got rid of it. This time it kept my cluster join information so the cluster is still intact.

 

[dustojnikhummer]

I posted too soon. I called pvecm del node (dead node name) even though it was not in the pvecm list and it worked. It removed the node from the corosync file. After that I removed the /etc/pve/nodes/(dead node name) and that got rid of it. This time it kept my cluster join information so the cluster is still intact.

I would add that I since I had only two nodes I also had to run "pvecm expected 1" to actually delete the node (since with default 2 both would have to be online to vote on the removal) https://forum.proxmox.com/threads/r...cluster-not-ready-no-quorum.23622/post-118347