Delete node - Node IP still in corosync.conf in totem section?

[dusty128]

Hi,
I have a little Proxmox 7.4 cluster I set up some time ago. Now my fist node died and I deleted it from the cluster. I plan to freshly install the node and add it to the cluster, but with the same name / IP address.

For that reason I checked every file in the /etc directory if there are any remains of the old hostname / ip address.

Interestingly, I still find the old IP address in corosync.conf (both in /etc/pve and /etc/corosync). The node is gone, but it is still in the totem section:

totem {
cluster_name: my_cluster
config_version: 11
interface {
bindnetaddr: 192.168.0.21
ringnumber: 0
}
ip_version: ipv4
secauth: on
version: 2
}

"192.168.0.21" is the IP address of the deleted node and is therefore unreachable.

Does anyone know why this (inexisten) IP address ist still in corosync.conf? Should I alter this IP address to some other Node IP?

Is there anything else to think about when I add the new node with the old hostname / IP Address? Maybe there are some other remains in /var/lib/pve-cluster/ I should be aware of?

Best Regards,
Hermann

 

[waltar]

What's the full /etc/pve/corosync.conf ?
Normally first remove any (lxc+vm) ha configs, then remove old node there, restart corosync (rewrites /etc/corosync/corosync.conf), then reconfig ha settings. Later reinstall your old pve node and rejoin it.

 

[dusty128]

Thanks for the quick reply! Unfortunately my issue was that the node had a single SSD as root disk and this SSD died. So there was no way to remove the node while still running. And unfortunately, I have only the /etc directory of the dead node backed up and not /var/lib, so there is no way to reconstruct it.

I don't use HA at all as I am unsure if it would work properly (split brain scenarios etc.).

My full corosync.conf looks like this:

---------- snip ------------
logging {
debug: off
to_syslog: yes
}

nodelist {
node {
name: node2
nodeid: 2
quorum_votes: 1
ring0_addr: 192.168.0.22
}
node {
name: node3
nodeid: 3
quorum_votes: 1
ring0_addr: 192.168.0.23
}
node {
name: node4
nodeid: 4
quorum_votes: 1
ring0_addr: 192.168.0.24
}
node {
name: node5
nodeid: 5
quorum_votes: 1
ring0_addr: 192.168.0.25
}
node {
name: node6
nodeid: 6
quorum_votes: 1
ring0_addr: 192.168.0.26
}
node {
name: node7
nodeid: 7
quorum_votes: 1
ring0_addr: 192.168.0.27
}
node {
name: node8
nodeid: 8
quorum_votes: 1
ring0_addr: 192.168.0.28
}
}

quorum {
provider: corosync_votequorum
}

totem {
cluster_name: pmx1
config_version: 11
interface {
bindnetaddr: 192.168.0.21
ringnumber: 0
}
ip_version: ipv4
secauth: on
version: 2
}

---------- snip ------------

 

[waltar]

Looks nearly like ours but the end is different:
...
totem {
cluster_name: pvelab
config_version: 6
interface {
linknumber: 0
}
ip_version: ipv4-6
link_mode: passive
secauth: on
version: 2
}

So I would remove your line "bindnetaddr" and maybe exchange "ringnumber: 0" to "linknumber: 0", we have link_mode passive also.
One very IMPORTANT thing I forgot - you MUST do your changes, then switch your line "config_version: <+1>" - then save ( not save earlier) !!!

 

[Hermann Himmelbauer]

Thank you for your reply, I altered my config as suggested (but the /etc/pve/corosync.conf) and it seems the cluster keeps running flawlessly.

So, problem is solved.