Need help on the Cluster Manager

[cocobanana]

Hi folks!

I need help on the cluster configuration

Apparently, we have 6 Nodes and we do a clustering with proxmox.

Our 6 servers are same server configuration
- 6 x 600GB SAS (RAID 10 raid card with BBU ) for Promox OS
- I create a /dev/sda4 on each server.
- All servers are with same specs and configuration

WE have our 7th nodes and it is empty.

The configuration runs perfect and very-very well ( THANKS TO PROXMOX ) for more than months. But today, 3 of our server need to replace the motherboard asap. The VM runs very well and there is no issue on the VM part.

I did ran :-
- systemctl stop pve-cluster corosync
- pmxcfs -l
- rm -rf /etc/corosync/*
- rm -rf /etc/pve/corosync.conf
- killall pmxcfs
- systemctl start pve-cluster

and the cluster has been destroy.

Now all the nodes are running as standing alone server.

May I know what is the best method to do cluster for all 6 NODES without backup all the VMs and without restart all the server/vm ?

Please help.

Thanks

 

[dcsapak]

hi,

I did ran :-
- systemctl stop pve-cluster corosync
- pmxcfs -l
- rm -rf /etc/corosync/*
- rm -rf /etc/pve/corosync.conf
- killall pmxcfs
- systemctl start pve-cluster

and the cluster has been destroy.

why did you run these commands? normally changing components like mainboard etc. (aside from the root disk) does not require anything special (besides from adapting hardware specific stuff like disk paths/ nic names)

you probably could have simply changed the mainboard without changing anything regarding the cluster

May I know what is the best method to do cluster for all 6 NODES without backup all the VMs and without restart all the server/vm ?

if you have a backup of the corosync config, you could try restoring it to /etc/pve (on all nodes) and restart
(note that you should have backups either way!)

there are sadly no built-in ways to (safely) join nodes that have vms/containers already on them

 

[cocobanana]

Hi,

Can you suggest a best way on how can I try to join cluster for all of my nodes. Those VMs are running with Windows OS and rebooting will be a very troublesome for us.

Please suggest and help.

Thanks