undo create cluster

[gijsbert]

I was playing with some settings and created a cluster doing:

===
root@prox1:~# pvecm create test-cluster-01

Corosync Cluster Engine Authentication key generator.
Gathering 1024 bits for key from /dev/urandom.
Writing corosync key to /etc/corosync/authkey.
Writing corosync config to /etc/pve/corosync.conf
Restart corosync and cluster filesystem
===

After this creation of the cluster I did not configure anything else. Is there an easy/safe way to undo this creation of the cluster, something like "pvm delete test-cluster-01" ? I cannot find something like this with pvecm --help

Any help will be appreciated.

Gijsbert

 

[t.lamprecht]

For the case were you really do not have other nodes, the cluster is healthy, and nothing was done I have a proven set of steps which destroy this simple cluster.
Do not change order, do not forget steps, all people which have done more or have multiple nodes, please stop here and read: https://pve.proxmox.com/pve-docs/chapter-pvecm.html#pvecm_separate_node_without_reinstall

# WARNING: only do on simple stand-alone clusters which are healthy and without HA!
rm -f /etc/pve/corosync.conf
rm -rf /etc/corosync/*
systemctl stop corosync pve-cluster
systemctl start pve-cluster

But also just as a note, if you want to keep the node standalone its "cluster" status does not hurt.

 

[gijsbert]

Dear T. Lamprecht,

Thanks for your reply. It was a simple stand-alone and healty cluster so I removed the corosync.conf and files like described. Seem to work, cluster status is gone and everything still ok.

One more question though. What happens when I reboot the server, will it start the corosync service? And if so, is that normal behaviour?

Thanks again !

Gijsbert

 

[t.lamprecht]

One more question though. What happens when I reboot the server, will it start the corosync service? And if so, is that normal behaviour?

No, if no "/etc/corosync/corosync.conf" exists, the corosync service will not be started, see

systemctl cat corosync.service

there are two conditions where the corosync service will not be started:
* ConditionKernelCommandLine=!nocluster - if you boot the node with a "nocluster" kernel commandline argument (useful when troubleshooting a cluster issue)
* ConditionPathExists=/etc/corosync/corosync.conf - if no corosync.conf exists there's no need for corosync to run.

So if you reboot now it won't run.

 

[gijsbert]

Thanks for this great (and fast) response. proxmox rocks