Full remove Cluster feature

[plastilin]

Hello All. I try to setup cluster with 2 nodes, but it doesn`t work properly. I remove second node from cluster with command pvecm delnode server2
Now i have error on cman when servers start. How i can full removwe cluster feateru and reinstall it without reinstalling physical servers? Thanks.

 

[dietmar]

I suggest you reinstall the node.

 

[Aksonx2]

This helped me

First, make a backup of the cluster:


cp -a /etc/pve /root/pve_backup
Stop cluster service:


/etc/init.d/pve-cluster stop
Umount /etc/pve if it is mounted:


umount /etc/pve
Stop corosync service:


/etc/init.d/cman stop
Remove cluster configuration:


# rm /etc/cluster/cluster.conf
# rm -rf /var/lib/pve-cluster/*
Start again cluster service:


/etc/init.d/pve-cluster start
Now, you can create new cluster:


# pvecm create newcluster

Restore cluster and virtual machines configuration from the backup:
# cp /root/pve_backup/*.cfg /etc/pve/
# cp /root/pve_backup/qemu-server/*.conf /etc/pve/qemu-server/
# cp /root/pve_backup/openvz/* /etc/pve/openvz/

 

[plastilin]

Thanks for replies. My solution for use node with reinstall is "pvecm e 1" on last cluster node. But big thanks for Aksonx2 for his solution.

 

[Tony]

Hi,


I did the following:


- installed Proxmox VE as per https://pve.proxmox.com/wiki/Proxmox_VE_2.0_Cluster
- after running 'pvecm create cluster' on 1st node, the command finished without error. however I checked and see that node has IP 10.0.0.1 (assigned to drbd link). Problem with /etc/hosts, as I found out after searching the forum. fixed.
- hence I need to re-create the cluster. Following various instructions found in the forum, here is the most recent attempt:


# remove all relevant packages: ie all packages that got installed when I did 'aptitude install proxmox-ve-2.6.32'
# remove the following dirs:
/var/lib/pve-cluster
/var/lib/pve-manager
/etc/cluster
/var/lib/cluster/
/var/run/cluster
/var/log/cluster
/etc/pve


# rerun 'aptitude install proxmox-ve-2.6.32'
# rerun 'pcevm create cluster'


But this time cman didn't start; it says:
"Starting cman... Timed-out waiting for cluster Check cluster logs for details"


When I checke log file, there is a lot of error messages like this:
Oct 22 22:57:32 corosync [TOTEM ] Totem is unable to form a cluster because of an operating system or network fault. The most common cause of this message is that the local firewall is configured improperly.


So I tried to turn iptables completely off, but the result is still the same.


I think the last choice is to reinstall the node completely from scratch.
Before doing that, I wonder if I need 2 nodes up when I run 'pvecm create
cluster'? Or it's ok to setup 1 node completely, and add the 2nd node later.

btw my nodes ar hosted, and multicast is not allowed.


thanks,
Tony

 

[udo]

...
btw my nodes ar hosted, and multicast is not allowed.
...

Hi Tony,
you have switched to unicast? https://pve.proxmox.com/wiki/Multicast_notes

Udo

 

[Tony]

Hi Udo,

sorry for the late reply. Yes I have switched to unicast but it didn't help.

I am going to reinstall everything from scratch then.

Tony