Fence Issues

A

adamb

Famous Member
Proxmox Subscriber
Mar 1, 2012
1,329
77
113
Just setup a new cluster. Having issues with fencing on one node after a reboot. After I reboot node #2 it will come up and try to fence node #1

1358439064 delay post_join_delay 6 quorate_from_last_update 1
1358439070 delay of 6s leaves 1 victims
1358439070 medprox1 not a cluster member after 6 sec post_join_delay
1358439070 fencing node medprox1
1358439090 fence medprox1 dev 0.0 agent fence_ipmilan result: error from agent
1358439090 fence medprox1 failed
1358439093 fencing node medprox1
1358439114 fence medprox1 dev 0.0 agent fence_ipmilan result: error from agent
1358439114 fence medprox1 failed
1358439117 fencing node medprox1

After a few minutes I then get this on node #2.

root@medprox2:~# fence_tool dump
fence_tool: dump: can't communicate with fenced

root@medprox2:~# fence_tool join
fence_tool: can't connect to cman

It seems the only way around this issue is to reboot node #1, then once that is up, reboot node #2 and all is well. If I reboot node #2 this process happens again. I do have FENCE_JOIN="yes" uncommented in /etc/default/redhat-cluster-pve on each node. I have two other nodes in the identical setup without this issue.
 
Last edited:
you run a two node cluster? if yes, I would be nervous two.

how do you make sure that you do not lose quorum on reboot?
 
tom said:
you run a two node cluster? if yes, I would be nervous two.

how do you make sure that you do not lose quorum on reboot?
Click to expand...

I have been running 2 node clusters for a bit over a year without any issues. I guess I should look into adding quorum disks to each one of my clusters. Im sure I could use a single node to provide quorum to all 3 clusters? I appreciate the input!
 
Just setup a cheap server with FreeNAS, and export iscsi targets for each cluster. It will take you 30 minutes from start to finish and give you a much safer cluster environment.
 
You must log in or register to reply here.
Top Bottom