[Linux-cluster] BladeCenter Fencing errors

jim parsons jparsons at redhat.com
Thu Jan 17 22:40:13 UTC 2008


On Thu, 2008-01-17 at 14:06 -0700, Gary Romo wrote:
> 
> I enabled telnet on the MM, now I am getting these messsages; 
> 
> Jan 17 14:00:24 node1 fenced[3229]: fence "node2" failed 
> Jan 17 14:00:29 node1 fenced[3229]: fencing node "node2" 
> Jan 17 14:00:40 node1 fenced[3229]: agent "fence_bladecenter" reports:
> pattern match timed-out at /sbin/fence_bladecenter line 189  
> 
> Jan 17 14:00:40 node1 fenced[3229]: fence "node2" failed 
> Jan 17 14:00:45 node1 fenced[3229]: fencing node "node2" 
> Jan 17 14:00:56 node1 fenced[3229]: agent "fence_bladecenter" reports:
> pattern match timed-out at /sbin/fence_bladecenter line 189  
> 
> Jan 17 14:00:56 node1 fenced[3229]: fence "node2" failed 
> Jan 17 14:01:01 node1 fenced[3229]: fencing node "node2" 
> Jan 17 14:01:12 node1 fenced[3229]: agent "fence_bladecenter" reports:
> pattern match timed-out at /sbin/fence_bladecenter line 189  
> 
> Line 189 looks like this; 
> 
>  ($text, $match) = $t->waitfor("/system:blade\\[$bladenum\\]>/"); 
> 
> 
> I am getting these on thesecond node; 
> 
> Jan 17 14:03:24 mode2 fenced[3340]: fence "node1" failed 
> Jan 17 14:03:29 node2 fenced[3340]: fencing node "node1" 
> Jan 17 14:03:29 node2 fenced[3340]: fence "node1" failed 
> Jan 17 14:03:34 node2 fenced[3340]: fencing node "node1" 
> Jan 17 14:03:34 node2 fenced[3340]: fence "node1" failed 
> 
Ah, yuck. Well, let's figure out what is going on here.
Can you post the clusternodes and fencedevices sections of your
cluster.conf here? Just XXXX out any passwords.

On one of the cluster nodes, can you run 
'/sbin/fence_bladecenter -a <ip or hostname of bladecenter> -l <login>
-p <passwd> -n <blade number of another running node> -o status -v'

Do you know firmware details about your bladecenter? The
fence_bladecenter script hasn't changed in years...The tested firmware
versions are in the top of the file. Maybe the interface has changed. If
so, the debuglog should give us information.

This will get us started.

-Jim




More information about the Linux-cluster mailing list