[Linux-cluster] cluster.conf fails to validate

Darrin De Groot d.degroot at griffith.edu.au
Fri Jun 20 00:45:17 UTC 2008


I have noticed recently that my RHEL5 cluster can no longer be managed 
through the luci interface - I see the following error:

An error occurred while attempting to get status information for this 
cluster. The information shown may be stale or inaccurate.

The odd thing is that I *can* actually update the cluster.conf on all the 
nodes through this interface...

I tried to validate the cluster.conf file, thinking that might have 
something to do with it:

xmllint   --relaxng    /usr/share/system-config-cluster/misc/cluster.ng 

The trailing portion of which showed this:

        <quorumd device="/dev/dm-1" interval="2" min_score="1" tko="20" 
Relax-NG validity error : Extra element quorumd in interleave
/etc/cluster/cluster.conf:2: element cluster: Relax-NG validity error : 
Element cluster failed to validate content
/etc/cluster/cluster.conf fails to validate

Is this relevant?

Does anyone have any ideas? Command line tools (clustat, cman_tool 
services|status) all work just fine. The only reason I noticed was because 
one of the nodes didn't reboot correctly (dlm took too long to start on 
boot, causing the node to not join the fence domain, and therefore all 
other cluster services to fail - the nodes just sit there "Starting 
fencing"). If anyone has any ideas about this I'd apreciate too.

One other small question - is there any way to query which node is the 
current qdisk master?


Darrin de Groot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20080620/03d90064/attachment.htm>

More information about the Linux-cluster mailing list