[Linux-cluster] fence_manual...

CAugustine at overlandstorage.com CAugustine at overlandstorage.com
Mon Apr 18 19:56:05 UTC 2005


Hi Everyone,

I have a two-node cluster. I have built and installed the cluster sources
from cluster_0406282100 snapshot. I can bring up both nodes successfully,
however, seems like I have a brain split cluster in that each node thinks 
it
is the only node. I ran the cman_tool on each node as follows:

        cman_tool join -c OVLCluster -2 -n nodename

Furthermore, in the log messages I often see messages that require one 
of the nodes to be rebooted. Some times I see the message in both nodes'
/var/log/messages files. In this case, I reboot the node that needs to be
rebooted and run the "fence_ack_maual -s rebooted-nodename" on the 
other system after the reboot. The problem is that then I see the same 
messages
again on the rebooted node's messages file. Seems like the cluster is in
some kind of a loop wanting to reboot that node over and over again after
the reboot. 

Can anyone tell me what is going on? Also, I am running the ccsd, 
cman_tool,
 fence_tool, clvmd, vgchange commands by hand. What version of the 
clustering
software has nice scripts such as "cluster start"?

Thanks,
Caroline

----------------------------------------------------------------------------------------------
See our award-winning line of tape and disk-based
backup & recovery solutions at http://www.overlandstorage.com
----------------------------------------------------------------------------------------------

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20050418/ba9a8680/attachment.htm>


More information about the Linux-cluster mailing list