[Linux-cluster] gfs2 mount hangs

[David Teigland]

On Thu, Aug 13, 2009 at 02:22:11PM +0800, Wengang Wang wrote:
> <cman two_node="1" expected_votes="2"/>

That's not a valid combination, two_node="1" requires expected_votes="1".

You didn't mention which userspace cluster version/release you're using, or
include any status about the cluster.  Before trying to mount gfs on either
node, collect from both nodes:

  cman_tool status
  cman_tool nodes
  group_tool

Then mount on the first node and collect the same information, then try
mounting on the second node, collect the same information, and look for any
errors in /var/log/messages.

Since you're using new kernels, you need to be using the cluster 3.0 userspace
code.  You're using the old manual fencing config.  There is no more
fence_manual; the new way to configure manual fencing is to not configure any
fencing at all.  So, your cluster.conf should look like this:

<?xml version="1.0"?>
<cluster name="testgfs2" config_version="1">
<cman two_node="1" expected_votes="1"/>
<clusternodes>
<clusternode name="cool" nodeid="1"/>
<clusternode name="desk" nodeid="2"/>
</clusternodes>
</cluster>

Dave