[Linux-cluster] Problem with more than one partition

[Robert Peterson]

aleava at libero.it wrote:
> Hi Bob,
> thanks for your help, but I have installed RHEL4 Update 4 AS, and the cluster doesn't work (the cluster.conf and the commands are the same).In the same manner, one node mount the shared partitions (4) without problems, but when the second node mounts the first partition, the mount command hangs.
> P.S.: if a node mounts only one partition, the second node mounts the same partition without problems. We have problems with a number of partition =>2.
> Any ideas?
>
> (At the moment, our solution is creating only a big shared partition with 4 directories, using quota...)
>
>
> Regards,
> Ale 
>   
Hi Ale,

This is odd.  I'm assuming that you rebooted all the cluster nodes after you
upgraded to U4, or at least brought down the whole cluster first.
Here are some things you can do to help us solve the problem:

(1) Please post the output from:

gfs_tool sb /dev/your_vg/lvol0 all

for both of the file systems you're trying to mount, substituting your 
vg and lv or
where your gfs file systems are located.

(2) Check for any unusual error messages posted in /var/log/messages 
right after
you recreate the hang.

(3)  Check for any unusual dmesg messages on both systems right after you
recreate the hang.

(4) Do this command and post the output, right after the hang:
cman_tool services

(5) Are you able to ctrl-c break out of the mount that hangs or do you 
have to reboot?

(6) If you can't break out of mount, on the system that has its mount 
hung, do this:
echo "t" > /proc/sysrq-trigger
and post the output so we can see where it's hanging.

We'll get to the bottom of this mystery!  Unfortunately, I will be out 
of the office next
week, along with most of the other Red Hat cluster developers.  So if I 
don't
hear back today or tomorrow, I won't be able to respond until the 2nd of 
January.

Regards,

Bob Peterson
Red Hat Cluster Suite