I'm wondering if that didn't start this whole thing off...
If you want to test it, get all your volumes created, shutdown one
machine, disable it's FC ports and restart it. Was clvmd barfing
because of this? Could it see some, but not all of the devices.
Would the fix have been as simple as simply reenabling the ports to all
the machines? I don't know...
I can't imagine your storage being trashed by the FC network. Even if
the logical volume headers were destroyed, they can be brought back by
vgcfgrestore. That would just leave the file systems. They have
journaling. So assuming some rouge process didn't mess with the disk,
they should be able to recover just fine as long as the storage (CLVM
volumes) are visible.
brassow
On Jan 24, 2007, at 1:29 PM, isplist@xxxxxxxxxxxx wrote:
Good point and that was indeed an issue. Some of the ports WERE
blocked. All
have since been cleared.
Mike
On Wed, 24 Jan 2007 13:28:28 -0600, Jonathan E Brassow wrote:
I would ensure that when the fencing occurred, that it didn't fence
off
the devices on the brocade switch. I would log into the brocade
switch
and ensure that all nodes are properly connected (and their ports are
not disabled).
brassow
On Jan 24, 2007, at 11:39 AM, isplist@xxxxxxxxxxxx wrote:
On Wed, 24 Jan 2007 11:15:28 -0600, Jonathan E Brassow wrote:
are you using fibre channel fencing in your cluster?
Yes, Brocade switches with Xyratex chassis.
Mike
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster