[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[dm-devel] RHEL-5.1-ia64 gfs/rhcs clvm pvcreate to wrong multipath device names



Hello Folks,

RE: RHEL-5.1-ia64 gfs/rhcs clvm pvcreate to wrong multipath device names.

I have a customer who did their pvcreate to the /dev/dm-n device names instead of the /dev/mapper/mpathN device names which is incorrect as per http://kbase.redhat.com/faq/FAQ_96_11196.shtm.

After some investigation, I believe this leads to cases where 'some" of the gfs file system built on the lvols can't be mounted in this cluster.

We are also seeing cases where a cluster node rhcs startup will stall shortly after starting cman on bootup or manually once the node is up.

The question is:  Is there a procedure to modify an existing clvm physical volume to replace the /dev/dm-n device name with the correct /dev/mapper/mpathN device name?

I have had this same question come up for customers who are migrating from a non-multipath or QLogic failover solution where they have specified /dev/sd[*] and now need to use the /dev/mapper/mpathN device name's.

My customer is looking for a solution that is less painful than to backup their data, purge and rebuild the lvm2 PV's, VG's, LV's & gfs and then restore their data.  In my search I have not found a solution, I am hoping I am just lacking in sufficient caffeine and that some one out here will have quick (obvious) answer that I missed.

Regards,
James Hofmeister
Hewlett Packard Linux Solutions Engineer


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]