vgchange -a s hangs in a cluster node

[José Enrique González]

Hi:

I am doing some tests to activate shared VG's in a cluster stand-by node. After some sucessfully vgchange -a s, vgdisplay and vgchange -a n to check operation, suddenly a vgchange -a s hangs with no obvious reason. No syslog neither dmesg neither any other message. Any subsecuent vg command hangs and server must be rebooted.

Something that catched my attention is the /etc/lvmconf/lvm_lock file. It has some entries with the text "Bad file number", and 000 file permission. Once hanging vgchange starts it hold the file locked.

All VG's are created in a XP512 disk array, and are active and perfectly working un the other two nodes of the cluster. CU/LDev numbers were carefully checked on working nodes and so disk device files paths in the failling node.

Any sugestion, in addition to checking LMV patches? Thanks in advance for your help.

[Mark Greene_1]

From the man page for vgchange:

"Because of these limitations, when sharing volume groups between
systems it is recommended that logical volumes be accessed only by one
system at a time. If logical volumes need to be accessed
simultaneously, the logical volumes should not be mirrored and should
not have bad-block relocation turned on, or all systems should use
read-only access to the logical volumes."

Are the LVs withing the VG in question mirrored and/or have bad block allocation set?

mark

[Steven E. Protter]

I think LVM patches are always a good idea. But if you have two machines trying to own the same physical disks at the same time, you are in for trouble.

ServiceGuard High availability works on a couple of simple principles. The most basic is the application fails over and its volume group can be activated on the failover node.

You should not have two nodes active on the same disk at the same time.

SEP

[melvyn burnard]

Are you using just Servicegiard? or do you also have SGeRAC or Serviceguard OPS Edition?

You can only use the vgchange -a s option if you have one of the latter versions, not standard Serviceguard