If you have non-clustered LVM volumes managed by RHCS (so called HA-LVM), you should have configured /etc/lvm/lvm.conf to have a host tag in the activation/volume_list configuration entry (as per the fine documentation). This stops the RHCS managed LVM VGs and LVs from being activated (and therefore becoming accessible) during normal operation on nodes that RHCS doesn’t think should have access. This is a safety feature to prevent a non-clustered filesystem from being mounted on two different hosts at the same time.
However, this can be a pain in some cases, usually when RHCS is down and you need to do some administrative work on the data in the LVM’s LVs. Changing the /etc/lvm/lvm.conf to remove the volume_list configuration is a pain because RHCS checks that it’s modification time is older than the systems initrd (to make sure you rebuild the initrd so it includes the new lvm.conf so when the system reboots it comes up with the right LVM configuration). There are two ways around this; first manually add and then delete the tag:
vgchange --addtag @$(uname -n) $VGNAME # Do the work here vgchange --deltag @$(uname -n) $VGNAME
Or secondly (and my favourite), override the LVM configuration on the command-line:
vgchange -ay $VGNAME --config "activation { volume_list = [ '$VGNAME' ] }"
Deactivating the VG doesn’t require the --config argument:
vgchange -an $VGNAME