cmgetconf and local vg

[Timo J]

2-node cluster named foobar, nodes nodeA & nodeB. When running 'cmgetconf -v -c foobar', it prints warnings:

"Error: Volume group /dev/vg00 on node nodeB does not appear to have a physical volume corresponding to /dev/dsk/c1t15d0 on node nodeA (5157705581148388940)."

Why cluster thinks as an error that situation where _local_ vg configuration differ between nodes?

cmgetconf also reports that "Error: Volume group /dev/vgfoo has mismatched cluster ids." That is shared volume which got one of it's disks changed few days ago. Looks like engineer forgot to update the cluster configuration after disk change or what?

[Denver Osborn]

Any chance that nodeB cloned from nodeA w/out using Ignite-UX?

Check the VGID of nodeA and nodeB's root disk.

# vgexport -p -v -s -m /tmp/vg00.map vg00
# grep VGID /tmp/vg00.map

Compare the VGID from both node's root disk. I'm betting they're the same.

You'll need to reinstall one of the nodes or use Ignite-UX. I'd use ignite.

For the 2nd part of your question, I think you can take care of it w/ a simple "vgchange -c n".

hope this helps,
-denver

[Stephen Doud]

For the first error, it's likely that /etc/lvmtab do not match content for vg00. Normally, cmgetconf doesn't report such an error for private VGs unless vg00 was cloned using a command like 'dd'. This would make both vg00's have the same VGID (LVM metadata). If this was done, use UXSGKBRC00007573 to resolve the condition.

As for the 2nd error, vgfoo should not have mismatched cluster IDs unless you have two different vgfoo's. Suggest you create a map file for both vgfoo's - one on each system:
# vgexport -pvs -m vgfoo.map /dev/vgfoo

Compare the VGID at the top of the file - if they differ, then vgfoo is not the same vg on each node. If this is the case, Serviceguard won't work with vgfoo properly.
Map the disk paths attached to each server and compare vgfoo's disks between nodes (strings /etc/lvmtab). If they differ, then you have 2 different VGs named vgfoo. Don't do that :)
vgexport one of the vgfoo VGs on one of the nodes and reimport it to a different name.
Then vgimport vgfoo using the map file created on the other node:
# mkdir /dev/vgfoo
# mknod /dev/vgfoo/group c 64 0x0n0000 (where n = a unique minor number)
# rcp otherhost: .
# vgimport -vs -m vgfoo.map /dev/vgfoo

If you have trouble - contact HP Support for guidance.