Change Disk on a Cluster

[Sharon Bi]

Dear all,

We just found out there are two defective disks on a two-node cluster. One disk has a mirrored copy. The other one does not. So we are planning to replace those disks.

For the one which has a mirrored copy:

1) remove defective disk
2) Place in new disk
3) boot system in single user mode
4) vgcfgrestore -n /dev/vg08 /dev/rdsk/c4t10d0

** Will this command let the system mirror all the data from the mirrored copy to this new disk? Before I do a " vgcfgrestore ", do I need to do any backup?

For the one does not have a mirrored copy:
1) Backup all the data on the defective disk on tape
2) Remove the defective disk
3) Place the new disk in
4) *** How do I do with this new disk in LVM? Re - do " pvcreate, vgcreate"? How do I make the other node in the cluster recognize it? Before I take out the disk, shall I do any " vgexport" ?

Thank you very much for your help!!

Sharon

[Cheryl Griffin · Accepted solution]

Sharon,

There is a document in the ITRC Knowledge Base that has the steps outlined. See "Procedure for replacing an LVM disk in HP-UX 10.x and 11.x" DocId: KBAN00000347

[CHRIS_ANORUO]

Yes, you have to do vgcfgbackup and vgexport on both disk. Before then, back up all you files on to a tape. You can also use Ignite-UX make_recovery and make_net_recovery for this actions.

[Thomas Schler_1]

1st question:
You do not have to boot the system in single
user mode. Because it is a mirrored
filesystem, you just have to replace the
disk. Then perform a 'pvcreate' to make
the new disk available to the system. Then
perform 'vgcfgrestore -n vg08 '.
Now, you have to set up the mirror:
vgsync vg08. This should be all. No other
action on the other node.

2nd question:
I assume, the defective disk is a member
of an existing volume group with more than
one disk. Backup all logical volumes that
do have data on the defective disk (see
lvdisplay -v ). Remove the logical
volumes you have backed up before
(lvreduce). Reduce the volume group
by taking away the defectice disk: vgreduce
. Replace the disk.
Then use pvcreate to make the new disk
available to the system. Use 'vgextend
' to put the new disk into the
volume group. Use 'lvcreate' to recreate
the logical volumes you backed up before.
Use 'newfs' to create filesystems. Restore
the data from the tape.

The following procedure is necessary for
question 2 and also for question 1 if
the replacement has a different capacity:

On the host with mounted filesystems:
Export the volume group information to
a map file:
vgexport -p -s -m /tmp/vg08.map vg08

On the other host (without having actually
mounted the filesystems):
Remove the old volume group information:
vgexport vg08
Create a new group character file:
mkdir /dev/vg08
mknod /dev/vg08/group c 64 0x??0000
'??' must be the same as on the first node.
vgimport -m /nfs/firstnode/tmp/vg08.map
-s -v vg08

That's how I changed defective disks.