Volume group recreation

[Terrence]

I have a simple two node active/passive cluster. I have one package running one application. There are two volume groups in the cluster; vg01 & vg02. vg01 is the important one which has the application. I need to remove and rebuild vg02 to accommodate larger pv's. vg02 will be identical after I am done, except it will have a larger pv's and larger max PE's per PV. All the logical volumes will be the exact same.

The application won't be affected by removing/recreating vg02. And there are no SG services/scripts which check for it's existence. Is it safe to rebuild it without bringing the cluster down?

The cluster lock disk is in vg01 and won't be affected. I know how to create the map file so I can recreate vg02 on the passive node.

Also is the sequence to remove the vg this?

vgchange -c n /dev/vg02
vgchange -a n /dev/vg02
vgexport /dev/vg02

Rebuild vg02

Then

vgchange -c y /dev/vg02
vgchange -a e /dev/vg02

[Torsten.]

What is your OS version? On 11.23 you can use vgmodify to adjust the VG settings for larger disks.

[Terrence]

I have 11iv2. I think vgmodify is only in 11iv3.

[Torsten.]

No, it came by patch for 11.23 too.

http://docs.hp.com/en/vgmodify/vgmodify_wp_modified_final.pdf

[James R. Ferguson]

Hi Terrence:

Native with 11.31, 'vgmodify' *has* been back-ported to 11.23 when you apply patches
PHCO_35524 and PHKL_32095.

Regards!

...JRF...

[Stephen Doud · Accepted solution]

A document on how to use vgmodify:
HP-UX ServiceGuard - Using vgmodify
Document ID: emr_na-c01316055-3

Whether you employ vgmodify or not, you should be able to continue to run a package assuming the applications either
a) allow the volumes in vg02 to be unmounted
or
b) the applications started by the package do not have any SERVICE monitors that will halt the package if you need to halt the application to umount the vg02 file systems
c) a package maintenance flag feature is employed to allow you to temporarily perform maintenance.