Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

mirrorvg rootvg w/o need to reboot for quorum

140 views
Skip to first unread message

Tomas Lindhoff

unread,
Jul 5, 2001, 3:13:55 AM7/5/01
to
Hi!
Anyone with tips on howto mirror rootvg on 2 disks with quorum
disabled and without the need to reboot for the quorum status to take
effect ???
(or maybe it's not possible?!)
For example:
If I use mirrorvg -c 2 rootvg hdisk1 it will mirror without quorum
but I will have to reboot for the system to know about this
(I cannot use the varyoff;varyon for rootvg)
The system in question is very critical and a reboot would like to be
avoided in the chance of a disk crash in rootvg. You should just have
to unmirror, reduce, change disk, extend, mirror with the quorum
actively disabled and no reboot.
Possible ??
Thanks /Tomas

Christian GILBERT

unread,
Jul 5, 2001, 7:42:40 PM7/5/01
to
Did you try this :
mklvcopy ....
syncvg rootvg
chvg -Qn rootvg ?

Tomas Lindhoff a écrit :

Gavin Wilson

unread,
Jul 5, 2001, 8:26:54 PM7/5/01
to
Changing the quorum setting using either the "mirrorvg" command or the
"chvg -Qn" command only changes the attribute flags for the volume group in
the ODM and has no effect on the behaviour of the volume group if it's
currently varied on. For the quorum changes to "take affect" the volume
group must be varied off and then varied on again. Unfortunately rootvg
cannot be varied off during normal operation and hence the only way to
achieve varyoff/varyon of rootvg is to reboot the system. If you don't
shutdown and reboot then if you lose one of the rootvg mirror disks, the
system will still fail as quorum will have been lost.

Note: Don't forget to generate a new bootlv (hd5) "bosboot -ad <device
name>" after the quorum setting has been turned off as the bootlv contains a
striped down version of the ODM (devices required for system boot),
obviously this contains the ODM entries for the rootvg and if you fail to
generate a new boot logical volume then when the rootvg is varied on during
the boot process...... quorum will still be enabled !!!!

Regards Gavin

"Tomas Lindhoff" <t...@integris.se> wrote in message
news:ff19ae82.01070...@posting.google.com...

Tomas Lindhoff

unread,
Jul 12, 2001, 9:45:20 AM7/12/01
to
Thanks for your confirmation Gavin !

I guess a 3 way mirroring is the only way to go if you'd like to avoid
a reboot in case of a disk failure in rootvg, huh ?!
And I assume you then have quorum set on ?!

rgds // Tomas

Steve Bradshaw

unread,
Jul 23, 2001, 4:03:33 AM7/23/01
to
t...@integris.se (Tomas Lindhoff) wrote in message news:<ff19ae82.01071...@posting.google.com>...

Tomas,

The reboot that is necessary after the mirroring of rootvg is as Gavin
explained to get the active volume group to switch to quorum being
off.

This is not the same as having to reboot in the event of a disk
failure. The reboot for the quorum change is a once only event -
after that reboot the rootvg quorum is set to no and the volume group
will behave accordingly.

In the event of a failure of a disk in rootvg, providing the volume
group has been mirrored correctly, there is no need for a reboot to
occur, this is the whole point of mirroring the rootvg, however on a
subsequent reboot of the system with a disk still failed there may be
some need to take manual action to get the system up. The mirroring
of rootvg allow the system to keep up and running in the event of a
disk failure in rootvg.

Triple mirroring is an option if you have 2 spare disks and cannot
afford the down time associated with a reboot to allow the quorum
changes to take effect. Or you could mirror the logical volumes over
2 disks and just add a third disk to the volume group to "Quorum bust"
- this used to be the usual way of creating a mirrored volume group
with out having to purchase three equal sized disks - you have say 2
9.1 GB disks and 1 2.2 GB disk, the smaller used only for the quorum
setup and not for data.

You can quite happily fail a disk in rootvg with mirroring and then
reintegrate that disk with out the system going down - if the disks
are hot swappable that is. This would be part of the standard tests I
would perform on such a system to check that it can stay up with a
failed disk.

Steve Bradshaw

0 new messages