Mirror that have a crash disk

[Javier Pérez F.]

Hellow, i have a problem with an disk array in a hp-ux, having installed the mirrordisk/ux software.
I have done a reboot on the system, before the reboot all is ok, but al rebooting the system has not mounted one filesystem /bd in a vg /dev/vg01 displaying warning. It seems that one disk in a mirror is crash, i try to activate the vg and the system says no, i try to do lvreduce on vg for deactivate the mirror and the system says that the vg is not activate. Can anyone to say me how can id do that?

[S.K. Chan · Accepted solution]

It is possible that vg01 has failed the quorum requirement (ie half or more number of disks in the VG has failed thus not able to activate the VG). Try to disable the quorum enforcement by running ..
# vgchange -q n /dev/vg01
the activate it ..
# vgchange -a y /dev/vg01
Does this work ? If it does you can then identify which disk is dead and how your data is impacted by this.
Another way to tackle this is to replace the mirror disk that's faulty. Assuming cxtXdX is the new disk.
# vgcfgrestore -n /dev/vg01 /dev/rdsk/cXtXdX
# vgchange -a y /dev/vg01
Now restore the lost data from tape (if needed).

[Sridhar Bhaskarla]

Hi,

If it is only missing the mirror, you can activate the VG by

vgchange -a y -q n vg01

But with your description, I think that the disk that failed has the data. You will have to replace the disk, make sure it is claimed through (ioscan -fnC disk). Once it is done, then do

vgcfgrestore -n vg01 /dev/dsk/cxtydz.
vgchange -a y vg01
vgsync vg01
mount -a

If the disk has only mirror, then the above should recover it. If not, you will need to restore your /bd from backups.

-Sri

[Javier Pérez F.]

Many many thanks.
Many thanks S.K.Chan and Sridar Bhaskarla.

Yours answers are the solution.

I have one lvol with 2 physical disks, one is wrong, the other seems ok.
I do:
vgchange -a y -q n /dev/vg01
and then the vg is activated.
then i do mount /bd
and all is ok.

Bets regards,
Javier P??rez

[Sridhar Bhaskarla]

Hi Javier,

It is only a temporary patch not a solution. You will need to do a permanent fix to your problem which is to replace your HD and then run vgcfgrestore followed by vgsync.

Under normal circumstance, you would not use -q -n option. If you can see the data now, I strongly suggest you to take a backup of your /bd filesystem and then replace the disk.

-Sri

[Javier Pérez F.]

Many thanks again Sridar Bhaskarla

Yes, i will do that. But i don't have support on week-end. I will do a backup on tape now, and ... "in system we trust"

Thanks
Javier Perez

[Burxi]

Xabier...Begira non aurkitu zaitudan. Hau surrealista da, oso, eta nire buruak ezin du egiten ari zarena ulertu. Zerbait jakin nahi nuke..., mesedez

[Burxi]

agur betiko, Javier