nothing to do for replace a mirrored non root disk...?

[thomas mornet]

Hello everybody,

first excuse me for my language.
I have to replace a failed disk in a non root Volume Group, in a MC/ServiceGuard Cluster environnement.( 2 nodes HP9000 with UP-UX11.00). I have two mirrored disks for VG01(just one logical volume lvol1), and one (/dev/dsk/c4t12d0) failed.
the disks are in a Storage Enclosure HASS, in RAID 1.
I think that i have to do the following:

1-vgcfbackup /dev/vg01
2-replace the failed disk by a new, without shutdown.
3-vgcfrestore -n /dev/vg01 /dev/rdsk/c4t12d0
4-vgchange -a y /dev/vg01
5-vgsync /dev/vg01
6-control with lvdisplay -v /dev/vg01/lvol1 | grep stale

But someone tells me that there is nothing to do because the HASS automatically do all I wrote in the procedure!
I'm lost, because i'm novice...
is there anybody to confirme or no this fact?

Thanks you very much for your help

[James R. Ferguson]

Thomas:

Document KBAN00000347 describes the procedures to follow for replacing a failed disk. I think you'll find this very helpful.

Your procedure looks good.

...JRF...

[Murray Jones]

I think it does depend on what you are using to mirror the disks. If your system uses hardware mirroring as opposed to software mirroring (Mirror/UX), I would tend to agree with the other viewpoint - replace the disk (is it hot pluggable?) and let the hardware do the rest. Unfortunately I am not familiar with the HASS enclosure you speak of.

If all else fails - find the manual.

Cheers,

Murray

[Manju Kampli]

The storage enclosure HASS does not have any Hardware mirroring capability. SO I think your configuration is a software mirroring using a Mirror-UX.
The steps you have mentioned above are correct. Once you do vgchange -a y it will do sync automatically, and no need to do vgsync. anyway there is no harm in doing that. But ... There is a bad LVM patch around and when the mirror disk fails and you replace it, this LVM patch does not do sync, and just thinks everything is synced. This might cause in data corruption. The bad patch is PHCO_19060. there is a superseeded patch which fixes the problem PHCO_21630. First install this patch, and then do the above procedure. The above procedure has been tested on production system by me and should work strightaway.

[thomas mornet]

Hello,
thanks to you to answer so quickly.

James, thanks for the document.
Murray, the system uses software mirroring, and the disk is hot pluggable.
Manju, i'm ok with you for the automatic vgsync with vgchange -a y. but i don't find these patches on my system.

Thanks for all
thomas