Replacing mirrored disk

[Staci Tribelhorn]

We are currently experiencing intermittent read errors on a mirrored disk. What we want to do is to replace it with a newly purchased disk and move the suspect disk to a new Volume group to see if the problem follows the disk or stays with the tray. Can you provide me with the commands I would use to accomplish this?

First, how do I replace the mirrored disk?
The disk in question /dev/dsk/c5t14d0 and belongs to a series of 4 disks /dev/dsk/c5t8d0 /dev/dskt10d0 /dev/dsk/c5t12d0 and /dev/dsk/c5t14d0. These are all assigned to physical volume pvg2 with volume group /dev/vgdata

My logical volumes are /dev/vgdata/ud52 /dev/vgdata/sb /dev/vgdata/sbdata /dev/vgdata/sbhold /dev/vgdata/sbtrain /dev/vgdata/sbiat /dev/vgdat/sbdev /dev/vgdata/sbcnv /dev/vgdata/olap /dev/vgdata/gw

Second, once I remove the suspect disk from the mirrored pvg, how do I reformat it once it gets to its new home so I can use it in the new volume group?

I've attached of copy of vgdisplay to help answer any additional question that you may have regarding our setup.

Thanks,
Staci

[Sanjay_6 · Accepted solution]

Hi Staci,

You can use the procedure in this doc to replace the the failed mirror disk.

http://www2.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000073905420

the itrc doc id is KBAN00000347.

Once you have replaced the disk, you can use vgextend to add the new disk into this VG and then do a pvmove to move the lv copy to the new disk.

Hope this helps.

Regds

[A. Clay Stephenson]

If this is a hot-plug disk module then you don't even need to shutdown. Simply pull the drive out an inch or so and allow it to spin down. Wait about two minutes or so and remove the drive completely. You box will continue to operate normally. Insert the new drive and allow it to spin up.

If this is not a hot-plug drive then simply shutdown and replace the drive.

From here on, the steps are the same.

vgcfgrestore -n /dev/vgdata /dev/rdsk/c5t14d0
vgchange -a y /dev/vgdata
vgsync /dev/vgdata


You are now completely restored.

[Staci Tribelhorn]

Thanks for the quick response and documentation. It appears at first glance that this will be alot easier than initially thought.