VxVM failed disk 2

[kHz]

I have two disks c1t1d0s2 disk01 rootdg and
c1t0d0s2 disk02 rootdg. These are on a V480.

c1t1d0 has failed. Using VxVM I know to remove the disk and replace the disk and can use vxdiskadm to perform the operations, but my question is what else do I need to do?

The V480 has hot-swappable disks, so do I have to reboot the server to encapsulate the replaced disk? And I assume I need to mirror the good disk to the failed disk?

Have used SVM and AIX to replace failed disks, but never VxVM.

Thanks for the help.

 

[Annihilannic]

The vxdiskadm options to "remove a disk for replacement" and "replace a removed disk" should automatically perform the "recovery" of the volume, in this case, remirroring.

If they are boot disks (which I would guess not from the names) you might just want to make sure that when it completes it has also applied the VTOC so that you can boot from underlying slices in an emergency. I think it does this automatically in recent versions of VxVM.

Annihilannic.

 

[cndcadams]

KHZ;

sun has an infodoc 40842 that gives the correct procedure for replaceing a failed drive in the v480. It deals with mirrored boot drive. As Annihilannic said it should rebuild itself.


You will need a contract login and password to get to the doc.


Thanks

CA

 

[kHz]

It is VxVM 3.5 and when I ran eeprom it has 'use-nvramrc?=false' but 'nvramrc=devalias vx-disk01 /pci...142,0:a devalias vx-disk02 /pci...17e,0:a' and the 'boot-device=/pci...35,0:a disk net'

The devalias vx-disk01 is the drive that has failed. The device in boot-device (...35,0:a) doesn't even exist on the server. It must have been replaced at one time. But with use-nvramrc?=false it wouldn't ever use the devalias's (vx-disk01 and vx-disk02) but will default to disk. But wouldn't disk be the devalias of vx-disk01? I also need to change the boot-device. I did find the disk failed and was replaced almost 2 years ago and that is where the wrong device is coming from; it was never changed.

My other server that has Veritas has 'use-nvramrc?=false' and 'nvramrc=devalias vx-rootdisk /pci...fe3,0:a devalias vx-disk01 /pci...216,0:a' and 'boot-device=/pci...216,0:a' and my question on that would be why the second disk c1t1d0s2 and is disk disk01 in rootdg while c1t0d0s2 is disk rootdisk and is in rootdg. I guess it really doesn't matter which is booted from, but I always like to have the primary as the first target and the other as the second target.

 

[kHz]

Infodoc 40842, step 9 states this:

If the disk is under VxVM control, be sure to write an SMI label (Solaris 9 4/03 OS or later):

# format -e /dev/rdsk/c1t1d0s2

I am running Solaris 8, and the disk is under Veritas control. So does it still need an SMI label or does it need an EFI lable (I doubt)? Or can I just run 'prtvtoc /dev/rdsk/c1t0d0s2 | fmthard -s - /dev/rdsk/c1t1dos2' to format the disk?

 

[Annihilannic]

Hmm... what are SMI and EFI?

I would normally just use the label command in the format utility before applying the VTOC or doing a vxdisksetup -i .... Presumably that uses an SMI label if it is the "older" type of label.

Annihilannic.