Veritas (3.x) RAID 5 Rebuild On Solaris (8)

It's probably a VERY good idea to read the Veritas/Symantic article "Best practice to recover a failed RAID-5 volume." This page is a shortcut, mainly because it seems our system has 2 disks fall out of RAID-5 almost at the same point in time and during low low activity. In particular, the document says you should dissociate the obsolete subdisk. The Obsolete one is the first disk to fall out of the RAID. Then, mend the valid subdisk which would be the second one to fall out. In our case these are close to identical data -- obsolete and valid.

Summary of flow

• Stop Volume
• VX Mend
• Start Volume
• Superstitious Wrap Up Just to Appease the Gods
• FS Check

Stop Volume

You want to make sure the volume is shutdown when making changes to it's disks.

vxvol -g ftsdg stop ftsvol
vxprint

VX Mend

vxmend -g ftsdg fix unstale ftsdg04-01
vxmend -g ftsdg fix unstale ftsdg05-01
vxprint

Start Volume

# delayrecover is important!
vxvol -g ftsdg -o delayrecover start ftsvol
vxprint

Superstitious Wrap Up Just to Appease the Gods

Had in notes from previously working with a Veritas Tech.

vxtask list
devfsadm -Cv
vxdctl init
vxdctl initdmp
vxdctl enable
vxdg flush ftsdg
vxdctl enable

FS Check =

/usr/lib/fs/vxfs/fsck -F vxfs /dev/vx/rdsk/ftsdg/ftsvol

Now we can mount it

mount -F vxfs /dev/vx/rdsk/ftsdg/ftsvol /FTS