Failing disk

[AKAT]

We have got Itanium Rx8620 with HP OS 11.23
You can see that the root disk is failing which is alternative bootpath

disk 1 0/0/0/2/0.5.0 sdisk CLAIMED DEVICE HP 73.4GMAX3073NC
/dev/dsk/c0t5d0 /dev/rdsk/c0t5d0
/dev/dsk/c0t5d0s1 /dev/rdsk/c0t5d0s1
/dev/dsk/c0t5d0s2 /dev/rdsk/c0t5d0s2 (ALTERNATIVE)
/dev/dsk/c0t5d0s3 /dev/rdsk/c0t5d0s3

disk 2 0/0/0/3/0.4.0 sdisk CLAIMED DEVICE HP 73.4GMAX3073NC
/dev/dsk/c2t4d0 /dev/rdsk/c2t4d0 (PRIMARY ROOT PATH)
/dev/dsk/c2t4d0s1 /dev/rdsk/c2t4d0s1
/dev/dsk/c2t4d0s2 /dev/rdsk/c2t4d0s2
/dev/dsk/c2t4d0s3 /dev/rdsk/c2t4d0s3r
vxdisk list
DEVICE TYPE DISK GROUP STATUS
c0t5d0s2 auto:hpdisk rootdisk02 rootdg online failing
c2t4d0s2 auto:hpdisk rootdisk01 rootdg online

root@fiscliv:/home/root> setboot
Primary bootpath : 0/0/0/3/0.4.0
HA Alternate bootpath : 0/0/0/1/0
Alternate bootpath : 0/0/0/2/0.5.0 (Failing DISK)


We are replacing the mirrored boot disk but we also want to increase the /opt and /stand size so I have drafted a workplan as below, Just wanted to check I am on the right path.
WORKPLAN
Shutdown Server
HP Engineer to Replace Faulty Disk
• Boot Server
• Load image from Ignite network and Install HPUX on the new disk; change the size of OPT and STAND as proposed
• New disk details - 0/0/0/2/0.5.0
• Boot the server from new HW path 0/0/0/2/0.5.0
• Mirror the Secondary disk - 0/0/0/3/0.4.0
o /etc/vx/bin/vxrootmir –v –b c2t6d0 ( Mirror root disk)
o vxprint –g rootdg ( Verify root are mirrored)
o Setboot ( To check it)
o If required use setboot
• # setboot -p x/x/x/x.y.y <-- Set primary disk
• # setboot -a x/x/x/x.y.y <-- Set alternate disk
• # setboot -h x/x/x/x.y.y <-- Set HA alt. boot
• Check final boot from Primary

[Steven E. Protter]

Shalom,

Hope you did an Ignite backup before you tried that.

http://docs.hp.com/en/5991-1236/When_Good_Disks_Go_Bad.pdf

SEP

[Tim Nelson · Accepted solution]

Looks like a good plan.

The only safe way to modify /dev/vg00 is to reload with ignite.

Best of luck.

[Armin Kunaschik]

Have you tried
# vxedit -g rootdg set failing=off c0t5d0s2 ?
Sometimes VxVM thinks, that the disk is broken, but it is not.
There is no risk. If the disk is really broken, the failing state reappears shortly.

If the disk is really broken, you can change the hot swap disk while the system is running.
You don't need to combine both actions, because they not depend on each other.
Remove the disk using vxdiskadm and follow the instructions. See also the mentioned document regarding bad disks.

Do you have OnlineJFS installed? Then it's possible to increase /opt (and even /stand).
But I never done this with VxVM. So I can't describe the exact way to do this. The basic way is to break the mirror of all volumes except lvol1 on one disk, then break the mirror of lvol1 on the other node and increase lvol1. Make sure, that lvol1, lvol2 and lvol3 are exactly continous. (I guess, the names are standvol, swapvol and rootvol in VxVM)
Be sure to have an ignite backup handy if things fail for some reason.

My 2 cents,
Armin