Please help restore a boot disk.

[Hai Nguyen_1]

Folks,

Both of my primary and alternate boot disks were able to boot before.
My HP consultant has just happened to run the command "mkboot -v /dev/dsk/c1t6d0" against my primary boot disk :-(.

Now my primary refused to boot, saying that "cannot find /stand/vmunix or /vmunix). Luckily, my alt. boot disk is still able to boot. I can split the mirror and go thru the sequence to put /dev/dsk/c1t6d0 into vg00 and have it remirrored. But I think that there may be a shorter, reliable way(s) to do it. Please advise. Thanks.

Hai

[Craig Rants · Accepted solution]

Your way should work, here is an alternate...

I've used your way and this way to rebuild. Both work well, do what your most comfortable with.

1. Identify the device file (/dev/dsk/c0t3d0 ) and hardware path (8/0.3.0 )
of the faulty disk. ioscan -fnCdisk
2. Identify disk type (If Possible) diskinfo /dev/rdsk/c0t3d0
3. Call HP to come out.
4. Identify volume group name ( vg00 ) pvdisplay /dev/dsk/c0t3d0
5. Identify the device file and hardware path of the remaining good mirrored disk. vgdisplay -v vg00|more & ioscan -fnCdisk
6. Save off a copy the correct file in /etc/lvmconf/vg00.conf to a safe place.
7. Change primary boot path to the good mirror copy setboot -p 8/0.6.0
8. Identify run level more /etc/inittab init: (3) = run level three
9. You need to override quorum. This is done after Hp has swapped out the disk.
v Stop the system at reboot, bo pri and interact with ISL
v Bring the system up without quorum in single user mode.
ISL> hpux -is -lq (;0)/stand/vmunix
10. Restore the LVM configuration headers. After system has been shutdown and the disk is replaced and quorum is off. vgcfgrestore -n /dev/vg00 /dev/rdsk/c0t3d0
11. Activate the volume group. vgchange -a y /dev/vg00
12. Start the Mirror synchronize process. vgsync /dev/vg00
13. Mount /usr to enable the mkboot commands. mount /usr
14. Place the boot utilities in the boot area.
mkboot /dev/rdsk/c0t3d0
mkboot -a "hpux (;0)/stand/vmunix" /dev/rdsk/c0t3d0
15. Place the diagnostic info in the lif.
If the server is 11.x, first determine if it is 32 or 64 bit:
getconf KERNEL_BITS
If the server is 10.20 or 11.x 32 bit:
mkboot -b /usr/sbin/diag/lif/updatediaglif -p ISL -p AUTO -p HPUX -p PAD -p LABEL /dev/rdsk/c0t3d0
If the server is 11.x 64 bit:
mkboot -b /usr/sbin/diag/lif/updatediaglif2 -p ISL -p AUTO -p HPUX -p PAD -p LABEL /dev/rdsk/c0t3d0

16. Update LIF's Table File with information contained in the BDRA (Boot Data Reserved Area) lvlnboot -Rv /dev/vg00
17. You are still in single user mode at the prompt type init 3 (step 8) this brings the system the rest of the way up.
18. Change back the primary boot path setboot -p 8/0.3.0


Don't attempt to lvreduce the mirrors. Or vgreduce out the disk. These will both hang because of the faulty disk. Over riding quorum with vgchange will also hang.


GL,
C

[Sanjay_6]

Hi,

Try to see if this works,

for your primary root disk,

mkboot -l /dev/rdsk/c1t6d0
mkboot -a "hpux (;0)/stand/vmunic" /dev/rdsk/c1t6d0

then try to boot your primary disk.

Hope this helps.

Regds

[S.K. Chan]

All you need to do is recreate the boot-track..

# mkboot /dev/rdsk/c1t6d0
# mkboot -a "hpux -lq" /dev/rdsk/c1t6d0

Look at what's in the AUTO LIF file in the secondary boot disk and use that string.

# lifcp /dev/rdsk/<2ndary-boot-disk>:AUTO -

[Sanjay_6]

Hi,

If the previos suggestin does not work, try this,

boot thru the mirror,
bo alt
at isl prompt type hpux -lq (;0)/stand/vmunix

pvcreate -f -B /dev/rdsk/c1t6d0
mkboot -l /dev/rdsk/c1t6d0
mkboot -a "hpux (;0)/stand/vmunix" /dev/rdsk/c1t6d0
vgcfgrestore -n vg00 /dev/rdsk/c1t6d0
vgchange -a y /dev/vg00
vgsync /dev/vg00
lvlnoot -b /dev/vg00/lvol1
lvlnoot -s /dev/vg00/lvol2
lvlnoot -r /dev/vg00/lvol3
lvlnoot -d /dev/vg00/lvol2
lvlnboot -v /dev/vg00

Hope this helps.

Regds

[Hai Nguyen_1]

Folks,

Thanks for your help. The problem has been fixed with this command sequence:

1# mkboot /dev/dsk/c1t6d0
2# mkboot -a "hpux -lq" /dev/rdsk/c1t6d0
3# lvlnboot -R

Note that in step one, it is a character device instead. Also, I ran step three as a precaution.

Thanks again.

Hai

[Hai Nguyen_1]

I really meant that it is a BLOCK device in step one.

Hai