Auto RAID 12 H Not activating Volume groups

[KapilRaj]

Hi friends,

I hv an HP9000 D370/1 Server with 4GB SCSI Internal disk

This is connected to An AUTORAID BOX of 8* 9.1 SE DISKS

The 12 H controller gave an error today morning "Disk B4 failed"

I shutdown the RAID Box

Replaced the drive with a new one (Same model)

Restated the RAID BOX

Did a Rebuild manually from the control panel(as Auto Rebuild was disabled).

Then i started the HP9000 system.

It was not coming up , the Volume Groups belong to Diskarray were not getting activated

the system was configured by another person so i hv less knowledge abt how it was. Can anybody help me ?

My /etc/lvmtab says

/dev/vg01

/dev/dsk/c0t0d0
/dev/dsk/c0t1d0

/dev/vg02

/dev/dsk/c0t0d1
/dev/dsk/c0t1d1

/dev/vg03

/dev/dsk/c0t0d2
/dev/dsk/c0t1d2

/dev/vg04

/dev/dsk/c0t0d3
/dev/dsk/c0t1d3

/dev/vg05

/dev/dsk/c0t0d4
/dev/dsk/c0t1d4

/dev/vg06

/dev/dsk/c0t0d5
/dev/dsk/c0t1d5

#ioscan -fnC disk output is as follows

It shows all LUN's of c0t1d* but nohting abt c0t0d*


Kapil Raj

[Bill McNAMARA_1]

what's c0t0 ?
send the ioscan -fnk
so we can have a look. If the t0 is
not showing on reboot, you probably have
a cabling problem to one of the controllers.
On your ioscan look for the ext_bus instance
number 0, should be okay, since c0t1 is okay,
but if t0 doesn't show up, you've either got
a scsi id conflict on c0 or a controller
/ cable failure on the autoraid.

Send: ioscan -fnk
arraydsp -h
and
Check SCSI cabling behind the autoraid,
you may have terminated at the wrong
position.

You vg's seem to be alternate paths, so
your vg should activate. .. try to activate
manually (ignoring quorum requirements)
vgchange -a y -q n vg01
etc..

Bill

[Bill McNAMARA_1]

also.. one of your autoraid controllers
may have gone offline..
send an arraydsp -c
to get controller info. You can verify from
the control panel SCSI area that the scsi id
is what you expect.... and you didn't change
it by accident

[KapilRaj]

Hi Bill

What you suspect is close to the scenario here.

We had two SPs, the backup SP is removed and now the Autoraid is on single SP ( Status - Warning - single controller )

Meanwhile, one of the disks failed and we replaced with a new disk (in B4).

Upon rebuild, the warning was - rebuild failed.

A manual vgchange reports - could not query physical volumes.

The arraydsp output shows everything to be normal. "arraydsp -a" attached.

[Insu Kim]

It's natural that you got a message, couldn't query physical volumes, because AutoRAID is on single controller.

Notice that lvmtab sees two device files for each volume group, which are c0t0* and c0t1* but c0t0* does not exist anymore.

I think that device files, c0t0*, are associated with removed controller.
You can check this out on display panel.
Try to see SCSI ID for removed controller.
It should be 0.


| |_________ SCSI ID for controller.
|___________ CARD instance number.

You know this throughout the ioscan output.
All that you can do now is to activate volume groups manually with quorum off.
# vgchange -a y -q -n

or you can remove device files starting with c0t0* for each volume group which will let lvmtab have only one entry for each volume group.

# vgreduce

For your information, you don't need to replaced the drive with disk array down.
You can do it online.

Regards.

[Insu Kim]

It's natural that you got a message, couldn't query physical volumes, because AutoRAID is on single controller.

Notice that lvmtab sees two device files for each volume group, which are c0t0* and c0t1* but c0t0* does not exist anymore.

I think that device files, c0t0*, are associated with removed controller.
You can check this out on display panel.
Try to see SCSI ID for removed controller.
It should be 0.


| |_________ SCSI ID for controller.
|___________ CARD instance number.

You know this throughout the ioscan output.
All that you can do now is to activate volume groups manually with quorum off.
# vgchange -a y -q -n

or you can remove device files starting with c0t0* for each volume group which will let lvmtab have only one entry for each volume group.

# vgreduce

For your information, you don't need to replace the drive with disk array down.
You can do it online.

Regards.

[KapilRaj]

hi friend,

Thanks and i tried the same, it activated the volume Group .

But i am not able to mount / do an fsck on the said logical volume.

And array daemon says " the data at /dev/c0t1d0 un available. Rebuild failed.

What does this mean. I hv enough disk space now as i added one as hot spare.

Regds,


Kapil

[Insu Kim]

Hello,
It looks like that HPUX is on AutoRAID.
Type lvlnboot -v and take a look at where boot disk is.
If boot disk is one of the device files which belong to AutoRAID,
then you'd better remove missing physical volume for root file system.
In this case, c0t0d* is missed.
# vgreduce
If it worked, try to reboot the system and see what is happening.

Or try to bring the system up with quorum disabled.
I'm not sure that this will work.
After power is up, you will see " To interrupt press any key within 10 seconds."
At that moment, press any key, which will bring you be main menu.
Do the followings.

Main menu> bo
Interact with ISL> Y
ISL>hpux -lq

[Insu Kim]

Another way of mounting root volume.
You can go directly to maintenance mode where root volume is not activated there.


Main menu> bo
Interact with ISL>Y
ISL> hpux -lm

Try to activate root volume with quorum off.
# vgchange -a y -q n
# mountall

Place the system in one of the run-levels, 0 to 6 by executing init command.

ex) init 2

Best regards,

[KapilRaj]

Hi friends,

Sorry for the late reply. As u said my root volume group is not on the array. I have a seperate disk 4.3 GB for that. Yesterday my controller card got repaired , As soon as i fixed it , the array started rebuilding the data. And after that i could mount all the filesystems.

Does this mean that my Old controller had some prob?. Or is it like this " For rebuilding the data , all the controllers have to be present?

Regds,

Kapil Raj