Integrity VM boot disk problem

[Coolmar]

Hi,

We have a particular VM that might be having problems. When I do an "lvlnboot -v" .... I get no output. When I do a diskinfo on the disk I know to be the boot disk I get an I/O error. Other than that, the system *seems* to be fine but we have not booted it yet (doing a backup first). No LBOLT or scsi errors in dmesg either.
Any ideas for more things I can check out?

Thanks,

[Mike Shilladay]

Hi,

Are you executing those commands on the VM host or the VM guest?

Mike.

[Coolmar]

On the guest itself.
The commands are fine on all the other guests.

[Torsten.]

What is the type of backing store (file/lvol/disk)? Just in case it is a disk - is it on SAN?

You won't see any errors from the HPVM regarding hardware problems. Virtual devices will never have a hardware problem - only virtual hardware problems ;-)

How about the VMHost? Does it report any errors?

[Mike Shilladay]

Hi,

If you do a hpvmstatus -P , does the disk you expect appear in the list?

Mike.

[Coolmar]

Yes, it is SAN disk.

Yes, the VM and disk both appear in the hpvmstatus.

[Torsten.]

Again: plain disk or file or lvol?
If it is a plain disk - do you have any multipath software in use?

[Coolmar]

Sorry...yes plain disk...scsi. We do have multipath running and the path is "active".

Oh, I should mention also. When I do an ioscan on the host, all the VMs' boot disks show with their slices, etc. This one shows as a normal disk.

IE:
VM boot disk:
disk 35 1/0/4/1/0.1.26.0.0.7.0 sdisk CLAIMED DEVICE HP OPEN-E
/dev/dsk/c10t7d0 /dev/dsk/c10t7d0s2 /dev/rdsk/c10t7d0 /dev/rdsk/c10t7d0s2
/dev/dsk/c10t7d0s1 /dev/dsk/c10t7d0s3 /dev/rdsk/c10t7d0s1 /dev/rdsk/c10t7d0s3

How this one is showing:
disk 225 1/0/4/1/0.1.28.0.0.0.6 sdisk CLAIMED DEVICE HP OPEN-E
/dev/dsk/c21t0d6 /dev/rdsk/c21t0d6

[Torsten.]

What multipathing? The hardware path does not look like this.
Can you provide some more information (e.g. hpvmstatus ...).

[Coolmar]

Autopath


Virtual Device File : /hpap/dsk/hpap179
Load Balancing Policy : Shortest Queue Length
Lun Timeout : Infinite Retry (-1)
==================================================================
Device Path Status
==================================================================
/dev/dsk/c19t8d6 Active
/dev/dsk/c21t0d6 Active

[Mike Shilladay]

Hi,

Does the SAN provide boot support ?

What type of SAN is it ?

Mike.

[Torsten.]

Did you assign to virtual disk device to your HPVM?
hpvmstatus will show you ...

[Coolmar]

MIke - Yes it does. We have many VMs here and they are all booting from the SAN disk.

Here is the hpvmstatus:

[Storage Interface Details]
Guest Physical
Device Adaptor Bus Dev Ftn Tgt Lun Storage Device
======= ========== === === === === === ========= =========================
disk scsi 0 0 0 0 0 disk /dev/rdsk/c21t0d5
disk scsi 0 0 0 1 0 disk /dev/rdsk/c21t0d6
disk scsi 0 1 0 0 0 disk /dev/rdsk/c21t0d4
disk scsi 6 6 0 0 0 file /var/opt/hpvm/guests/vndevdb/tape6
tape scsi 6 6 0 2 0 attach /dev/rscsi/c31t0d2
tape scsi 6 6 0 4 0 attach /dev/rscsi/c32t0d1
disk scsi 6 7 0 0 0 file /var/opt/hpvm/guests/vndevdb/VTL7

[Mike Shilladay]

Hi,

Not wanting to state the obvious, but if the are not s1, s2 and s3 chunks, won't that cause it to behave oddly. It looks as though there hasn't been an "idisk" done on the disk.

Other than that, I am puzzled.

I guess it is over to you Torsten.

Mike.

[Coolmar]

Well that's how it appears, but the server seems fine. I haven't rebooted yet though! Holding that off for a bit. All our other vm's have the idisk layout when you ioscan, except one other one. On that one though, I can do an lvlnboot -v and diskinfo and everyhting is fine. So I figured that the ioscan might be funky for those two systems for some reason.

[Coolmar]

Just rebooted and the VM won't come up. Can't find the boot disk. Does anyone know of a way to rebuild it?

[Torsten.]

It's a bit too late for more tests from the running guest...

Regarding the backing store disk:
If you just created the guest and installed, the VMhost doesn't know about the partitions.

Unless you reboot the VMhost or run an insf you won't see the "sx" device files - but this is no problem at all.

If you connect to the vMP of the guest and run an EFI shell, what do you see when you run

> reconnect -r
> map -r

[Coolmar]

I did an idisk -p and it said that the headers and whatnot were lost and to do a "-r" ... did the -r and it said it grabbed the headers from the alternate disk and everything looked fine. Tried to boot and no go. So re-igniting and restoring.