11i upgrade failure - mirroring issue ?

[Ian Foster]

My problem is a little long on explanation and short on information - apologies - but I always value any opinion I get here.

While I was on leave another administrator attempted to upgrade our N class 9000 server from HP-UX 11.00 to 11i following a procedure I had already prepared.

The plan was to split off the root disk mirror (root lv's mirrored strictly using MirrordiskUX) as described in the forums, prove we could reboot from the split mirror for regress then boot back to the primary and run the upgrade.

The detail as to what actually occurred is a little sketchy unfortunately, but everything seems to have gone ok with testing the regress procedure and the team could boot from either mirror and tested switching this several times.

However, after running the update-ux routine on the (still split) primary they could not boot from either mirror, and eventually could only boot from a recovery disk to the 11.00 system and ,with errors, using a recovery cd to the 11i (primary) disk.

The suggestion that has been passed back to me is that the LIF area was somehow shared or had been modified during the upgrade causing the problem.

Does this ring true with anyone, or has something else gone wrong ?

The mirroring setup/policy appeared to be correct as far as I can determine.

Unfortunately I have no firm details of the errors they were encountering although at one point they got an error mesage stating that init was respawning too rapidly when trying to boot the new system.

I need to try and get a better handle on what went wrong so I can come to some reasonably informed decision as to how to move forward (we are now running on the original mirror with a supposedly unbootable version of 11i on the primary disk, so effectively I have lost my safety net now).

Any suggestions very gratefully recieved !

[Pete Randall · Accepted solution]

Ian,

Without more specific errors it's hard to come up with any sort of reliable guesses. I would suggest, when you're ready to try this again, that you first do an Ignite make_tape_recovery backup before you start. Then split the mirrors and do a cold install. I can see no reason why the LIF area would be shared in any way shape or form in a mirrored configuration.


Pete

[Steven E. Protter]

Two possible suggestions now that the box has been thoroughly trashed.

1) Cold Install 11.11, the apps will have to be upgraded or reinstalled anyway.
2) Stick that Ignite tape in the tape drive, boot and restore the system to before your backup messed up your day.

SEP

[Geoff Wild]

Does the box have the correct firmware revision?

One of the prerequisites is that the firmware be atleast at 41.02 or higher.

To check:

>/usr/sbin/cstm
Running Command File (/usr/sbin/stm/ui/config/.stmrc).

-- Information --
Support Tools Manager


Version A.38.00

Product Number B4708AA

(C) Copyright Hewlett Packard Co. 1995-2002
All Rights Reserved

Use of this program is subject to the licensing restrictions described
in "Help-->On Version". HP shall not be liable for any damages resulting
from misuse or unauthorized use of this program.

cstm>scl type proc
cstm>info
cstm>il
-- Converting multiple raw log files to text. --
Preparing the Information Tool Log for each selected device...

.... system3 : XXX.XXX.XXX.XXX ....

-- Information Tool Log for CPU on path 33 --

Log creation time: Mon Jul 7 09:06:42 2003

Hardware path: 33


Product ID: CPU Module Type: 0
Hardware Model: 0x5cc Software Model: 0x4
Hardware Revision: 0 Software Revision: 0
Hardware ID: 0 Software ID: 601329313
Boot ID: 0x2 Software Option: 0x91
Processor Number: 0 Path: 33
Hard Physical Address: 0xfffffffffed21000 Soft Physical Address: 0

Slot Number: 8 Software Capability: 0x100000f0
PDC Firmware Revision: 40.25 IODC Revision: 0
Instruction Cache [Kbyte]: 512 Processor Speed: N/A
Processor State: N/A
Monarch: Yes Active: Yes
Data Cache [Kbyte]: 1024
Instruction TLB [entry]: 160 Processor Chip Revisions: 2.3
Data TLB Size [entry]: 160 2nd Level Cache Size:[KB] N/A
Serial Number: 0


So, on that box, we got HP to update the PDC first - then the upgrade went flawlessly...

BTW - I also split the root mirror and had no issues.


Rgds...Geoff

[Ian Foster]

Thanks Guys. This confirmed my gut reaction.

I have since done some more digging around on what's left of the 'upgraded' system disk and what really happened was that the update-ux process encountered errors which caused the failure - the shared LIF area / mirroring issue theory was definitely somebody barking up the wrong tree and missing the crux of the problem.

I am tentatively investigating the errors that were encountered by update-ux, as the business were planning to dump an ignite image of the working 11.00 system from the good disk back to the screwed disk (the original primary) then try running the upgrade yet again, but I have no real confidence that this can be made to work at present.

My guess is we will be securing an ignite image of our working system from the current disk,resynching the mirrors and planning a cold-install.

Thanks again.

[Kevin Wallace]

Didn't see what OE you upgraded to. There is an issue with the technical OEs (11i-MTOE and 11i-TCOE) for which you must reinstall PHKL_23333 even though it is included with the update. Got this info from the HP class "Upgrading to HP-UX 11i"