HP 9000 K220 800

[David Gwin_1]

I posted this issue a couple of weeks ago and received a few suggestions but I am still having a problem.
I have a HP 9000 K220 running HPUX 10.20 (with all of the latest patches), 512 mb ram, disk array Model 20.
The problem is that the system just stops responding to all activity (no warning, no consol message... nothing) and I have to shut it off and reboot. This is happening at different times under different conditions. I have not found anything in any log file to give me a clue where to look. I don't know how to start a TOC or trace. My educated guess is telling me that it is a hardware issue. The system had been shutdown for the past 2 years and does not have any type of HW support.

Thank you for any help you can provide.

[Mark Ellzey]

David,

I think your educated guess is correct. I really don't have any suggestions other than to check all the connections on the drives, cpus, memory and backplane. Do be careful, tho, you don't want to bend or break a pin in a connector. Also, use a ground strap to avoid zapping a board with static electricity.

Without a HW support contract, you are pretty much on your own. I know, I'm in the same situation, and it sucks.

If you do find a faulty piece of hardware, checkout www.configsys.com. They have a lot of hardware for HP K series boxes, and at very reasonable prices.

HTH,
Mark

[Stefan Stechemesser]

Hi David,

I have some hints how I would troubleshoot this situation ( maybe you want to call the HP support ? ;-)

If a hang is happening, then you should start a TOC (Transfer of Control) to get a memory dump of the system. You can do this by setting the key switch to "service" and then enter "Control-b" on the Console to get the CM prompt. Then enter "TC" to initiate the TOC. If everything works fine, a dump is written to swap and during the bootup to /var/adm/crash.
Read the software recovery handbook (ITRC=>Knowledge Trees) how to analyze them. Maybe you find some useful hints in the message buffer (f.e. SCSI problems etc.).
Disk Problems are the most common source for hangs (is the system mirrored ?).
Another way of troubleshooting is to deconfigure CPUs in the BCH Service Menue and check if the error happens again.

I hope this helps troubleshooting the system ...

best regards

Stefan

[Bill Hassell]

You'll need to setup for a crash dump before a TOC will help. Check /etc/rc.config.d/savecore to make sure it has SAVECORE=1 and SAVECORE_DIR=/var/adm/crash. Now this assumes that you have at least 512Mb of space left in /var/adm/crash (hint: bdf /var/adm/crash). If not, look around for an lvol with enough space and then either change SAVECORE_DIR to point to it, or mount the lvol at /var/adm/crash.

Once completed, you can use the CTRL-B and TC commands. Note that you'll have to turn the front panel switch to service (all the way clockwise) for CTRL-B to work.

As you might expect, a hang with no console access means that nothing will be logged either, thus the need to use a crash dump. But this will very likely require HP to analyze. The K-box is ancient in computer years and 10.20 is obsolete and unsupported so this may be a costly adventure. You may be able to obtain another K-class box just for shipping costs (they are heavy though). A working K460 with a lot more memory and extra CPUs may cost just a few hundred dollars (which is a lot less than having HP analyze your crash dump).

[Jannik]

when you say all activities you meen ALL, or is it just LAN?
If it is lan it could be switch port fault:
use another port and module.
It could be a duplex problem, many of the old K-Class only use 10MBit Half-duplex look at output from lanadmin:
lanadmin -sx 4
Speed = 10
Current Speed = 10 Half-Duplex

If the problem is extenal storrage SCSI I would disconnect it and see if the pinns is looking ok.

[Ray H. Gay]

Did you find the prob? I have a 9000/800 with similar behavior. It fails to boot at times with a console error from the boot utility - don't know if isl or bch. I think it is either a disk or controller failure, but can't catch it in a log.

[David Gwin_1]

I still have not found the real problem. I have tried dropped back to 256 ram and swapping it around but it is still shutting down. The first indication that the system is going to crash is when i can not connect to the system via workstation. Then i goto the consol and i can do one or maybe two commands and the system is dead.

[mostefai mourad_1]

Hi

I think you should check if you got an HPMC in your system, to do that, check the latest file under /var/tombstone/tsxx.
if there is some thing in the file, you can send to me and I will troubleshoot it.

regards

[Marek Smejkal]

David,

If you have installed online diagnostics (STM) you can run some HW verifications and you may get one day password from HP to run more tests.
Marek

[Bill Hassell]

Make sure you have setup the crash feature I mentioned above. And turn the key on the K220 to the Service position. The reason is that you'll now be able to see hardware problems and at least be able to type CTRL-B to get a processor (not HP-UX) prompt. Type "co" to get back to HP-UX. Note that all the symptoms sound like a hardware failure and unless you have a spare K220 to use for parts, knowing that a processor has failed won't help. You'll still need a service call. However, the K-series is quite old now and you can likely get a replacement (with a lot more memory) for just a couple of hundred dollars US. Note that shipping costs may be a $50 to $250 depending on distance and time. You can probably get a much faster K-series for the same price (a K460 or maybe even a K580). Checkout eBay.