topic Re: Frequently Alpha DS20E get halted... in Operating System - OpenVMS

Frequently Alpha DS20E get halted...

Alanjones_1 — Fri, 22 Jun 2007 00:01:18 GMT

Dear Friends,

i m working with 4 node alpha server cluster environment, in that one node DS20E frequnetly goes halted state. what might be the cause? when i analysed after boot the server there is no errors found in error log and operator log.

is there any other analysis method to analyse the entrie server? can any one...

Alan

Re: Frequently Alpha DS20E get halted...

Volker Halle — Fri, 22 Jun 2007 00:16:21 GMT

Alan,

if you get an unexpected halt on an OpenVMS Alpha system, it's best to set the console enviroment variable AUTO_ACTION to RESTART. This will cause the console to restart OpenVMS after an unexpected HALT and cause OpenVMS to write a crashdump and reboot.

This will save errlog entries and make the halt analyzeable.

Alternatively, you need to capture the console output from the halt, this at least gives some clue about the underlying problem.

Volker.

Re: Frequently Alpha DS20E get halted...

labadie_1 — Fri, 22 Jun 2007 00:25:52 GMT

Hello

You have not said which vms version you use.
Depending upon this version, different tools show the hardware errors
_ ana/error
- diagnose
- Compaq analyze (SEA)
- ana/err/elv translate

Check the memory with the command
$ sh memory
Do you see some bad pages ?

Check the Cpu and memory errors with a procedure such as
http://dcl.openvms.org/stories.php?story=06/03/21/8098045

Post your Vms version.

Re: Frequently Alpha DS20E get halted...

Volker Halle — Fri, 22 Jun 2007 00:30:31 GMT

Gerard,

if an OpenVMS system halts and you haven't set AUTO_ACTION RESTART, you need to go to the console and type >>> B to get the system up again.

In this case NO ERRLOG buffers are saved from the time preceeding the halt and absolutely no information is available from the halt situation.

Volker.

Re: Frequently Alpha DS20E get halted...

Peter Zeiszler — Fri, 22 Jun 2007 11:21:23 GMT

Also make sure your dump file is not a shadowed disk. Check your parameters if your dump file is different than the system disk.

Logical - CLUE$DOSD_DEVICE
sysgen - DUMPSTYLE
http://h71000.www7.hp.com/DOC/82FINAL/6549/6549pro_001.html#rb_tab1

Re: Frequently Alpha DS20E get halted...

Andy Bustamante — Fri, 22 Jun 2007 11:36:42 GMT

You should also check sysgen parameter DUMPSTYLE. For an Alphaserver I'd generally use 9, but see the help in SYSGEN. Make sure the sysdump.dmp file is large enough for your crash dump, autogen will tell you if changes are recommended.

Another option would be to schedule downtime and run console diagnostics. I'd recommend using a PC and terminal emulator to be able to scroll back display.

Andy

Re: Frequently Alpha DS20E get halted...

Dean McGorrill — Fri, 22 Jun 2007 12:01:31 GMT

hi Alan,
if it happens again, do a >>> exam pc
exam ps, exam r26. on the running system see
if you can find out where the processer was
with an anal/system and exam the pc address.

Re: Frequently Alpha DS20E get halted...

kari salminen — Mon, 25 Jun 2007 07:16:23 GMT

I've seen this few times and the reason has been a bad memory module.

If the bad memory resides within the system area, and the EXCEPTION code
gets hit by the bad memory, then no dump file or Error Log entry is written,
just plain HALT instruction.

Ask your local HP service to perform a full memory diagnose.

Re: Frequently Alpha DS20E get halted...

Wim Van den Wyngaert — Mon, 25 Jun 2007 08:30:38 GMT

Just this weekend I had bad memory and this resulted in a hang, not a crash. control-p no longer reacting (GS160).

After a few restarts we got a crash instead of a hang and then the ecc errors were reported.

Wim

Re: Frequently Alpha DS20E get halted...

Jim Geier_1 — Mon, 25 Jun 2007 13:10:55 GMT

We have a DS20e, and it was also halting by itself with no crash and no entries in the error log, no evidence at all as to the nature of the problem. Some investigation revealed an obscure problem with the front control panel, called the OCP board, at other sites. Since the front control panel board was replaced, that DS20e has not halted once.