CPU EXCEPTION in uerf

[kkdurrani]

Hi all,
We have some performance isssues on alphaserver 4100. Just wondering if these "cpu exception" are hardware related issue.

Kairm

EVENT CLASS ERROR EVENT
OS EVENT TYPE 100. CPU EXCEPTION
SEQUENCE NUMBER 123.
OPERATING SYSTEM DEC OSF/1
OCCURRED/LOGGED ON Wed Aug 10 21:01:12 2005
OCCURRED ON SYSTEM cpaprd05
SYSTEM ID x00070016
SYSTYPE x00000000
PROCESSOR COUNT 4.
PROCESSOR WHO LOGGED x00000000

----- UNIT INFORMATION -----

UNIT CLASS CPU

********************************* ENTRY 2. *********************************

----- EVENT INFORMATION -----

EVENT CLASS ERROR EVENT
OS EVENT TYPE 100. CPU EXCEPTION
SEQUENCE NUMBER 122.
OPERATING SYSTEM DEC OSF/1
OCCURRED/LOGGED ON Wed Aug 10 21:01:12 2005
OCCURRED ON SYSTEM cpaprd05
SYSTEM ID x00070016
SYSTYPE x00000000
PROCESSOR COUNT 4.
PROCESSOR WHO LOGGED x00000000

----- UNIT INFORMATION -----

[Ivan Ferreira]

CPU EXCEPTION messages are often related to memory problems, BUT:

If you are using Tru64 V5 or above, you should not use uerf, you must install and use DECEVENT, the dia command replaces the uerf command. The uerf command wont translate correctly the binary error log.

[Stuart Fuller_2]

For each of the entries, is there more information, or was this posting edited down? If edited down, then all the good information has been edited out.

[kkdurrani]

Yeah this posting is edited down for the good view. attached is the full file

[Stuart Fuller_2]

Yep, looks like a job for DECEVENT.

[Aaron Biver_2]

Hi,
uerf will not help much here. WEBES/WSEA (Compaq Analyze) would be ideal. Support personnel are very adept at interpreting the output.

Was there a crash associated with the event? If so, can you provide the console output (maybe from the crash dump or the /var/adm/messages file) from before and after that event? A stack trace from the crash dump is sometimes useful here, though not likely.

[kkdurrani]

No There was no crash. I have run dia -i cpus. Please help me to figure out if anything suspecious.

[Aaron Biver_2]

These appear to be hardware-correctable ECC errors. They occur every month or so. Indicates a bit gets whacked every now and again.

Not a big deal, based on what I see in the dia output.

[kkdurrani]

No hardware problems. they are correctable errors

[Michael Worthington]

Encountered similar issue on Alpha 1200 - identified as faulty memory - resolved by replacement