System crash to analyze from VAX 4000

[Piet Timmers_1]

Hi,

Our VAX crashed every 10/12 our.

An analyze/error is included.

I need some help to come to a correct conclusion, who dares.

Greetings,

Piet

[Volker Halle]

Piet,

what exactly do you mean with 'crashed every 10/12 hours' ?

The attached errlog entry looks like an MSCP LOG message, reporting some kind of memory error on an MSCP storage controller connected to your VAX 4000-700A.

As the system is running OpenVMS VAX V7.1, there should be a CLUE file from the crash, if a valid SYSDUMP.DMP is available and the system has written a real crashdump.

Look for CLUE$OUTPUT:CLUE$LAST_HG4S26.LIS. If such a file exists, please post as an attachment with your next reply.

If no dump is being written, could you provide the console output from the system at the time of 'the crash' ?

Volker.

[Piet Timmers_1]

Volker that is the problem. There is no information in the errorlog from the time the system crashed, and there is no valid dump, and there is no console output (VT terminal on console). So this is all there is.

The node crashes every 10 to 12 hour, the first time at 00:11 the second time at 11:20 and so on.

I known this is not very must info, but it is all there is.

Greetings,

Piet

[Volker Halle]

Piet,

how about connecting some PC/Laptop with a serial port to the console and recording the console output with Hyperterm or Putty...

Did you check for power (supply) problems ?

Volker.

[Volker Halle]

Piet,

except for hardware reasons, there is one possible 'software' reason, which could cause a behaviour like this:

If the system is incuring some HALT in kernel mode (should not happen) and >>> SET HALT BOOT is set, it will leave no trace of the problem (except messages on the console !), because the console with just BOOT the system again. If this could be happening, use >>> SET HALT RESTART to cause a restart attempt, which would write a restart crash.

Volker.

[Volker Halle]

Piet,

may the MSCP LOG message be in any way related ? What kind of 'controller' is this ? DSSI ?

Does SDA> SHOW CLUS/SCS show a SCS system with a system id of

CDDB$B_SYSTEMID 1101FF68 4200

Volker.

[Volker Halle]

Piet,

and if you decide to capture the console output, make sure to set DUMPSYTLE bit 1, so that maximum console output will be provided. This will help in cases, where the dumpfile cannot be successfully written.

Volker.