possible bug in 6.2.36

[Justin T. Gibbs]

> We have a raid array consisting of 4 external promise raid  devices
> attatched to an Adaptec 29160 SCSI adapter in a redhat 7.3 box running
> kernel 2.4.24. One afternoon the kernel paniced, we thought because
> of a problem with JFS. We rebooted and noticed in syslog that
> there are many SCSI errors, including the one immmediatly 
> below which asks us to file a bug report.

File the bug report with Promise.  The firmware on your external
controller does not properly support the echo buffer used to perform
domain validation.  I believe that Promise has updated firmware that
resolves this issue.

> We have dozens of these in syslog as well:
> 
> 

The lines just above this will indicate why the dump card state
was generated.  I'm guessing there was a timeout.  Not an uncommon
problem with at least early firmware revisions of these *marginal*
external controllers.  You might be able to prevent these from happening
if you lower the queue depth - at least until you get new firmware.

--
Justin