RE: aacraid failure? Dell PE2500, RH8, kernel 2.4.18-27.8.0smp ( RH Errata version), Perc 3/Di

I have seen a similar panic in a different driver that was triggered by the
device offline status being set in the middle of error recovery. This caused
a recursive loop in the scsi error handler.

Although in the `other' driver, the problem was caused by an inappropriate
action of that driver, the debug I did looked like it could also occur due
to a timing bug in the SCSI subsystem as a device is being taken offline
naturally. The aacraid driver in 2.4.18-27.8.0 does not perform any
modifications of the device status, and as a result, I expect this is the
possible bug in the SCSI subsystem being triggered.

Yes, this is an *ungraceful* handling of a disk error or timeout causing the
array to go offline, and then subsequently panicking the operating system.
The current discussion in this list regarding the "Host adapter reset
request. SCSI hang?" is being looked at with a great deal of interest as
this is probably the root cause of your problem. All driver variants appear
to suffer from this problem, this is not a regression, but a problem that
has always been there, but has become more prevalent with faster processors
...

I can *not* comment further on the failure until we get that spark of
epiphany to understand the root cause of the timeouts. I have not been able
to duplicate this issue to investigate it.

Sincerely -- Mark Salyzyn

-----Original Message-----
From: Peter Smith [mailto:[email protected]]
Sent: Monday, June 23, 2003 11:25 AM
To: [email protected]; [email protected]
Subject: aacraid failure? Dell PE2500, RH8, kernel 2.4.18-27.8.0smp (RH
Errata version), Perc 3/Di


This morning I spontaneously got the following on the (serial) console.  
Hopefully someone might be able to tell me "it is the driver" or "it is 
the hardware."  I don't think it is the hardware.  Also attached is the 
controller details (from a second machine) which are identical to the 
(currently downed) machine in question.  And yes, the second machine 
does, at times, die in the same manor as this particular machine just 
not as often.



Peter Smith




_______________________________________________
Linux-aacraid-devel mailing list
[email protected]
http://lists.us.dell.com/mailman/listinfo/linux-aacraid-devel
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/