watchdog timeout

[Marius Strobl]

On Sun, Jun 23, 2013 at 01:24:19PM +0200, Riccardo Veraldi wrote:
> 
> it is 9.1-RELEASE
> 
> I have no way to reproduce it, because it appeared suddently without any 
> specific cause.
> The system was not under heavy load, I Was not compiling anything...
> 
> But I had other errors recently  which are really not related I think, 
> mainly parity SCSI errors on isp0
> 
> Jun 12 04:48:38 blade kernel: (da3:isp0:0:1:0): CAM status: SCSI Status 
> Error
> Jun 12 04:48:38 blade kernel: (da3:isp0:0:1:0): SCSI sense: ABORTED 
> COMMAND asc:47,0 (SCSI parity error)

In fact, there's one open issue with isp(4) causing timeouts of the
kind you have seen but which never got quite resolved. Investigating
that as a potential culprit for the case you have encountered would
require updating to stable/9 and possibly giving a patch a try,
though. Besides, we'd need a way to reliably detect whether the
problem is gone.
On the other hand, the parity errors indicate that this system is
suffering from hardware problems, including the possibility that a
disk is defective or has broken firmware and the combination of some
components not getting along for some reason.

Marius