Re: kernel panic after 5.2 RELEASE --> 5.2.1, DELL PE2650, SCSI

Raphael H. Becker wrote:
> Hi,
>
> yesterday I got a new DELL PE2650 with PERC and two RAIDs (complete
> dmesg [1]).
> Installed with 5.2 from "miniinst.iso" and updated with fresh sources
> from CVS, -r RELENG_5_2. buildworld and buildkernel went fine. Just
> after reboot with the new kernel and after "installworld" the system
> died with some strange SCSI-Errors, kernel panic. "Calling boot() for
> CPU1 ... " something like that. Then the well known few lines, syncing
> disc etc ...
> While syncing disks there were some more SCSI-Errors, Timeout after ...
> 75 ... 255 ...  seconds and so on, no more acting drives.
>
> I powercycled after 7 minutes, the console was messed up with the
> scsi-errors, so I have no more details about the panic, powercycle
> flushed the dmesg .oO(why not dumping kernelmessages to an USB-stick??)
>
> System rebootet, but with a nearly empty /usr/sbin.
> Retry of "make installworld" failed, because it wants to copy some tools
> in an temporary dir in /tmp and didn't find some tools in /usr/sbin.
> Managed that manually, installworld succeeded.
>
>
> Kernel ist a GENERIC with only
> -cpu            I486_CPU
> -cpu            I586_CPU
> +#cpu           I486_CPU
> +#cpu           I586_CPU
> ( BTW: has anyone a perfect DELL PE2650 kernelconfig for 5.x? )
>
> Anyone some hints about the panic?
> How to avoid this?
>
> Regards,
> Raphael Becker
>
> PS: The actual dmesg.boot:
> [1] http://rhb.uugrn.org/temp/dmesg.boot.pinserv4

Without seeing the actual SCSI errors it is hard to say, but I was
experiencing a very similar issue with my PE2650.  Disabling write caching
on the RAID 1 container cleared it up.  I do not know if that is the best
way to go about it, but this server had to be in production pronto, so I
was happy.


:Fuzz