Mysterious reboot, suspecting hardware issue

[Robert Slade]

On Tue, 2006-02-28 at 04:25, Olivier Nicole wrote:
> Hi,
> 
> Running an Amanda server on a dual P III, with 256 MB ram and dual
> Symbios SCSI adapter. Part of the hard disk are mounted in RAID with
> vinum on the first SCSI adapter. The Tandberg tape drive is on the
> second SCSI adapter. FreeBSD is:
> 
> FreeBSD amanda.cs.ait.ac.th 4.11-RELEASE-p14 FreeBSD 4.11-RELEASE-p14 #3: Thu Feb 23 14:03:05 ICT 2006     root at amanda.cs.ait.ac.th:/usr/obj/usr/src/sys/SMALL  i386
> 
> Recently (mid February) it started rebooting around 01:20, that is
> about 40 minutes after amanda started the nightly dumps of the various
> servers on my network. It's about time when amanda reaches full speed
> and the machine may be busy doing some high gzip compress.
> 
> The first reboot occured while I was running 4.10-RELEASE-p19 so it
> cannot be related to the version of FreeBSD.
> 
> That machine has been running like that for about one year without
> any problem.
> 
> I set dumpdev="/dev/rda0s1b" in /etc/rc.conf, expecting to get a
> kernel dump in time of the crash, but I got nothing.
> 
> - did I missed something?
> 
> - how to prevent the machine from rebooting when it crashes?
> 
> Other idea?
> 
> Best regards,
> 
> Olivier

Oliver,

There has been problems with the SMP kernel in the past, but it was
supposed to be fixed around 5.4. From memory there was a patch for 5.3
which was incorporated in 5.4 and 6.0. I say supposed as there have been
reports of random reboots with 5.4 and 6.0 running SMP - see the
archives. The usual response is that it is a hardware related, but I am
not too sure. In my case it was on a dual redundant power supply etc
quad processor machine which ran the diagnostics with no problem for
several days. It looks like a load problem to me.

I ran out of time trying to fix the machine and had to go down a
different route.

Hope this helps a little

Rob