Spontaneous reboot problem

[R D L Smith]

On 21 Jan, 2005, at 11:56, Colin J. Raven wrote:

> On Jan 21 at 11:36, R D L Smith asked the panel:
>
>> I have a teramac R310-1U from DNUK running 5.3-RELEASE-p1. It is the 
>> only
>> server I have running 5.x. ISTR that the SiS chip set prevented me 
>> loading
>> 4.x.
>>
>> The uptime of this server is good unless its under heavy load. When 
>> MySQL
>> (with InnoDB) is being worked hard, the server will reboot after a 
>> few hours.
>> However, build world will not crash it.
>>
>> I have added the following to rc.conf:
>>
>> dumpdev="/dev/ad0s1b"
>>
>> and the following to the kernel configuration file:
>>
>> makeoptions DEBUG=-g
>> options     KDB
>> options     PANIC_REBOOT_WAIT_TIME=-1
>>
>> I have also connected a serial console to the serial port and logged 
>> in as
>> "root".
>>
>> I never get a dump file and I never get a message at the console. I 
>> don't
>> know what else I can do other than strike this hardware vendor off my 
>> list
>> and dump the server. This machine has already had its motherboard 
>> replaced.
>
> What about the power supply?
>
> There has been quite a lot of traffic of late concerning so-called
> "spontaneous reboots" and so far - *IIRC* - none of the incidents
> were directly attributed to causes _other than_ balky power spupplies,
> and in one recent case interestingly, a defective UPS!!
>
> I think it's probably worth shoving a new PS in there, these days they
> are far from expensive, in order to see if that eliminates further
> instances of the problem.

I did consider that the previous time I was trying to get to the bottom 
of the problem. The problem is easy to reproduce, simply by starting 
mysqld and getting it to replicate.

I do not normally blame FreeBSD. I have been using it very successfully 
for years in a production environment. This is the first server I have 
had with this particular chip set, and I have read about problems with 
the chip set in the past.

It seems to be disk related - it could certainly be a hardware fault 
(cables or disk drive - as I said before, the motherboard has been 
replaced already) - if anyone out there is successfully using this 
hardware with FreeBSD, I could at least eliminate a device driver 
related problem.

My other concern is that I do not understand why I do not get any 
diagnostics.


> I suppose much depends on whether or not you really want to -as you put
> it - 'dump the server'. Should you elect to do so, please feel free to
> ship it across the North Sea to me, I'll be happy to pay for the
> shipping, deal with its little idiosyncracies and give it a "second
> life" :-)

I will keep that in mind ;-)

--
R D L Smith
Network Systems Director
Satamatics Limited
Gloucester Road, Tewkesbury, GL20 5TT, United Kingdom
T: +44 1684 278610
F: +44 1684 278611
E: rdls at satamatics.com


________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________