Troubleshooting a Freeze

[Matthew Seaman]

On Sat, Dec 20, 2003 at 12:30:22PM -0500, Ryan Sandridge wrote:
> Are there any general or specific tips on how to figure out what is  
> causing FreeBSD to just freeze up?  The machine is not at my physical  
> location, and I log in using ssh.  I've checked all the logs in  
> /var/log for some kind of clue, but I haven't found anything helpful.   
> The machine has worked fine for over 6 months, until the last few days  
> it freezes up at apparently random times, normally within the first 12  
> hours of being up.  While many ports have been added over the past 6  
> months, nothing was added on or around the time I started experiencing  
> the problem.

Sounds hardware-ish to me.  Thermal cut out?  Fans going awol?  Power
supply going flakey?  Try using the sysutils/xmbmon port to monitor
fan speed, CPU and motherboard temperature, PSU voltages. (Works well
if you hook it into rrdtool or mrtg to produce graphs against time).
Get the NOC people to check that the machine is getting an
unobstructed flow of cooling air and that filters are not clogged by
dust.  While they're there make sure there isn't anything printed out
to the system console when the system goes down.

If that's all OK, try running a few cycles of memtest86
(http://www.memtest86.com) to see if it can turn up any anomalies.  

Failing finding any memory problems, then try enabling crash dumps by
setting eg:

    dumpdir="/dev/da0s1b"

in /etc/rc.conf (Nb. requires that your swap partition contains
slightly more space than the amount of RAM in you machine).  Compile a
kernel with debugging symbols.  When the machine conks out again, see
if you can get a core image.  Use gdb to try and get a traceback
showing what went wrong.  If that doesn't give you any clues as to how
you can fix things, submit a PR with the traceback information and
anything else pertinent, and see if any of the developers have any
suggestions.

   http://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/kerneldebug.html

explains all about kernel crash dumps.

	Cheers,

	Matthew

-- 
Dr Matthew J Seaman MA, D.Phil.                       26 The Paddocks
                                                      Savill Way
PGP: http://www.infracaninophile.co.uk/pgpkey         Marlow
Tel: +44 1628 476614                                  Bucks., SL7 1TH UK
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-questions/attachments/20031221/926d2400/attachment.bin