sudden lockups w/ 2.4.26 on an E450

[Mike Edwards <sauron-debian-sparc@psychology.rutgers.edu>]

A few weeks ago, our E450 (quad 296 MHz USIIs, 4 GB memory) started locking
up with nothing to indicate why it died.

I'm using serial console on this beast.  If I send break to get into the
PROM, the cursor moves down a line (sent from the E450, so it DOES respond),
and then... nothing.  Never get an 'ok' prompt, and sending additional breaks
fails to elicit a response from the E450.

When the box first hangs, I've noticed that it will still respond to ping.
In addition, one can connect to ports that are open - however, the services
using those ports never respond.  Once break is sent, the box stops
responding on the network (as one would expect), and at that point, it must
be power cycled before it will work again.

This issue is starting to occur more and more frequently, as well.  The
first two times happened about 2 weeks apart.  This last one happened a few
days after the previous hang.

Typically, nothing is printed on the console when the lockup occurs.
With the last one, though, I found the following message:
request_module[net-pf-10]: fork failed, errno 1

Any ideas?  I'm not quite sure if this is software (kernel?) or hardware at
fault at this point.  It's running 2.4.26-sparc64-smp (dpkg reports version
37 for kernel-image-2.4.26-sparc64-smp).

Thanks.


-- 
Mike Edwards <sauron-debian-sparc@psychology.rutgers.edu>
System Administrator
Psychology Department, Rutgers University, Newark campus
Office: 973.353.5440 x246
Cell  : 732.407.2502