Sun Blade 2000 crashing mysteriously

[gtay]

Hi there,
We've got a Sun Blade 2000, it's pretty much stock except for a third party SATA card. It has been sitting doing nothing for months, but recently started crashing most days. I can't see much on the OBP as all the interesting information must have scrolled past. There are no core dumps in /var/crash/hostname and nothing of note in /var/adm/messages. To all intents and purposes, it looks fine. I've also run SunVTS for 2 days solid, and it reports no faults.

So I'm wondering where I go from here... Maybe my swap device is not big enough to hold the dump? But if that's the case, what exactly can I do if I've no free slices on the disks to dump to? Is it possible to get it to truncate dumps so at least I see some of it? And finally, is it possible to somehow scroll up the OBP to see more meaningful error messages?

One thing I've noticed is that if the machine is active, then it seems not to crash, but if it sits doing nothing, a crash within 24 hours is likely. This would lead me to believe maybe something is going to sleep on the PCI, maybe the third party card, but we have turned off all power saving.

Anyone got any ideas?

All the best

Garry

 

[Annihilannic]

Try connecting a laptop running some terminal software to the serial console with a null modem cable, that way you'll have the dump messages captured in the software's buffer (enable logging if you are expecting a lot or if the software doesn't buffer much).

Annihilannic.