On 10/3/2011 21:03, ..I'd rather be coding ASM! wrote:
How about a terminal program (tip, putty, minicom, telix, JrComm, etc..) with a massive scroll back buffer logging to disk and let it run? Have you tried provoking it with heavy network, or memory usage, or disk activity? I have an IPX with a weitek powerUp! cpu + rom that was aaround 2003 getting a bit on the unstable side with similar lock-ups. I ended up de-soldering and re-capping the entire board + PSU, and thus far touch wood it's not sneezed. Another thing that may be worth thinking about is an expect script or perl script driving the serial device that'll send escape sequences through?
The only problem with a terminal program logging to disk is that both serial ports are occupied. This machine is supposed to be my GPS based NTP server. If I unplug the GPS then it's not loaded like it normally would be. Then again if I unplug it and it goes for two weeks straight without dying then I know what the culprit is. I hadn't noticed the problem before because I was rebooting constantly to try modified kernels or restarting lots of software. I think I'll start by disabling gpsd (but leave ntpd running) and see if there's something going on in gpsd. Not likely but possible.
Recapping might be needed, not sure yet. There's no Weitek in this one but if it locks up again I'll end up opening and looking. Or I'll swap the motherboard with my other IPX and see if the problem goes away. I am having difficulty installing NetBSD on the second IPX for some reason. It locks up in the middle of the bootloader.