James Wilkinson wrote:
I had someone restart the machine four days ago. Since then it's been sitting in the logon prompt, and it's still running. When I came into the office this morning, the screen was in power saver mode. This seems consistent with the theory that the culprit may be the screen saver. On the other hand, if the screen saver isn't running, the machine isn't doing much...
I'm having a bit of trouble parsing that. Are you saying that if the computer is sitting there, with no-one logged in at the console, and there's not even a screen saver to burn CPU cycles, then the computer is basically idle, and you suspect that the problems might be load related?
Basically. The person who restarted the machine just rebooted it and walked away. That means that it was sitting in the X logon screen for four days. (I assume, but haven't taken the time to check, that that means that the screen saver was not running. - when I came in this morning, the display was in stand-by mode, and when I woke the system up, it was still in the logon screen.)
I'll take your word for it that the computer is idle if no-one is logged in at the console (it's certainly not true for all Unix-like systems).
Yes, I know. This is a system that I use for software development, so when I'm not actively doing that, the system is basically idle. It runs an FTP server and an NTPD and a few other thing, but it's main purpose in life is to hang around and wait for me to do software development...
It sounds like a recent Pentium 4 or Celeron machine. These are known to be (electrical) power-crazed, and can output a *lot* of heat. But they have thermal protection in the CPU: they throttle down speed and power consumption if the temperature gets too high.
It's a 3G Pentium 4 w/ hyperthreading. Other posters have suggested that the problem is related to hyperthreading. I know that multi-processor / threading issues are often load related, so I was guessing that the added load of the screen saver might be a factor.
I suppose that it *is* possible that on a badly designed system the extra heat from a loaded Pentium 4 could induce temperature related problems in the rest of the system.
Well, it's a commercial IBM product, I doubt that the problem is design related...
But my money would still be on the screen saver (if I was a betting man...)
This sounds like an ideal candidate for the SETI at Home client, or something similar: get the machine doing useful work in the background, and see if it affects system stability.
Might be worth a try!
James.
Regards, Eric