Mike McCarty wrote:
Dougie Nisbet wrote:
Um, if the current release is the problem, then it will never run stablyI thought you said it has been running without a change for some time. I quote your exact words:A server which has been running steadily for years is beginning to reboot. To the best of my knowledge, nothing has changed. It is a dual-processor PIII. It runs stable.Therefore, my dear Watson, logically, the problem is not the load, since that has not changed. Furthermore, the symptoms sound more like hardware, anyway.
I'm confused. I'm not suggesting it is load. And yes, the symptoms do sound like hardware. But the timeline for this machine is interesting. I had a problem with it crashing when I upgraded from 2.4 to 2.6.4 about two years ago. I couldn't track it down so I reverted to 2.4. A few months later I upgraded it 2.6.8 and that has run stably for well for a year.
Now it starts playing up. And nothing has changed in the OS. So everything points at hardware.
again. More to the point, it's easy to test. It's already rebooted twiceIf you want to argue, go to someone else. If you want expertadvice, then listen.
I'll put it another way. I won't have time to physically spend time on this box doing the sort of things you suggested for a few days. I think hardware is far more likely to be a the problem than the kernel version. But a hunch is a hunch, and I'm entitled to explore it. It costs me 20 seconds to rerun lilo and boot of a different kernel. With the box rebooting several times a day it should eliminate the kernel from the enquiries PDQ. That way, I'll have eliminated the impossible, and can explore the improbable.
I'm not going to try to fix a machine which has a bunch of fiddling going on in the background.
Fair enough. But with an attitude like that I'll happily live without your advice. Which is a shame, because what you said in your original post was useful.
Dougie