On Wed, Nov 23, 2005 at 07:24:52AM +1100, Felix Karpfen wrote: > On Sun, 20 Nov 2005 21:13:09 +0000, Thomas Adam wrote > The root directory switched to "read-only" mode and the computer > could not be shut down cleanly. In fact it could only be shut down by > switching off the power.
This is symptomatic of a much larger problem. Typically if this happens, it's because there's been a disk error, and that mount has remounted the partition in RO mode -- most recent distributions employ this methodology, that I have seen. > And after switching the power back, I had to run fsck manually to let > the computer clean up the mess - which took quite a time. And I lost > several of my few remaining hairs while biting my fingernails and > wondering whether the crash was beyond fsck's ability to fix. Fsck is very good at what it does, Felix. I usually find that running it four times consecutively helps. It really will try its best -- and if it can't fix your partition, you can generally be assured your hard disk is on the way out. Indeed, if you: tail -f /var/log/messages ... do you see any I/O errors? > I have never attempted to use a "corefile" but, from a very distant > memory, I believe it sits in /var/log/wtmp. Is that correct? Not necessarily. Usually corefiles are dumped in the CWD at the time of the crash. > While running fvwm2 with the Gnome Session Manager opened up > attractive new routines and possibilities, I think the price is too > high. In 5 years of Linux, I have never encountered crashes such as > the two crashes that I have managed to survive in the past few weeks. Check your logs -- these symptoms, as I say, sound like something on a much larger scale than just a software glitch in this instance. Session management within FVWM is very good. I've used it. I don't like it, but it works. :) -- Thomas Adam -- I'm brutal, honest, and afraid of you.