I'm having a problem with one of the linux machines. It's running a RedHat 7 (i believe) and is used frequently by several users at a time, mostly just telneting and vi'ing documents, and maybe internet browsing. But the problem is that at random times (and random because no indication of overload) the box just feezes up and you can't do anyhting to it but restart. Now I have no idea, what is going on with it. Some ideas were, maybe /tmp folder was running out of space, because users are telneting and vi'ing, and also maybe some application is craping out (samba?). I've checked the /var/log/messages, but it doesn't seem to indicate any problems (or maybe i just can't see any).

Any ideas on what may be causing these lockups, or how can i trace the problem?

TIA

try emptying /tmp and, really, RH 7 is really old now, it just might be an age thing. try a newer install, such as fedora.

titanium_geek

/tmp running out of space can do some scary stuff (especially to KDE) but it hasn't crashed my machine yet. I am running the slack though. Also, normally, a crapping out application shouldn't be able to take down your machine (unless you are running Windows, hihi). Maybe you should check your memory and hd for bad blocks, 'cause that'll definitely do it.

well, how do i see what amount of space /tmp was allocated and how much it has used? It's not a separate partition, it's a part of hda partition. After the machine freezes/crashes, sometimes the disk check finds bad blocks and sometimes it doesn't. So i'm not sure how to go about emptying out /tmp, because what if it gets full again fast? Can I monitor it?

you should be able to cd to the directory and rm - but I would be darn careful about doing that. umm.. does df help you?

titanium_geek

no df only shows the size of the actual partitioning of the hd, but not the filesystem

cd /tmp
du -sh

Since /tmp is on the root filesystem in your case, it will expand to fill all free space on the root filesystem and that would definitely cause bad things to happen (which is why untrusted users should not be allowed to write to the root fs unless there are quotas).

Also, bad blocks being found is bad -- you can try running badblocks directly. Running memtest would probably also be a good idea.

Is there any way to bring up the console (/dev/console) to see if the system is dumping some errors right before it freezes?
Any other place errors might be dumped after it locks-up

In var/log/messages, this is what is there right before the system locked up, as you can see the fourth line is the actual restart. The first line probably is nothing, since it happens 22 minutes before the lock-up.
Does this provide any clues to my problem?

Jun 30 09:48:53 localhost kernel: udf: registering filesystem
Jun 30 10:10:29 localhost kernel: kernel BUG in header file at line 254
Jun 30 10:10:29 localhost kernel: ------------[ cut here ]------------
Jun 30 10:15:41 localhost syslogd 1.4.1: restart.

perhaps you could try recompiling your kernel? That looks to me to be standard programming bug notification- there's a problem with the kernel, perhaps corrupted with the bad blocks.

titanium_geek