PDA

View Full Version : X-windows stuck in treacle - help



urilabob
28th April 2006, 02:55 PM
My X-windows system has been behaving decidedly oddly for about the last week. I'm running FC5 on a dual-core 2GHz machine, 1G of memory. The system _mostly_ runs OK. But program startup for graphical programs takes huge amounts of time. For example, I just timed starting up a terminal session at 1min 20sec!! Yet once started, it runs fine. Starting a second terminal window from the first is instantaneous. I can't see any obvious problems in the logs (but maybe I don't know where to look).

The problem _could_ be related to kernel compilation; because of problems with the sky2 driver (marvell yukon and syskonnect NICs) in 2.6.16, I recompiled the 2.6.16 kernel with the sk98lin driver recently. Any hints on any options I might have missed would be appreciated. However I think the problem actually appeared a day or so later, so my best guess is that the problem and the kernel recompilation are unrelated.

wintermute000
28th April 2006, 03:26 PM
hmmm, u could check (as root) /var/log/messages

also try good ol' dmesg | tail as soon as you start something and see what msgs it comes up with.

Also how about your video card drivers? (don't say ATI, if so just do a quick search on the forum for mega painful stories from the likes of me! hehe)

you could also try installing an alternative desktop and go from there. I recommend XFCE (you can do it via yum and select it in sessions at the gdm login screen, no pain required), its lite and fast as all heck. Your problem may not be X but Gnome / KDE (as much as I love my nice looking gnome desktop I know its bloating something shocking hehe, when I was running XFCE I was getting easily 20% faster performance.

urilabob
28th April 2006, 03:38 PM
Thank you for the suggestions..... dmesg doesn't help, but checking /var/log led me to a very interesting discovery.... all my logs stopped 5 days ago (but according to the service configuration
gui, syslogd and klogd are running....). I also got a strange message the same day

/etc/cron.weekly/makewhatis.cron:

sort: write failed: standard output: No space left on device
sort: write error


....but df -a shows most devices under 50% full (a specific file system, /tftpboot, is 94% full, but
that filesystem is not used for any system purposes). In particular, /var shows as 50% full.

wintermute000
28th April 2006, 04:02 PM
is there any clue as to what is trying to write? (swap space? making guesses hehe)
what happens when you restart your log daemons?
has this problem always been there or did it start 5 days ago? sorry I'm no expert but you may want to google what that cron job was if this has any correlation.