Using an updated rawhide upgrade. I have noticed a few pecular items with the operation of the system.
The system seems to slow down to a snails pace at times and may be unresponsive. The processor seems to be overburdoned at times and allows my usual 45 degree temperature for the processor to escelate to higher than usual temperatures. Today, the temperature got to 80 C and the system shut itself off. I run acpi.
Selinux from the upgrade to FC3 to rawhide from Feb 13th mix left the logs stagnated and inaccessable for days past the installation. Using touch /.autorelabel got the logs working again. Later, experimenting to use the strict policy (vs. targeted) and relabeling the system would not allow me to start x from runlevel 3. I could launch X if setenforce 0 was run before running the startx script.
selinux-policy-targeted-1.21.14-1 policycoreutils-1.21.18-1 checkpolicy-1.21.4-1 selinux-policy-strict-1.21.14-1
General preferences, I like the Apps, places and desktop divisions within the menu structure. It just might be good to not add/remove and juggle the menu structures around as frequently. Hopefully the reorganizing settles down to more subtle structure changes.
Also, there seems to be a problem with icon registration for certain apps within the menus. gnucash and KDE items do not show up on the menu. If you add the program to the panel launcher, it errors out as having no icon. If you check the properties, the icons are there. When the item was checked for a second time, there is no icon selected in the properties for the item.
The largest pecularity is a white flash when starting X from runlevel 3. The screen is completely white for an instant, then the loading of GNOME continues with its normal sequence.
Overall, GNOME seems to be slower than it was for FC3. KDE was very slow after launching mozilla. KDE seemed pretty stripped down from previous usage of the windows manager for short periods.
Is anyone else experiencing similar issues with current rawhide?
Jim
Jim Cornette wrote:
Using an updated rawhide upgrade. I have noticed a few pecular items with the operation of the system.
The system seems to slow down to a snails pace at times and may be unresponsive. The processor seems to be overburdoned at times and allows my usual 45 degree temperature for the processor to escelate to higher than usual temperatures. Today, the temperature got to 80 C and the system shut itself off. I run acpi.
CPU temp seems like a good indicator that some process has gotten stuck in a loop and is using 100% cpu. Using top will usually show the culprit.
Last time that was happenning to me it was /usr/bin/esd. It is impossible to "rpm -e esound" because everything claims to depend on it, so my solution was to "mv /usr/bin/esd /usr/bin/esd.broken" and reboot. Sound seems to work much better now.
John
John Ellson wrote:
CPU temp seems like a good indicator that some process has gotten stuck in a loop and is using 100% cpu. Using top will usually show the culprit.
Last time that was happenning to me it was /usr/bin/esd. It is impossible to "rpm -e esound" because everything claims to depend on it, so my solution was to "mv /usr/bin/esd /usr/bin/esd.broken" and reboot. Sound seems to work much better now.
John
It amazes me to see all the depencies on ESD. Most of the alleged depencies are apps that have optional ESD support (which can be turned off upon recompilation). When is this abomination finally going to be killed for good?
-sb
John Ellson wrote:
Jim Cornette wrote:
Using an updated rawhide upgrade. I have noticed a few pecular items with the operation of the system.
The system seems to slow down to a snails pace at times and may be unresponsive. The processor seems to be overburdoned at times and allows my usual 45 degree temperature for the processor to escelate to higher than usual temperatures. Today, the temperature got to 80 C and the system shut itself off. I run acpi.
CPU temp seems like a good indicator that some process has gotten stuck in a loop and is using 100% cpu. Using top will usually show the culprit.
Last time that was happenning to me it was /usr/bin/esd. It is impossible to "rpm -e esound" because everything claims to depend on it, so my solution was to "mv /usr/bin/esd /usr/bin/esd.broken" and reboot. Sound seems to work much better now.
John
Using top to view memory usage, the programs gnome-vfs-daemon, gnome-settings, nautilus and X seem to be the leading processes for consuming cpu time. I added the cpufrequency control applet to my panel and am able to control the cpu frequuency to prevent the 80 C temp then shutdown. These listed processes take 28 -32 percent each in usage percentage. X takes around 1 to 6.6 percent usage range.
Even though the computer hovers at high speeds, it takes a long while for a responsive desktop. lowering the speed does not help or hinder the system responsiveness. ESD is not listed in top.
Thanks for the hints regarding top. I'll use it to see what's going on when the system is extremely slow. (it responds like a 100 MHz processor, on initial startup, but is a 1.5 GHz processor)
Jim
Jim Cornette wrote:
Using top to view memory usage, the programs gnome-vfs-daemon, gnome-settings, nautilus and X seem to be the leading processes for consuming cpu time. I added the cpufrequency control applet to my panel and am able to control the cpu frequuency to prevent the 80 C temp then shutdown. These listed processes take 28 -32 percent each in usage percentage. X takes around 1 to 6.6 percent usage range.
Woh. Clearly there is something going wrong here. From your description it seems between X, gnome-vfs-daemon, gnome-settings, and nautilus you are using 100% CPU? This would make everything else crawl. So now the question is: Why are these processes running out of control?
Even though the computer hovers at high speeds, it takes a long while for a responsive desktop. lowering the speed does not help or hinder the system responsiveness. ESD is not listed in top.
If you are at 100 CPU usage constantly then there is no mystery here except why.
Thanks for the hints regarding top. I'll use it to see what's going on when the system is extremely slow. (it responds like a 100 MHz processor, on initial startup, but is a 1.5 GHz processor)
Jim
-sb
Stan Bubrouski wrote:
Jim Cornette wrote:
Using top to view memory usage, the programs gnome-vfs-daemon, gnome-settings, nautilus and X seem to be the leading processes for consuming cpu time. I added the cpufrequency control applet to my panel and am able to control the cpu frequuency to prevent the 80 C temp then shutdown. These listed processes take 28 -32 percent each in usage percentage. X takes around 1 to 6.6 percent usage range.
Woh. Clearly there is something going wrong here. From your description it seems between X, gnome-vfs-daemon, gnome-settings, and nautilus you are using 100% CPU? This would make everything else crawl. So now the question is: Why are these processes running out of control?
I glanced through the logs and there were no entries for when the trouble was most noticable. I mainly asked to see if this behavior is common or maybe my install is corrupted enough to cause the runaway processes.
Even though the computer hovers at high speeds, it takes a long while for a responsive desktop. lowering the speed does not help or hinder the system responsiveness. ESD is not listed in top.
If you are at 100 CPU usage constantly then there is no mystery here except why.
The processes are not running as wild as they were yesterday. I applied the latest rpms from rawhide and things settled down some. The cpufrequency did upshift and downshift a few times when the cpu usage climbed to higher levels. It did not peg at 1.52 GHz and stick, causing high temps and eventually a shutdown. The processes mentioned earlier do seem to hover around 20 percent for the nautilus gnome-vfs-daemon (better behaved than yesterday) and gam_server is not even listed in top today. It seems that gam_server was the main culpret. At least the average load is down to mid 80s to low 90s and the processor speed is back down to 662 MHz (43 %) like was normal for the computer since severn.
Thanks! At least I am now more familiar with program cpu usage and what effects it has on other processes. I was surprised that the memory usage was low, but the cpu usage was high for the programs that caused the system slowdown/ processor max speed and system shutdown.
Jim
On Thu, 2005-02-24 at 01:48 -0500, Stan Bubrouski wrote:
Jim Cornette wrote:
Using top to view memory usage, the programs gnome-vfs-daemon, gnome-settings, nautilus and X seem to be the leading processes for consuming cpu time. I added the cpufrequency control applet to my panel and am able to control the cpu frequuency to prevent the 80 C temp then shutdown. These listed processes take 28 -32 percent each in usage percentage. X takes around 1 to 6.6 percent usage range.
Woh. Clearly there is something going wrong here. From your description it seems between X, gnome-vfs-daemon, gnome-settings, and nautilus you are using 100% CPU? This would make everything else crawl. So now the question is: Why are these processes running out of control?
That's the question. After updating to the most recent rawhide yesterday I've started to see this too. (I update to rawhide each day without fear or fail, so it was something on 20050224 that did it I guess).
The first time I couldn't figure out what it was, but the second time I found killing gnome-panel fixed it (I'd tried gamin, gnome-vfs-daemon, gnome-settings-daemon and Nautilus before this.)
I'm not sure if this is related, but I can get a similar (same result) by doing the following:
1. Open Applications > Accessories > Dictionary 2. Type Role into the Word field and hit enter 3. Click on the hyperlink Roll (this should bring up the meaning of Roll)
Doing this sends my CPU 100% (about 30% user and 70% system)
Closing the Dictionary fixes this immediately.
Rodd
Rodd Clarkson wrote:
On Thu, 2005-02-24 at 01:48 -0500, Stan Bubrouski wrote:
Jim Cornette wrote:
Using top to view memory usage, the programs gnome-vfs-daemon, gnome-settings, nautilus and X seem to be the leading processes for consuming cpu time. I added the cpufrequency control applet to my panel and am able to control the cpu frequuency to prevent the 80 C temp then shutdown. These listed processes take 28 -32 percent each in usage percentage. X takes around 1 to 6.6 percent usage range.
Woh. Clearly there is something going wrong here. From your description it seems between X, gnome-vfs-daemon, gnome-settings, and nautilus you are using 100% CPU? This would make everything else crawl. So now the question is: Why are these processes running out of control?
That's the question. After updating to the most recent rawhide yesterday I've started to see this too. (I update to rawhide each day without fear or fail, so it was something on 20050224 that did it I guess).
The first time I couldn't figure out what it was, but the second time I found killing gnome-panel fixed it (I'd tried gamin, gnome-vfs-daemon, gnome-settings-daemon and Nautilus before this.)
I'm not sure if this is related, but I can get a similar (same result) by doing the following:
- Open Applications > Accessories > Dictionary
- Type Role into the Word field and hit enter
- Click on the hyperlink Roll (this should bring up the meaning of
Roll)
Doing this sends my CPU 100% (about 30% user and 70% system)
Closing the Dictionary fixes this immediately.
Rodd
The thing that got my computer into a more usable state was to undo the prelinking for all with 'prelink -uav' - Once the prelinking was undone, the random slowdown of programs does not seem to be a problem any longer. My cpu usage numbers are now down in the 2 to 3 percent range. The loading does not seem to be any better or worse than before though.
At first, I tried to 'prelink -av' the system and it was no different than before trying to prelink processes.
Using the dictionary program did not make any difference to my cpu usage problem.
Your results might vary from my tests. This "seems" to work, but I just guessed as to what to try next.
Jim
-
Jim Cornette -
John Ellson -
Rodd Clarkson -
Stan Bubrouski