System freeze....need help with diagnostics

[heinkel_111]

Hi
I am not really getting anywhere on my computer right now because when I log in the system hangs up almost completely not a long time after login ( less than 5 minutes in general).
When the system stalls, I can switch to console with Ctrl+Alt+Del+F1 or write new commands in a konsole window. For some reason windows continue to give output to screen until a litle while after input becomes impossible.
I have been able to run "top" in a command window and I cannot see anything like a runaway process taking up all my memory or CPU. XORG log shows nothing particularly alarming (some kind of bogus length for keyboard description but can that cause this serious problems?)
Having checked these things I am at a loss as to what to do next. Someone suggested on the IRC channel that I do a memory test but I really don't think that is a problem as this system is dual boot and I just finished a 1 hour round of multiplayer game with lots of graphics and sound effects.

What to do to find the problem?
Please help me with brainstorm!

[heinkel_111]

Re: System freeze....need help with diagnostics

I am trying to compile a list of things to check:

1. Set up a Konsole window to run "top" while system freeze occurs. Found nothing special (no processes consuming lots of CPU or memory are apparent).
2. Checked X.org log files, found nothing special except a report on bogus length for a keyboard description. Is that bad? Could it cause desktop lockup?
3. Out of suspicion that a full harddisk can create unexpected and not easily explainable problems, I checked hard disk partitions with

df - h

Result showed all hard disk partitions to have more than enough free space. Full harddisks should not be the problem.
4. Memory. Someone mentioned this, I didn't think it should be the problem because the machine runs resource-intensive applications (games :P) well on the windows XP side of my dualboot.
Currently checkin this using the memtest option at boot (the standard one in the grub menu). This will take a while but I can do it now as I am supposed to do some work-related work on a laptop while sitting at the same desk. I'll just let it run while I sit here and work...currently it is at 5 % with no errors.
5. Graphics driver:
I have a version of Nvidia driver that is one newer than the one in the kubuntu repos. I could check if that is the problem (i think) by rewriting my xorg.conf to use the nv driver instead and see if I still get the system freeze problem.

What else can I check for finding the problem that causes system freeze?

[klerfayt]

Re: System freeze....need help with diagnostics

run memtest overnight to be sure

[heinkel_111]

Re: System freeze....need help with diagnostics

I ran 7 laps on the memtest and finished with 2 errors on the first run through and apart from that, none. While I guess it is possible that this causes problems once in a while, it does not appear to be the cause for total lockup of my desktop.

The fact that the errors were not repeating indicate that there may be instabilities in power supply or something else which at that point in time triggered a memory error, I guess?

[dibl]

Re: System freeze....need help with diagnostics

there may be instabilities in power supply

Hey, that could cause instability in the output of the graphics card, too!

[heinkel_111]

Re: System freeze....need help with diagnostics

I agree with that but it should be more random and have no links to events like resizing windows etc.

And furthermore, shouldn't it be equally problematic when using (blush) WinXP?

These problems are linux specific or very much more sever on linux compared to windows. That is my best indication of a "software" rather than a "hardware" root cause for my problems.

[dibl]

Re: System freeze....need help with diagnostics

Well, device drivers translate the signals to the hardware, so it seems possible to me that the Linux behavior could vary from the Windows behavior, in this regard. When you're resizing windows on screen, you are varying the demands on the graphics card, I would think, and varying demands would change the power consumption.

[heinkel_111]

Re: System freeze....need help with diagnostics

Replying to self here with more info on anatomy of recent crash.

bootup+ login ~ as normal
starting up and running konsole window (no apparent problems, full functionality)
starting up and running konversation (also with no apparent problems)
starting digikam. initially no problems
starting konqueror to read a documentation page for digikam
trouble starts when opening moving digikam window around and maybe doing some resizing.
The Digikam and konqueror windows do not redraw in any sensible way (many blank areas, the windows are frozen)
When returning to konsole window I can give input, but there is no output, cursor stalls as I hit enter
Several times something like this has happened

I suspect that some graphical kde component may be having a very hard time.

[heinkel_111]

Re: System freeze....need help with diagnostics

Adding one more test to my list of things to test when the system freezes up:

Check if there is something gone wrong with KDE settings:
Create a new user and run same applications as far as possible using that user.

If you do't crash in the same manner it is an indication that the problem may be related to configuration choices and files related to the user account that you usually use...

(YES if you have noticed it I am turning this thread into a draft for a "Guide on how to troubleshoot Lock-up Kubuntu Krisis!). Please keep contributing!

[heinkel_111]

Re: System freeze....need help with diagnostics

I think I have found that there is something wrong with my KDE. When I created a new account log in as another user than the one I usually use, my computer works OK, with no system freezes.

I have had my computer up and running for more than 60 minutes now and that compares with something like 60 seconds on the regular account.

Furthermore, due to most of the problems used to appear when resizing and moving windows (especially konqueror, also digikam) I think it may be something related to these more graphic applications. But what?

I am toying with the idea of copying my KDE setups one by one from the alternative user account, but the problem is that it is very time consuming and also difficult to avoid creating more problems, loosing data and such. Advice anyone?

[klerfayt]

Re: System freeze....need help with diagnostics

I ran 7 laps on the memtest and finished with 2 errors on the first run through and apart from that, none.

you should get no errors, what test number was it?

[heinkel_111]

Re: System freeze....need help with diagnostics

I agree I should get no errors but tests themselves are sometimes prone to error. The errors in this case was reported for test number 5, run 1 of 6. If I strongly believed that memory problems were to blame for my troubles then I would have continued the memtest and let it run all night, to see if I got more reliable indications of errors.
However, I have narrowed down my search cone now: as running my kubuntu with another (fresh) user account seemed to be unproblematic, there are a lot of potential causes for problems which can be written off almost completely: Everything common to both accounts, such as hardware, drivers and installed programmes. I am left with causes related to the specific user account, and after some more trial and error I am now quite confident the problem may be Konqueror related (it also shows up with Digikam but maybe digikam uses the same base kde component that is affected by these problems).

[klerfayt]

Re: System freeze....need help with diagnostics

you don't get it; memtest should give 0 zero errors

[dibl]

Re: System freeze....need help with diagnostics

I agree with klerfayt.

Do you have any Live CD for another OS? Are you using a proprietary video driver or VESA or one of the open ones?

If you could boot a live CD that doesn't use KDE, like E-Live or Ubuntu, and run that for awhile, it would be interesting to see if there's a problem.

[heinkel_111]

Re: System freeze....need help with diagnostics

I appreciate your feedback and contributions here, gentlemen.

However, the problem here is: my system freezes up only while running kubuntu from a specific user account.

A problem or not a problem with memory appears to be of no consequence to this, because my computer apparently runs 1) kubuntu from another user account 2) windows xp problem-free.

Furthermore I have used the time since my previous post elaborating on the issue, and I am becoming more and more confident that the specific user account has a troble related to konqueror or some kde component used by konqueror to retrieve urls. Due to the fact that I use konqueror a lot and have a dual head display with 3200x1200 total screen size, there is always a window or three running konqueror open.

I previously thought that the problem occured due to me moving windows around. That appears not be the case, but I have a habit of checking my email and konversation windows while waiting for konqueror to load a window which could have led to me not noticing that every time the computer got a system freeze I was trying to open a website in konqueror.

If I was misleading any of my helpers here I apologize, but I was misled myself. I never had this kind of problem with konqueror before, and the other user account on this machine does not have that trouble either.

Furthermore, I don't think the trouble is network related as I am browsing kubuntuforums and posting this status report from the konqueror-troubled account, but using (heaven forbid :P) firefox.

I have tried the following trouble shooting steps that did not fix the issues:

mv .kde/share/config/konq_history .kde/share/config/old_konq_history
mv .kde/share/config/konquerorrc .kde/share/config/old_konquerorrc

Other places to look for user-account specific information that konqueror use to freeze up my entire system?

EDIT: and oh yea I know there may be hardware problems on my systems, but my computer's memory is far less corrupted compared to my own :P