Page 1 of 2 12 LastLast
Results 1 to 10 of 11

Thread: Help with Diagnostics

  1. #1
    Join Date
    Oct 2005
    Beans
    21

    Help with Diagnostics

    I've recently installed Hardy, and it's going pretty well. No major complaints just yet. There's just something that irritates me every now and again: the cpu gets maxed, the computer freezes for just a couple of seconds, and returns to normal. Instead of asking if anyone else has had the same irksome little experience, I want to know how you diagnose problems of this nature (i.e. infrequent, unexpected, difficult to pinpoint). What software do you use? Are there any bash commands that could help me log cpu usage of processes, or any errors they output without monitoring specific processes one at a time? Thanks in advance.

  2. #2
    Join Date
    Feb 2007
    Beans
    9

    Re: Help with Diagnostics

    Have you tried the top command? This should tell you which processes are utilizing the most resources on your system.

  3. #3
    Join Date
    Mar 2008
    Location
    Copenhagen Denmark
    Beans
    722
    Distro
    Ubuntu 11.10 Oneiric Ocelot

    Re: Help with Diagnostics

    or htop. it is a tad better.
    Ubuntu 10.10 Maverick | ASUS A6Rp | Intel(R) Celeron(R) M CPU 420 @ 1.60GHz | 4 GB ram |
    Graphic Card: ATI Technologies inc RC410 [Radeon Xpress 200M]

  4. #4
    Join Date
    Aug 2006
    Beans
    841

    Re: Help with Diagnostics

    i had this problem right when gutsy was out.

    the culprit was the vino vnc server.

    run on a terminal ps -e |grep vino
    if its running, disable it. and see if your problem goes away.

  5. #5
    Join Date
    Oct 2005
    Beans
    21

    Re: Help with Diagnostics

    I tried top and htop (thanks to those who suggested them ). They should prove useful, but I'm more looking for a daemon that can help log these things over long periods of time. The problem occurs extremely sporadically, and I don't have the time (nor the cpu cycles) to open a terminal, enter in $top, and find the process responsible before the problem goes again.

  6. #6
    Join Date
    Jan 2007
    Beans
    Hidden!

    Re: Help with Diagnostics

    Daemonize top. It will take some CPU cycles every 2 seconds(change the -d 2 option below if you want some other interval), but only until you diagnose your problems.
    Code:
    top -b -c -d 2 > /tmp/topResults
    Just after you have your freeze-ups, look around the last part of the file /tmp/topResults and you might find your culprit.

  7. #7
    Join Date
    Oct 2005
    Beans
    21

    Re: Help with Diagnostics

    Thanks! That's exactly what I wanted!

  8. #8
    Join Date
    Oct 2005
    Beans
    21

    Re: Help with Diagnostics

    OK, I've tried it, and I was lucky enough to catch the bug before the file became too massive. Unfortunately, even though the bug went on for more than two seconds, it didn't log any usage over 10% (despite the CPU monitor showing it to be maxed out). I'm guessing whatever is taking all those cycles has a higher priority, so I might try the same process with a very low nice value.

    One thing I did find out is that I have a "zombie" process: python. Would that account for anything?

  9. #9
    Join Date
    Jan 2007
    Beans
    Hidden!

    Re: Help with Diagnostics

    In top's output, include the "wa" cpu percentage in your analysis. Your CPU monitor could be showing this as almost 100%. Which CPU monitor do you use?

  10. #10
    Join Date
    Oct 2005
    Beans
    21

    Re: Help with Diagnostics

    wa is minimal. "id" is larger, but I'm guessing it stands for idle. I use the taskbar widget that comes with ubuntu.

    I'm still yet to isolate the bug while logging. I used the command "sudo nice -n -20 top -b -c -d 2 > ~/topResults" I'll assume that it's correct and safe until anyone tells me otherwise.

Page 1 of 2 12 LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •