Page 1 of 3 123 LastLast
Results 1 to 10 of 24

Thread: Persistent system crash

  1. #1
    Join Date
    Aug 2008
    Location
    England
    Posts
    63

    Default Persistent system crash

    I am finding 42.3 very unstable on a brand new system.

    uname -a gives

    Code:
    Linux hostname 4.4.85-22-default #1 SMP Fri Sep 1 14:21:21 UTC 2017 (0c39a1f) x86_64 x86_64 x86_64 GNU/Linux
    The motherboard is an Asus Prime B250-Pro with Intel Core i3-7100 with 16GB DDR4 RAM. Graphics is the integrated Intel device. HD is a Samsung 500GB SSD.

    The install is a clean install from a USB stick

    The system works for a while but after an hour or more the typical failure is the crash of a running application and then the complete seizure of the keyboard and mouse. No <alt><ctrl>F1.
    I can sometimes SSH into the system but often even that is slow and crashes.

    An example is a crash of firtefox which was run from a konsole command line. This looks like this:

    Code:
    AiOGest: init
    AiOGest: end init
    ExceptionHandler::WaitForContinueSignal waiting for continue signal...
    ExceptionHandler::GenerateDump cloned child 6921
    ExceptionHandler::SendContinueSignalToChild sent continue signal to child
    When I move to iceWM in an attempt to evade plasma problems that also seizes after a while.

    Can I supply more info and if so what?

    Thanks
    Peter

  2. #2

    Default Re: Persistent system crash

    Quote Originally Posted by pblewis View Post
    Graphics is the integrated Intel device.
    Try to uninstall the kernel graphics stack update, drm-kmp-default, and see if it helps.

    Or uninstall the intel Xorg driver, xf86-video-intel, and use the generic modesetting driver (which is even said to have better performance on some systems...).

  3. #3
    Join Date
    Aug 2008
    Location
    England
    Posts
    63

    Default Re: Persistent system crash

    Hi Wolfi,

    I uninstalled drm-kmp-default and the display was dropped into a fixed 1024 x 768 resolution with no option to change the resolution back to my normal 1920 x 1080.
    System settings did not have the resolution change feature normally found in the Dispalys tab.
    Is there any way round this please?

    peter

  4. #4

    Default Re: Persistent system crash

    Quote Originally Posted by pblewis View Post
    I uninstalled drm-kmp-default and the display was dropped into a fixed 1024 x 768 resolution with no option to change the resolution back to my normal 1920 x 1080.
    Then apparently your GPU is quite new and not supported by kernel 4.4 yet.
    So the system probably uses the generic fbdev driver now, which doesn't allow resolution changes.
    /var/log/Xorg.0.log should tell more though.

    Is there any way round this please?
    Yes, install it again.

    And try to uninstall xf86-video-intel instead.

  5. #5
    Join Date
    Aug 2008
    Location
    England
    Posts
    63

    Default Re: Persistent system crash

    Unfortunately xf86-video-intel is not installed on the system.
    Would it be worth trying with it installed?

    Peter

  6. #6
    Join Date
    Aug 2008
    Location
    England
    Posts
    63

    Default Re: Persistent system crash

    I have now installed xf86-video-intel without any noticeable problems but it I have since had a system crash. I was able to ssh into the system but an attempt to halt the system from the remote host failed. Having su'd into root:

    Code:
    bardsey:/home/peter # halt
    
    Broadcast message from systemd-journald@bardsey (Fri 2017-09-15 14:34:46 BST):
    
    systemd[1]: Caught <SEGV>, dumped core as pid 19702.
    
    
    Broadcast message from systemd-journald@bardsey (Fri 2017-09-15 14:34:46 BST):
    
    systemd[1]: Freezing execution.
    
    Warning! D-Bus connection terminated.
    Failed to wait for response: Connection reset by peer
    Failed to open /dev/initctl: No such device or address
    Failed to talk to init daemon.
    bardsey:/home/peter # 
    bardsey:/home/peter # halt
    Failed to start halt.target: Connection timed out
    Failed to open /dev/initctl: No such device or address
    Failed to talk to init daemon.
    bardsey:/home/peter # halt
    Failed to start halt.target: Activation of org.freedesktop.systemd1 timed out
    Failed to open /dev/initctl: No such device or address
    Failed to talk to init daemon.
    bardsey:/home/peter # exit
    exit
    So this is looks to me as a complete failure of the system daemon.

    Peter

  7. #7

    Default Re: Persistent system crash

    Quote Originally Posted by pblewis View Post
    Unfortunately xf86-video-intel is not installed on the system.
    Would it be worth trying with it installed?
    Well, if it is not installed, it cannot be the cause of the problems either. (I mainly suggested to uninstall it because it is quite buggy on certain chipsets it seems)

    Installing it probably won't change anything at all, because if it is not installed by default it probably doesn't support your graphics chip anyway (which means it won't get used even if installed).

    Again, /var/log/Xorg.0.log should tell what happens.

    Quote Originally Posted by pblewis View Post
    I was able to ssh into the system but an attempt to halt the system from the remote host failed. Having su'd into root:

    Code:
    bardsey:/home/peter # halt
    
    Broadcast message from systemd-journald@bardsey (Fri 2017-09-15 14:34:46 BST):
    
    systemd[1]: Caught <SEGV>, dumped core as pid 19702.
    Looks like systemd crashed. (and that would also explain why the subsequent "halt" doesn't work at all, as there's no init system any more to talk to and initiate the halt)
    Btw, it would probably be better to run "systemctl poweroff", "shutdown", or "halt -p", as "halt" alone will *not* poweroff the system.

    If systemd crashes, that can cause all sort of problems of course, but the question is whether this is related to your other problems, or rather only happens when you run "halt" via ssh.

    coredumpctl should list all crashes, so having a look at that via ssh after the system froze may give more clues.
    And the output of dmesg might be interesting too.

    As this seems to be brand new hardware (as e.g. the GPU apparently is not even supported by the standard kernel), the best thing probably would be trying to install the latest kernel version though, available from the Kernel:stable repo.
    http://download.opensuse.org/reposit...able/standard/

    E.g. add this URL as repo with YaST->Repositories and then install the highest version of kernel-default in YaST->Software Management (click on "Versions" below the package list to see all available versions).
    The standard kernel will be kept (and available to boot in "Advanced Options" in the boot menu) if you do this, so you can easily switch back in case of problems.
    Last edited by wolfi323; 16-Sep-2017 at 05:15.

  8. #8
    Join Date
    Aug 2008
    Location
    Little Exuma, Bahamas
    Posts
    699

    Default Re: Persistent system crash

    FWIW, I have 'enjoyed' many of the issues described by the OP on four very new systems (Kabylake, released by Intel about the same time as the B250 series) and found simply updating the Kernel to resolve most of those issues, however it's important to know what version to update to.

    Based on my experience my highly subjective opinion is as follows:

    4.4 -not able to handle Kabylake, particularly graphics, very unstable
    4.9 - better but not much better
    4.11 - should be ok based on posts elsewhere, but that's not my experience.
    4.12 - very stable
    4.13 - only using it on two systems for two days, but seems OK
    MS user 1988-2008, Linux user 1998-present, openSUSE user since 2004
    (The first computer I used had a punch card reader)

  9. #9
    Join Date
    Aug 2008
    Location
    England
    Posts
    63

    Default Re: Persistent system crash

    Thank you for that advice.

    I have upgraded as Wolfi suggests and now have version 4.13.2-1.1.g68f4aee installed.

    I am guessing that I will have to re-install virtualbox to get that running.

    Thanks for the stability advice caprus. I will report back on the results of this upgrade.

    The oldest computer that I have run code on is a Mercury computer which ran Pegasus autocode. One hot room full of valves and drum memory!

  10. #10

    Default Re: Persistent system crash

    Quote Originally Posted by pblewis View Post
    I am guessing that I will have to re-install virtualbox to get that running.
    Basically you need to compile/install the kernel modules for the new kernel.

    If you use Oracle's RPM, you can do that by running "sudo vboxconfig setup".

    If you use openSUSE's RPM, this won't work out of the box, as it only comes with a pre-compiled module that fits the distribution kernel, but no source code.
    Additionally installing virtualbox-host-source should make that work (but I never tried that myself).
    OTOH, there is a Virtualization repo that comes with a kmp (kernel module package) built against the latest kernel from Kernel:stable:
    http://download.opensuse.org/reposit...able_standard/
    (you should also install virtualbox from there then, to avoid version mismatches)

Page 1 of 3 123 LastLast

Tags for this Thread

Posting Permissions

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