Make 'no_timer_check' Kernel parameter for Fedora cloud images

[Kashyap Chamarthy]

Heya,

This is a brief summary of discussions re `no_timer_check` Kernel
parameter from different bug trackers (see references below for more
details) 


What?
-----

Make 'no_timer_check' Kernel parameter default for Fedora cloud images.


Why?
----

It has been reported (recently and in the past) many times[1][2][3] that
Kernel timer checks are not reliable/helpful in the context of
virtualization and it's useful to disable

This is also useful to elimiate false-positives in upstream
OpenStack CI testing[1][2].


Notes
-----

  - 'no_timer_check' parameter is added implicitly when the kernel
    detects it is booting under KVM.

    However when booting under QEMU TCG (software emulation) you have to
    add this to the guest kernel command line explicitly.

    (Thanks Richard WM Jones.)


  - About 'no_timer_check', from Kernel documentation[4]:

        'no_timer_check'  [X86,APIC] Disables the code which tests for
                          broken timer IRQ sources. 


References
----------

 [1] https://bugs.launchpad.net/cirros/+bug/1312199 -- cirros 0.3.1
     fails to boot 
       - Relevant comments: #1, #2, and #4

 [2] https://bugzilla.redhat.com/show_bug.cgi?id=1102592 -- F20 could
     image affected by timing issues when using with soft-qemu

 [3] https://bugzilla.redhat.com/show_bug.cgi?id=502058#c15 -- qemu
     -no-kvm guest hangs at during timer setup; works with noapic

 [4] https://www.kernel.org/doc/Documentation/kernel-parameters.txt


-- 
/kashyap

[Daniel P. Berrange]

On Wed, Jun 18, 2014 at 10:42:51PM +0530, Kashyap Chamarthy wrote:
>  Heya,
>  
>  This is a brief summary of discussions re `no_timer_check` Kernel
>  parameter from different bug trackers (see references below for more
>  details) 
>  
>  
>  What?
>  -----
>  
>  Make 'no_timer_check' Kernel parameter default for Fedora cloud images.
>  
>  
>  Why?
>  ----
>  
>  It has been reported (recently and in the past) many times[1][2][3] that
>  Kernel timer checks are not reliable/helpful in the context of
>  virtualization and it's useful to disable
>  
>  This is also useful to elimiate false-positives in upstream
>  OpenStack CI testing[1][2].
>  
>  
>  Notes
>  -----
>  
>    - 'no_timer_check' parameter is added implicitly when the kernel
>      detects it is booting under KVM.
>  
>      However when booting under QEMU TCG (software emulation) you have to
>      add this to the guest kernel command line explicitly. 
FYI, upstream OpenStack now explicitly adds this kernel parameter when
possible, but it can only do this when booting from a kernel/initrd.
If the guest is booting from a virtual BIOS, there's no option but
to add it to the guest image grub.conf manually.

commit 6b86a61fee15ce1237303fab2f7896f8c3bcad47
Author: Attila Fazekas <afazekas(a)redhat.com&gt;
Date:   Wed May 28 09:19:29 2014 +0200

    Use no_timer_check with soft-qemu
    
    The Linux kernel timer check not working properly
    when the hypervisor's thread preempted by the host CPU scheduler.
    
    The timer check is automatically disabled with other types
    of hypervisors including the hardware accelerated kvm,
    but timer_check is not disabled when qemu used without hardware acceleration.
    
    This issue is frequently mischaracterized as an SSH connectivity issue and
    causes rechecks and occasional boot failures.
    
    This change adds no_timer_check kernel parameter when we are using
    uec images with qemu.
    
    Closes-Bug: #1312199
    Change-Id: I3cfdfe9048fe219fc12cdac8a399b496f237e55e


Regards,
Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|

[Kashyap Chamarthy]

On Wed, Jun 18, 2014 at 06:16:04PM +0100, Daniel P. Berrange wrote:
>  On Wed, Jun 18, 2014 at 10:42:51PM +0530, Kashyap Chamarthy wrote:
[. . .]

>  > Notes
>  > -----
>  > 
>  >   - 'no_timer_check' parameter is added implicitly when the kernel
>  >     detects it is booting under KVM.
>  > 
>  >     However when booting under QEMU TCG (software emulation) you have to
>  >     add this to the guest kernel command line explicitly.
>  
>  FYI, upstream OpenStack now explicitly adds this kernel parameter when
>  possible, but it can only do this when booting from a kernel/initrd.
>  If the guest is booting from a virtual BIOS, there's no option but
>  to add it to the guest image grub.conf manually. 
Yup, noticed this fix from LP bug -- 1312199. I see exactly your
point (it works only when you use '--append' w/ external
Kernel/initrd) echoed in comment #14[1], by rjones.

  [1] https://bugs.launchpad.net/cirros/+bug/1312199

>  
>  commit 6b86a61fee15ce1237303fab2f7896f8c3bcad47
>  Author: Attila Fazekas <afazekas(a)redhat.com&gt;
>  Date:   Wed May 28 09:19:29 2014 +0200
>  
>      Use no_timer_check with soft-qemu
>      
>      The Linux kernel timer check not working properly
>      when the hypervisor's thread preempted by the host CPU scheduler.
>      
>      The timer check is automatically disabled with other types
>      of hypervisors including the hardware accelerated kvm,
>      but timer_check is not disabled when qemu used without hardware acceleration.
>      
>      This issue is frequently mischaracterized as an SSH connectivity issue and
>      causes rechecks and occasional boot failures.
>      
>      This change adds no_timer_check kernel parameter when we are using
>      uec images with qemu.
>      
>      Closes-Bug: #1312199
>      Change-Id: I3cfdfe9048fe219fc12cdac8a399b496f237e55e 
 

-- 
/kashyap