New and Current Xen Systems no longer fully boot and pause themselves

[Ricky Burgin]

Suddenly out of nowhere, Xen instances, new and current will no longer boot fully. Instead, they will come up for a brief moment, then go into a paused state, unpausing them only results in them being paused shortly after. I am not using HVM, as others seem to have initially thought. Systems that have not had to be rebooted are still running fine.

Ideas?


_______________________________________________
Xen-users mailing list
Xen-users@...
http://lists.xensource.com/xen-users

[Pasi Kärkkäinen]

On Tue, Oct 27, 2009 at 04:00:43PM +0000, Ricky Burgin wrote:
>    Suddenly out of nowhere, Xen instances, new and current will no longer
>    boot fully. Instead, they will come up for a brief moment, then go into a
>    paused state, unpausing them only results in them being paused shortly
>    after. I am not using HVM, as others seem to have initially thought.
>    Systems that have not had to be rebooted are still running fine.
>
>    Ideas?

Do you have xenconsoled running?

Sometimes it dies on its own, and I've also seen it getting stuck; in
that you need to kill and restart it.

broken xenconsoled could cause problems you described..

-- Pasi


_______________________________________________
Xen-users mailing list
Xen-users@...
http://lists.xensource.com/xen-users

[Ricky Burgin]

Hi Pasi,

Thanks for the response. This seems to have been caused by an update from CentOS 5.3 to 5.4. I'm not exactly sure what packages were specifically affected during the update, but after reluctantly rebooting the server, it came back up and Xen started functioning completely normally. Could quite easily have been xenconsoled, I suppose.

Many thanks,
Ricky Burgin

On 28/10/2009 12:06, Pasi Kärkkäinen wrote:

On Tue, Oct 27, 2009 at 04:00:43PM +0000, Ricky Burgin wrote:
  

   Suddenly out of nowhere, Xen instances, new and current will no longer
   boot fully. Instead, they will come up for a brief moment, then go into a
   paused state, unpausing them only results in them being paused shortly
   after. I am not using HVM, as others seem to have initially thought.
   Systems that have not had to be rebooted are still running fine.

   Ideas?
    

Do you have xenconsoled running? 

Sometimes it dies on its own, and I've also seen it getting stuck; in
that you need to kill and restart it.

broken xenconsoled could cause problems you described..

-- Pasi

  

_______________________________________________
Xen-users mailing list
Xen-users@...
http://lists.xensource.com/xen-users

[Fajar A. Nugraha-3]

On Wed, Oct 28, 2009 at 7:09 PM, Ricky Burgin <ricky@...> wrote:
> Hi Pasi,
>
> Thanks for the response. This seems to have been caused by an update from
> CentOS 5.3 to 5.4. I'm not exactly sure what packages were specifically
> affected during the update, but after reluctantly rebooting the server, it
> came back up and Xen started functioning completely normally. Could quite
> easily have been xenconsoled, I suppose.

Xen userland must match the version of hypervisor you're running.
Since upgrading to Centos 5.4 should've include kernel-xen (which also
contain the hypervisor) and xen (the userland), it makes sense that
you have to reboot to have it function properly.

I usually exclude these packages from normal update, and only update
them just before I reboot the server:
kernel kernel-devel kernel-xen kernel-xen-devel xen xen-libs glibc
glibc-common glibc-devel glibc-headers glibc-utils

--
Fajar

_______________________________________________
Xen-users mailing list
Xen-users@...
http://lists.xensource.com/xen-users

[Ricky Burgin]

Actually I believe the kernel wasn't updated as it was one from the openvz repositories which allowed me to run Xen instances and OpenVZ containers simultaneously on the same system. This did not update during the 5.3 to 5.4 update.

On 29/10/2009 02:36, Fajar A. Nugraha wrote:

On Wed, Oct 28, 2009 at 7:09 PM, Ricky Burgin ricky@... wrote:
  

Hi Pasi,

Thanks for the response. This seems to have been caused by an update from
CentOS 5.3 to 5.4. I'm not exactly sure what packages were specifically
affected during the update, but after reluctantly rebooting the server, it
came back up and Xen started functioning completely normally. Could quite
easily have been xenconsoled, I suppose.
    

Xen userland must match the version of hypervisor you're running.
Since upgrading to Centos 5.4 should've include kernel-xen (which also
contain the hypervisor) and xen (the userland), it makes sense that
you have to reboot to have it function properly.

I usually exclude these packages from normal update, and only update
them just before I reboot the server:
kernel kernel-devel kernel-xen kernel-xen-devel xen xen-libs glibc
glibc-common glibc-devel glibc-headers glibc-utils

  

_______________________________________________
Xen-users mailing list
Xen-users@...
http://lists.xensource.com/xen-users

[Fajar A. Nugraha-3]

On Thu, Oct 29, 2009 at 9:40 AM, Ricky Burgin <ricky@...> wrote:
> Actually I believe the kernel wasn't updated as it was one from the openvz
> repositories which allowed me to run Xen instances and OpenVZ containers
> simultaneously on the same system. This did not update during the 5.3 to 5.4
> update.

Was the userland updated? Perhaps somehow xenstored was deleted or
restarted. While both xend and xenconsoled can be restarted safely,
other components (like xenstored) can not.

--
Fajar

_______________________________________________
Xen-users mailing list
Xen-users@...
http://lists.xensource.com/xen-users