[CentOS-virt] Domain0 with Kernel 2.6.18-92.1.22.el5xen fails to boot Guests with Kernel 2.6.18-164.el5xen

Pasi Kärkkäinen pasik at iki.fi
Sun Sep 27 11:21:51 UTC 2009


On Sun, Sep 27, 2009 at 12:46:46PM +1000, Devraj Mukherjee wrote:
> Hi all,
> 
> I am in a bit of a fix here, I have several Guests paravirtualized
> running on a Dom0 that is currently running the 2.6.18-92.1.22.el5xen
> Kernel. All my guests (administered by different people) have
> subsequently been updated to later versions of the kernel (namely
> 2.6.18-164.el5xen or 2.6.18-128.4.1.el5xen).
> 
> Ones that are still running are using the Kernel 2.6.18-128.4.1.el5xen
> while Dom0 is still booted up into 2.6.18-92.1.22.el5xen
> 
> I restarted a Guest that's now trying to boot with the
> 2.6.18-164.el5xen Kernel and it spits out messages like
> XENBUS: Waiting for devices to initialise: 295s...290s...
> 
> and eventually reboots.
> 
> If I try and interrupt the boot and choose a previous version of the
> Kernel on the Guest, it says the following and fails to boot. The
> guests use a single disk image that has the SWAP and Root partitions.
> 
> Is this happening because the version of the kernel on the Dom0 is way
> too old for guests to boot? I am just a bit hesitant in rebooting my
> server for the new kernel to take effect unless I am sure :)
>

No, that shouldn't be the reason. 

The layer between dom0 and guests is Xen hypervisor, not the kernel.
Although Redhat ships the hypervisor in the same rpm as the kernel :)

The output looks like the initrd image is broken; it doesn't set up the root
device.. does it load the xenblk driver? Does it detect the xvd* disks?

Please paste the full domU kernel boot log..

-- Pasi




More information about the CentOS-virt mailing list