Roadmap: Xen revisited

[Stefan Bader]

On 20.07.2011 11:30, Mark Goldfinch wrote:
> On 20/07/11 21:12, Stefan Bader wrote:
>> The several issues rather being related to run the dom0. I still see some
>> warnings getting triggered which seem to be harmless enough. And HVM is not
>> working. There is another minor issue when you have sd* disk names in your pv
>> configuration (which is actually discouraged).
> 
> Great, thanks for the clarification..
> 
>> Otherwise I have PVM domUs working with our 3.0-rc7 based dom0. Not sure what
>> the Debian 6 dom0 is based on. The blkback was the one thing that went upstream
>> quite late. But since we had been following the 3.0 kernel, we would have picked
>> up any fixes there automatically.
> 
> Debian 6 dom0 uses 2.6.32 Linux kernel and 4.0.1 Xen Hypervisor kernel.  I'm
> using amd64 architecture.
> 
> What can I do to help diagnose what's going on here?
> 
> Thanks,
> Mark.
> 
In the end (as its Debians dom0) there should be a debian bug opened. Of
interest likely is:

- clear statement about last running domU (2.6.? -> 3.0 regression?)
- the used configuration file
- is root= using a device name or label or uuid (boot loader)?
- dmesg from any last working case and the failing one
- check xm log / xm dmesg for hints
- try to rule out the boot loader by extracting initrd and vmlinuz and
  using kernel and ramdisk pointing to those in the cfg file instead.

So I'd probably play around with those hints, especially the boot loader test. I
am using pvgrub on both the Oneiric dom0 and a CentOS based dom0 and both were
working for me.

-Stefan