Update 2015-04-26: a better solution is now provided in this bug report

In 2013, Debian wheezy was released with a number of great virtualization options include the Xen Cloud Platform (XCP / Xen-API) toolstack packaged by Thomas Goirand to run in a native Debian host environment.

Unfortunately, XCP is not available as a host (dom0) solution for the upcoming Debian 8 (jessie) release. However, it is possible to continue running a Debian wheezy system as the dom0 host and run virtualized (domU) jessie systems inside it. It may also be possible to use the packages from wheezy on a jessie system, but I haven't looked into that myself so far.

Newer kernel boot failures in Xen

After successfully upgrading a VM (domU in Xen terminology) from wheezy to jessie, I tried to reboot the VM and found that it wouldn't start. People have reported similar problems booting newer versions of Ubuntu and Fedora in XCP and XenServer environments. PyGrub displayed an error on the dom0 console:

# xe vm-start name-label=server05
Error code: Using  to parse /grub/grub.cfg
Error parameters: Traceback (most recent call last):,
   File "/usr/lib/xcp/lib/pygrub.xcp", line 853, in ,
     raise RuntimeError, "Unable to find partition containing kernel"

There is a quick and easy workaround. Hard-code the kernel and initrd filenames into config values that will be used to boot. A more thorough solution will probably involve using a newer version of PyGrub in wheezy.

If the /boot tree is a separate filesystem inside the VM, use commands like the following (substitute the correct UUID for the VM and the exact names/versions of the vmlinuz and initrd.img files):

xe vm-param-set uuid=da654fd0-74db-11e4-82f8-0800200c9a66 \

xe vm-param-set uuid=da654fd0-74db-11e4-82f8-0800200c9a66 \
   PV-args="root=/dev/mapper/vg00-root ro quiet"

and if /boot is on the root filesystem of the VM, this will do the trick:

xe vm-param-set uuid=da654fd0-74db-11e4-82f8-0800200c9a66 \

xe vm-param-set uuid=da654fd0-74db-11e4-82f8-0800200c9a66 \
   PV-args="root=/dev/mapper/vg00-root ro quiet"

Future strategy

Once a comprehensive XCP solution appears in Debian again, hopefully it will be possible to migrate running VMs into the new platform without any downtime and retire the wheezy dom0.

Other upgrade/migration options exist and the choice will depend on various factors, such as whether or not you have built your own tools around the XCP API and whether you use a solution like OpenStack that depends on it. Debian's pkg-xen-devel mailing list may be a good place to discuss these options further.