Applying hotfixes XS71ECU1024, XS71ECU1026 and XS71ECU1027 on a XS 7.1 pool leads to many VMs unable to be migrated because of internal changes to the CPU masks. The only recourse then is to shut down and restart the VMs manually on an updated host or, fortunately, you can still do: xe vm-migrate uuid=(UUID-of-VM) host-uuid=(UUID-of-target-host) --live force=true . Some updated hosts could migrate VMs to already upgraded hosts and not yet updated hosts, but host that have not been upgraded could not move their remaining VMs to an updated host.
Note that I am not sure which one or combination of these hotfixes triggers this.
With all hosts finally patched, it's actually worse. Even now with identical hosts all patched to the same level, most cannot migrate VMs via XenCenter to any others because of claiming a CPU mismatch. Not a good situation. The master apparently can migrate VMs from itself to other hosts, but the remaining hosts now cannot seem to migrate VMs anywhere else. All four hosts in the pool are identical in HW, have the same BIOS versions and hotfixes. This was not an issue before these three hotfixes were applied. This is a licensed version of XS 7.1 ECU1.