Hi,
it is a limitation - and always was - that others (like in this case openstack) 
treat host-capabilities CPU name as something it isn't - a name vs an 
approximation of the CPU.

I'm glad upstream now documented it after working with them on getting
this finally more outspoken.

Libvirt would never want you to say "CPU name" and more likely "Name +/- 
features".
Which is just what libvirt does when it is asked to express the host model.

I thought Openstack does support host-model [1], which can be the
easiest but only good if you have consistent CPUs across all your nodes.
But it also has custom [2], which allows the use of CPU Features [3]
mostly in the same way that libvirt does.

Due to that - no matter if security issues, CPU errata and more change
the features after a name was defined. You can still express "any-name +
features" there as well and thereby get all the new CPU features right?
(host-model would do the same for you, but if you want a common
denominator you can define it via custom+flags here).

Adding a few openstack people to the bug to get their POV to all of
this.

[1]: https://docs.openstack.org/nova/latest/admin/cpu-models.html#host-model
[2]: https://docs.openstack.org/nova/latest/admin/cpu-models.html#custom
[3]: 
https://docs.openstack.org/nova/latest/admin/cpu-models.html#cpu-feature-flags

** Also affects: nova (Ubuntu)
   Importance: Undecided
       Status: New

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2084102

Title:
  Sapphire rapids not recognized by virsh capabilities

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/2084102/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to