Those patches look good to my eyes. Thank you Lee.
The date sounds sensible, I am unsure on the usual timeframe, but that sounds like some warning combined with getting this information to our users as soon as we can.
I think that description looks OK. I do wonder if we want to say the VM reverts to using the libvirt XML it used on the source host after a soft reboot. I guess the patches make that very clear.
In terms of mitigations, could you ask users to hard reboot instances that have been live-migrated via the API/horizon. I think that would also reset the persistent libvirt XML? Is that correct, or is it worse than that? I think operators could look at the actions list for each instance to determine if it has been affected by a live-migration followed by a soft reboot, and target those instances for a hard reboot?
Maybe that is too much detail, especially for something we would need to test to be sure it helps?
Those patches look good to my eyes. Thank you Lee.
The date sounds sensible, I am unsure on the usual timeframe, but that sounds like some warning combined with getting this information to our users as soon as we can.
I think that description looks OK. I do wonder if we want to say the VM reverts to using the libvirt XML it used on the source host after a soft reboot. I guess the patches make that very clear.
In terms of mitigations, could you ask users to hard reboot instances that have been live-migrated via the API/horizon. I think that would also reset the persistent libvirt XML? Is that correct, or is it worse than that? I think operators could look at the actions list for each instance to determine if it has been affected by a live-migration followed by a soft reboot, and target those instances for a hard reboot?
Maybe that is too much detail, especially for something we would need to test to be sure it helps?