Thanks! I am now starting to investigate if we can re-use parts of migration code after some refactoring, and how much code should be written from scratch. I am not very clear about how the new flow looks like. I think the migration protocol v3 is quite reasonable for dealing with various cases and errors in the migration. After all, QEMU live upgrade is implemented as migration under the hood, so I think common migration management logic would be applied to it too. For example, if QEMU live upgrade migration fails, target domain should be destroyed and source domain should be resumed. Maybe I can re-use some of the parts in the migration protocol v3 flow.

By the way. It seems my mail sent to libvir-list(a)redhat.com is not appearing in the list. I'm already a member in the list. I see in the subscribing web page saying first time posts need moderator approval. I thought I've posted before. Maybe it's too long since my last activity. I also tried to join the list again and the system sent me a mail saying I was already a member. I sent a mail to libvir-list-owner(a)redhat.com asking for help but no response yet. Am I blocked? Or is it just because I didn't add a [libvirt] tag in the title? I thought the tag is added automatically when I was posting in other lists. Would you have a look at my subscription status? Thank you!