On 27.11.2012 22:42, Eric Blake wrote:
This patch set re-implements migration with storage for enough new qemu.
How does this series interact with Li Guang's efforts to add offline migration? In particular,
I guess you mean [1] more precisely, right? Although, event the origin offline migration patch [2] does something similar like I am doing in 2/11: propagating 'unsigned long flags' deeper in the stack of functions to make some decisions. But that's not what are you asking later, so ignore this comment :)
1) src -> dest: (QEMU_MIGRATION_PHASE_BEGIN3 -> QEMU_MIGRATION_PHASE_PREPARE) <nbd> <disk src='/var/lib/libvirt/images/f17.img' size='17179869184'/> </nbd>
Both sets of patches need to pass size information across in the cookies; so is tying it to <nbd> appropriate, or should we be rethinking this XML to be shared between both patches?
I've commented his patch [1] yesterday. We need the same piece of functionality. Actually, my migration cookie is just a superset of his (because I do need the port attribute as pointed out in comment to 6/11). However, I am not hesitating to rename it. But from a quick look at his patch - we seem to implement the same thing, more or less. His patch allows us to pre-create images for older qemu which doesn't support nbd-server-* yet. But who is using such ancient qemu? :) 1: https://www.redhat.com/archives/libvir-list/2012-November/msg01022.html 2: https://www.redhat.com/archives/libvir-list/2012-November/msg00886.html