Comment 21 for bug 1539271

FWIW, adding some related notes from observing the logs above:

From the Gate logs: Block migration didn't even begin -- on the source node, 'drive-mirror' didn't even start; and on the target node, NBD server was not started either.

However, from Paul's logs: We can at least see that 'drive-mirror' began on the source node. And on the target node, NBD server was started, and the disk ('drive-virtio-disk0') that's being was migrated was added, _however_ the NBD server was not stopped yet (by issuing the command 'nbd-server-stop').

Thus, confirming Dan's theory above that the Gate failures and Paul's failures are different.

- - -

In a successful live block migration (with 'drive-mirror'+ NBD approach): From my previous testing (a few months ago, as you can see from time-stamps; tested with: libvirt-daemon-driver-qemu-1.2.13.1-2 qemu-system-x86-2.3.1-7), on the source node, you should see messages that indicate drive mirroring was complete:

...
2015-09-29 18:47:45.021+0000: 8285: debug : qemuMigrationDriveMirror:1800 : Drive mirroring of 'drive-virtio-disk0' completed
...

IIUC, this message on the source means the out going migration has been completed successfully:

...
2015-09-26 12:52:07.908+0000: 758: debug : qemuDomainObjSetJobPhase:1231 : Setting 'migration out' phase to 'perform3_done'
...

- - -