Update patch set 1

Patch Set 1: I would prefer that you didn't merge this

(1 inline comment)

Not really a -1, but not sure what to put here.
 
I believe the stuff I mention in the comment needs at least consideration as part of this blueprint. I strongly believe it is a design issue that is the cause of a number of bugs some of which are related to the live mig just because it is so related to handling disks.

Patch-set: 1
Reviewer: Gerrit User 5511 <5511@4a232e18-c5a9-48ee-94c0-e04e7cca6543>
Label: Code-Review=-1

be6ac8eff346d062fefd53cdea49ab52f9661db6

@@ -0,0 +1,21 @@
+{
+  "comments": [
+    {
+      "key": {
+        "uuid": "AAAAWH//glY\u003d",
+        "filename": "juno/live-migration",
+        "patchSetId": 1
+      },
+      "lineNbr": 20,
+      "author": {
+        "id": 5511
+      },
+      "writtenOn": "2014-03-17T13:28:47Z",
+      "side": 1,
+      "message": "The root cause of many libvirt driver issues and I believe this one as well is get_instance_disk_info method.\n\nBasically instead of consulting the nova database, any code paths that use this method, attempt to guess the layout of instance disks from the libvirt xml.\n\nThere are several issues with this but I will name the two that I think are the biggest ones and a cause of most grief:\n\n* This makes any code inside the libvirt driver more difficult to reason about as there are now 2 completely different sources of data that are being used.\n\n* the libvirt xml does not contain all the information needed for nova to be able to manage VMs.\n\nIf you ask me - fixing live migration in libvirt (and hence in the gate) should start from nuking this completely from the libvirt driver and using only Nova as the source of information.",
+      "revId": "be6ac8eff346d062fefd53cdea49ab52f9661db6",
+      "serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543",
+      "unresolved": false
+    }
+  ]
+}