Update patch set 6

Patch Set 6:

(4 comments)

Patch-set: 6
Label: Verified=0

68fd5aa8be6c12d2b59d22aa6594bd32da2d7fd1

@@ -0,0 +1,96 @@
+{
+  "comments": [
+    {
+      "key": {
+        "uuid": "5ff73747_cf822001",
+        "filename": "specs/newton/address-scope-subnet-pool-mapping.rst",
+        "patchSetId": 6
+      },
+      "lineNbr": 55,
+      "author": {
+        "id": 1689
+      },
+      "writtenOn": "2017-05-03T15:49:58Z",
+      "side": 1,
+      "message": "Do you mean \"with the aim_mapping driver\"? Maybe you should say \"with the ml2plus core plugin\" instead. It should be possible to use ml2plus in place of the ml2 with other (non-AIM) mechanism and policy drivers.\n\nI\u0027m becoming more concerned about this implicit GBP L3P workflow depending on this implicit subnetpool extension (which was my idea, I admit). One reason is that only ml2plus supports this extension, so there is at least a question of whether the implicit workflow is really a core GBP feature, or is a vendor-specific feature. Portability is seriously compromised if basic GBP functionality like this doesn\u0027t work the same with any backend. \n\nThe other reason is that the existence of an implicit subnetpool changes Neutron semantics in a way that could break lots of existing Neutron client code. We shouldn\u0027t force users to choose between compliance with standard Neutron behavior and usage of a GBP feature.\n\nOne solution to consider would be having GBP use the default subnetpool extension instead of (or in addition to) the impliciit subnetpool extension. The default subnetpool extension would still provide implicit v4 and/or v6 subnetpools for L3P creation, but does not break the Neutron API like the implicit subnetpool extension does. Neutron users need to pass the --use_default_subnetpool if they want the default subnetpool, and GBP could do this as well.\n\nIt seems that if a deployment using the apic_aim MD and aim_mapping PD is using implicit subnetpools for Neutron workflows, they would want these same subnetpools to be used for GBP workflows. I think this could be accomplished by setting the same subnetpools as being both implicit and default. If so, I think GBP should define its implicit L3 policy creation as using just the default subnetpool extension, and thus make this feature available with and backend, and without breaking standard Neutron semantics.",
+      "range": {
+        "startLine": 55,
+        "startChar": 49,
+        "endLine": 55,
+        "endChar": 69
+      },
+      "revId": "68fd5aa8be6c12d2b59d22aa6594bd32da2d7fd1",
+      "serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543",
+      "unresolved": false
+    },
+    {
+      "key": {
+        "uuid": "5ff73747_4ffcd03d",
+        "filename": "specs/newton/address-scope-subnet-pool-mapping.rst",
+        "patchSetId": 6
+      },
+      "lineNbr": 68,
+      "author": {
+        "id": 1689
+      },
+      "writtenOn": "2017-05-03T15:49:58Z",
+      "side": 1,
+      "message": "Again, not sure what this means. Anything other than the aim_mapping PD? What about the resource_mapping PD or other PDs that derive from it? I think we need a general solution, and using the default subnetpool extension may be it.",
+      "range": {
+        "startLine": 68,
+        "startChar": 39,
+        "endLine": 68,
+        "endChar": 50
+      },
+      "revId": "68fd5aa8be6c12d2b59d22aa6594bd32da2d7fd1",
+      "serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543",
+      "unresolved": false
+    },
+    {
+      "key": {
+        "uuid": "5ff73747_af491361",
+        "filename": "specs/newton/address-scope-subnet-pool-mapping.rst",
+        "patchSetId": 6
+      },
+      "lineNbr": 332,
+      "author": {
+        "id": 1689
+      },
+      "writtenOn": "2017-05-03T15:49:58Z",
+      "side": 1,
+      "message": "If a single external_segment can be dual-stack (ip_version \u003d\u003d 46), how are CIDRs for each stack represented. Does this need to become a list?",
+      "range": {
+        "startLine": 330,
+        "startChar": 0,
+        "endLine": 332,
+        "endChar": 65
+      },
+      "revId": "68fd5aa8be6c12d2b59d22aa6594bd32da2d7fd1",
+      "serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543",
+      "unresolved": false
+    },
+    {
+      "key": {
+        "uuid": "5ff73747_8f189775",
+        "filename": "specs/newton/address-scope-subnet-pool-mapping.rst",
+        "patchSetId": 6
+      },
+      "lineNbr": 342,
+      "author": {
+        "id": 1689
+      },
+      "writtenOn": "2017-05-03T15:49:58Z",
+      "side": 1,
+      "message": "I don\u0027t see the change?",
+      "range": {
+        "startLine": 338,
+        "startChar": 0,
+        "endLine": 342,
+        "endChar": 6
+      },
+      "revId": "68fd5aa8be6c12d2b59d22aa6594bd32da2d7fd1",
+      "serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543",
+      "unresolved": false
+    }
+  ]
+}