Update patch set 13

Patch Set 13:

(1 comment)

Patch-set: 13

0a92b33b61ec519d1afbbed6217cf225a02927a6

@@ -34,6 +34,24 @@
       "revId": "0a92b33b61ec519d1afbbed6217cf225a02927a6",
       "serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543",
       "unresolved": true
+    },
+    {
+      "key": {
+        "uuid": "c7717e63_02423089",
+        "filename": "magnum/drivers/common/templates/kubernetes/fragments/upgrade-kubernetes.sh",
+        "patchSetId": 12
+      },
+      "lineNbr": 6,
+      "author": {
+        "id": 16036
+      },
+      "writtenOn": "2021-04-27T12:20:00Z",
+      "side": 1,
+      "message": "Thanks, I kind of expected that to be the case based on my previous findings but hope dies last - perhaps you could make the wording more explicit? Something like \"Node rebuilt as part of the cluster update\"?\n\nI have to say, trying to figure out what exactly can be done in Heat has been an exercise in frustration. There was a blueprint draft started here that talked about adding support for what we seemingly need: https://blueprints.launchpad.net/heat/+spec/update-hooks but the comment only mentions another blueprint that\u0027s supposed to \"cover 99% or more of [usecases]\". The aforementioned blueprint is https://blueprints.launchpad.net/heat/+spec/action-aware-sw-config but it\u0027s not clear to me what is actually implemented and how it works - I have a feeling that it doesn\u0027t, as SoftwareDeployment \"depends\" on the Server so I assume it will only start updating after server is rebuilt. Also, I don\u0027t think it\u0027s safe to upgrade master nodes (and etcd cluster) without some sort of global/shared lock anyway so that may be dead end anyway.\n\nPerhaps you are right, and pre-update hooks are what we really should be using - I\u0027ve looked into them but came to the same conclusion that some extra component running in the cluster are needed. But perhaps that\u0027s the case and there is no reason to look for something else? If so, what would the upgrade process look like? Is there a way for magnum to push information about pending update to the cluster component, or would it have to query magnum periodically?",
+      "parentUuid": "ccc74f49_554dd87b",
+      "revId": "0a92b33b61ec519d1afbbed6217cf225a02927a6",
+      "serverId": "4a232e18-c5a9-48ee-94c0-e04e7cca6543",
+      "unresolved": true
     }
   ]
 }