the waiter is buggy for a nodegroup

The waiter will continue waiting when the nodes are ready.
I had pointed this out and wrote a custom funciton and it
got replaced by the waiter at some point. :/ this will
revert it back, but I need to test it first./

Signed-off-by: vsoch <[email protected]>

kubescaler/scaler/aws/cluster.py

@@ -778,6 +778,22 @@ def _create_nodegroup(self, node_group, nodegroup_name):
         if node_group is None:
             raise ValueError("Could not create nodegroup")
 
+        # DO NOT USE THE WAITER or uncomment, it is buggy and does not work.
+        # self.wait_for_nodegroup(nodegroup_name)
+        self.wait_for_nodes()
+
+        # Retrieve the same metadata if we had retrieved it
+        return self.eks.describe_nodegroup(
+            clusterName=self.cluster_name, nodegroupName=nodegroup_name
+        )
+
+    def wait_for_nodegroup(self, nodegroup_name):
+        """
+        Wait for the nodegroup to create.
+
+        In practice, this waiter is very buggy. It's easier sometimes
+        to query the nodes and check their status is Ready.
+        """
         try:
             print(f"Waiting for {nodegroup_name} nodegroup...")
             waiter = self.eks.get_waiter("nodegroup_active")
@@ -788,11 +804,6 @@ def _create_nodegroup(self, node_group, nodegroup_name):
             print(f"Waiting for nodegroup creation exceeded wait time: {e}")
             time.sleep(180)
 
-        # Retrieve the same metadata if we had retrieved it
-        return self.eks.describe_nodegroup(
-            clusterName=self.cluster_name, nodegroupName=nodegroup_name
-        )
-
     def delete_stack(self, stack_name):
         """
         Delete a stack and wait for it to be deleted