add checks for endpoint/endpoints field in status (#5)

The endpoint and endpoints field in the Domain.Status struct are filled
in when the domain's endpoint URL or URLs (if VPC endpoints) are
established. Adds an e2e test to verify these fields on the Status
struct are indeed populated, which after adding
requeue_on_success_seconds to the generator.yaml file, is indeed true.

Signed-off-by: Jay Pipes <[email protected]>

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

Author: jaypipes

apis/v1alpha1/ack-generate-metadata.yaml

@@ -1,13 +1,13 @@
 ack_generate_info:
-  build_date: "2021-11-04T22:37:28Z"
+  build_date: "2021-11-05T21:04:24Z"
   build_hash: 0a8fc32cdf2d33693e919e4a59da05599b9f916e
   go_version: go1.17
   version: v0.15.1
 api_directory_checksum: 550c87ef2158e5b3d58bc5a8ff5d3367c192b04a
 api_version: v1alpha1
 aws_sdk_go_version: v1.40.51
 generator_config_info:
-  file_checksum: 1aa3cbb5df5c91de25b9935978e30138af9b696d
+  file_checksum: bd75f4fe50f216466e8e62e7f5c40bd1550da7e3
   original_file_name: generator.yaml
 last_modification:
   reason: API generation

apis/v1alpha1/generator.yaml

@@ -9,6 +9,11 @@ ignore:
     - CreateDomainInput.SnapshotOptions
 resources:
   Domain:
+    reconcile:
+      # Doing this because it takes a LONG time for the Domain's
+      # endpoint/endpoints field to be populated, even after the Domain's
+      # Processing field is set to False after creation...
+      requeue_on_success_seconds: 60
     exceptions:
       errors:
         404:

generator.yaml

@@ -9,6 +9,11 @@ ignore:
     - CreateDomainInput.SnapshotOptions
 resources:
   Domain:
+    reconcile:
+      # Doing this because it takes a LONG time for the Domain's
+      # endpoint/endpoints field to be populated, even after the Domain's
+      # Processing field is set to False after creation...
+      requeue_on_success_seconds: 60
     exceptions:
       errors:
         404:

pkg/resource/domain/manager_factory.go

@@ -126,3 +126,19 @@ def get(domain_name):
     except c.exceptions.ResourceNotFoundException:
         return None
 
+
+# Apparently, there is an 'endpoint' and an 'endpoints' field for a domain.
+# The 'endpoint' field is filled in with a URL when the domain does *not* use
+# VPC endpoints. The 'endpoints' field is filled in with a map when the domain
+# *does* use VPC endpoints. The following two assertion functions verify this
+# "one or the other" behaviour
+def assert_endpoint(cr):
+    assert 'endpoint' in cr['status']
+    assert cr['status']['endpoint'] != ""
+    assert 'endpoints' not in cr['status'] or len(cr['status']['endpoints']) == 0
+
+
+def assert_endpoints(cr):
+    assert 'endpoint' not in cr['status'] or cr['status']['endpoint'] == ""
+    assert 'endpoints' in cr['status']
+    assert len(cr['status']['endpoints']) > 0

pkg/resource/domain/sdk.go

@@ -37,6 +37,13 @@
 # Status.Conditions contains a True ResourceSynced condition.
 CHECK_STATUS_WAIT_SECONDS = 60
 
+# It can take a LONG time for the domain's endpoint/endpoints field to be
+# returned, even after Processing=False and the ResourceSynced condition is set
+# to True on a domain. Domain resources are requeued on success which means the
+# controller will poll for latest status, including endpoint/endpoints, every
+# 30 seconds, so 2 minutes *should* be enough here.
+CHECK_ENDPOINT_WAIT_SECONDS = 60*2
+
 
 @dataclass
 class Domain:
@@ -93,7 +100,7 @@ def es_7_9_domain(os_client):
     time.sleep(CHECK_STATUS_WAIT_SECONDS)
     condition.assert_synced(ref)
 
-    yield resource
+    yield ref, resource
 
     # Delete the k8s resource on teardown of the module
     k8s.delete_custom_resource(ref)
@@ -135,7 +142,7 @@ def es_2d3m_multi_az_no_vpc_7_9_domain(os_client):
     time.sleep(CHECK_STATUS_WAIT_SECONDS)
     condition.assert_synced(ref)
 
-    yield resource
+    yield ref, resource
 
     # Delete the k8s resource on teardown of the module
     k8s.delete_custom_resource(ref)
@@ -187,7 +194,7 @@ def es_2d3m_multi_az_vpc_2_subnet7_9_domain(os_client, resources: BootstrapResou
     time.sleep(CHECK_STATUS_WAIT_SECONDS)
     condition.assert_synced(ref)
 
-    yield resource
+    yield ref, resource
 
     # Delete the k8s resource on teardown of the module
     k8s.delete_custom_resource(ref)
@@ -203,35 +210,56 @@ def es_2d3m_multi_az_vpc_2_subnet7_9_domain(os_client, resources: BootstrapResou
 @pytest.mark.canary
 class TestDomain:
     def test_create_delete_es_7_9(self, es_7_9_domain):
-        resource = es_7_9_domain
+        ref, resource = es_7_9_domain
+
+        latest = domain.get(resource.name)
 
-        aws_res = domain.get(resource.name)
+        assert latest['DomainStatus']['EngineVersion'] == 'Elasticsearch_7.9'
+        assert latest['DomainStatus']['Created'] == True
+        assert latest['DomainStatus']['ClusterConfig']['InstanceCount'] == resource.data_node_count
+        assert latest['DomainStatus']['ClusterConfig']['ZoneAwarenessEnabled'] == resource.is_zone_aware
 
-        assert aws_res['DomainStatus']['EngineVersion'] == 'Elasticsearch_7.9'
-        assert aws_res['DomainStatus']['Created'] == True
-        assert aws_res['DomainStatus']['ClusterConfig']['InstanceCount'] == resource.data_node_count
-        assert aws_res['DomainStatus']['ClusterConfig']['ZoneAwarenessEnabled'] == resource.is_zone_aware
+        time.sleep(CHECK_ENDPOINT_WAIT_SECONDS)
+
+        cr = k8s.get_resource(ref)
+        assert cr is not None
+        assert 'status' in cr
+        domain.assert_endpoint(cr)
 
     def test_create_delete_es_2d3m_multi_az_no_vpc_7_9(self, es_2d3m_multi_az_no_vpc_7_9_domain):
-        resource = es_2d3m_multi_az_no_vpc_7_9_domain
+        ref, resource = es_2d3m_multi_az_no_vpc_7_9_domain
+
+        latest = domain.get(resource.name)
 
-        aws_res = domain.get(resource.name)
+        assert latest['DomainStatus']['EngineVersion'] == 'Elasticsearch_7.9'
+        assert latest['DomainStatus']['Created'] == True
+        assert latest['DomainStatus']['ClusterConfig']['InstanceCount'] == resource.data_node_count
+        assert latest['DomainStatus']['ClusterConfig']['DedicatedMasterCount'] == resource.master_node_count
+        assert latest['DomainStatus']['ClusterConfig']['ZoneAwarenessEnabled'] == resource.is_zone_aware
 
-        assert aws_res['DomainStatus']['EngineVersion'] == 'Elasticsearch_7.9'
-        assert aws_res['DomainStatus']['Created'] == True
-        assert aws_res['DomainStatus']['ClusterConfig']['InstanceCount'] == resource.data_node_count
-        assert aws_res['DomainStatus']['ClusterConfig']['DedicatedMasterCount'] == resource.master_node_count
-        assert aws_res['DomainStatus']['ClusterConfig']['ZoneAwarenessEnabled'] == resource.is_zone_aware
+        time.sleep(CHECK_ENDPOINT_WAIT_SECONDS)
+
+        cr = k8s.get_resource(ref)
+        assert cr is not None
+        assert 'status' in cr
+        domain.assert_endpoint(cr)
 
     def test_create_delete_es_2d3m_multi_az_vpc_2_subnet7_9(self, es_2d3m_multi_az_vpc_2_subnet7_9_domain):
-        resource = es_2d3m_multi_az_vpc_2_subnet7_9_domain
+        ref, resource = es_2d3m_multi_az_vpc_2_subnet7_9_domain
+
+        latest = domain.get(resource.name)
+
+        assert latest['DomainStatus']['EngineVersion'] == 'Elasticsearch_7.9'
+        assert latest['DomainStatus']['Created'] == True
+        assert latest['DomainStatus']['ClusterConfig']['InstanceCount'] == resource.data_node_count
+        assert latest['DomainStatus']['ClusterConfig']['DedicatedMasterCount'] == resource.master_node_count
+        assert latest['DomainStatus']['ClusterConfig']['ZoneAwarenessEnabled'] == resource.is_zone_aware
+        assert latest['DomainStatus']['VPCOptions']['VPCId'] == resource.vpc_id
+        assert set(latest['DomainStatus']['VPCOptions']['SubnetIds']) == set(resource.vpc_subnets)
 
-        aws_res = domain.get(resource.name)
+        time.sleep(CHECK_ENDPOINT_WAIT_SECONDS)
 
-        assert aws_res['DomainStatus']['EngineVersion'] == 'Elasticsearch_7.9'
-        assert aws_res['DomainStatus']['Created'] == True
-        assert aws_res['DomainStatus']['ClusterConfig']['InstanceCount'] == resource.data_node_count
-        assert aws_res['DomainStatus']['ClusterConfig']['DedicatedMasterCount'] == resource.master_node_count
-        assert aws_res['DomainStatus']['ClusterConfig']['ZoneAwarenessEnabled'] == resource.is_zone_aware
-        assert aws_res['DomainStatus']['VPCOptions']['VPCId'] == resource.vpc_id
-        assert set(aws_res['DomainStatus']['VPCOptions']['SubnetIds']) == set(resource.vpc_subnets)
+        cr = k8s.get_resource(ref)
+        assert cr is not None
+        assert 'status' in cr
+        domain.assert_endpoints(cr)