CLOUDP-316400: Update upgrade tests for 1.1 (#87)

# Summary

Upgrade tests are currently doing the path MEKO -> MCK 1.0

We need to make a few updates to the fixtures to change that to: latest
released MCK -> next MCK. E.g `1.0` -> `1.1` at the time this PR is
opened.

# Proof of Work

E2E tests pass + they install the right operator versions (helm commands
in logs)

Author: Julien-Ben

.evergreen-tasks.yml

@@ -170,7 +170,7 @@ tasks:
     commands:
       - func: "e2e_test"
 
-  - name: e2e_operator_upgrade_sharded_cluster
+  - name: e2e_sharded_cluster_operator_upgrade_v1_27_to_mck
     tags: [ "patch-run" ]
     commands:
       - func: "e2e_test"
@@ -180,7 +180,7 @@ tasks:
     commands:
       - func: "e2e_test"
 
-  - name: e2e_operator_upgrade_appdb_tls
+  - name: e2e_appdb_tls_operator_upgrade_v1_32_to_mck
     tags: [ "patch-run" ]
     commands:
       - func: "e2e_test"
@@ -840,7 +840,7 @@ tasks:
     commands:
       - func: "e2e_test"
 
-  - name: e2e_multi_cluster_appdb_state_operator_upgrade_downgrade
+  - name: e2e_multi_cluster_appdb_upgrade_downgrade_v1_27_to_mck
     tags: [ "patch-run" ]
     commands:
       - func: "e2e_test"

.evergreen.yml

@@ -778,13 +778,13 @@ task_groups:
     <<: *setup_and_teardown_task_cloudqa
     tasks:
       - e2e_operator_upgrade_replica_set
-      - e2e_operator_upgrade_sharded_cluster
+      - e2e_sharded_cluster_operator_upgrade_v1_27_to_mck
       - e2e_operator_upgrade_ops_manager
       - e2e_operator_proxy
       - e2e_operator_partial_crd
       - e2e_operator_clusterwide
       - e2e_operator_multi_namespaces
-      - e2e_operator_upgrade_appdb_tls
+      - e2e_appdb_tls_operator_upgrade_v1_32_to_mck
       - e2e_meko_mck_upgrade
     <<: *teardown_group
 
@@ -806,13 +806,13 @@ task_groups:
     <<: *setup_and_teardown_task_cloudqa
     tasks:
       - e2e_operator_upgrade_replica_set
-      - e2e_operator_upgrade_sharded_cluster
+      - e2e_sharded_cluster_operator_upgrade_v1_27_to_mck
       - e2e_operator_upgrade_ops_manager
       - e2e_operator_proxy
       - e2e_operator_partial_crd
       - e2e_operator_clusterwide
       - e2e_operator_multi_namespaces
-      - e2e_operator_upgrade_appdb_tls
+      - e2e_appdb_tls_operator_upgrade_v1_32_to_mck
       - e2e_meko_mck_upgrade
     <<: *teardown_group
 
@@ -910,7 +910,7 @@ task_groups:
       - e2e_multi_cluster_om_networking_clusterwide
       - e2e_multi_cluster_om_appdb_no_mesh
       # Reused OM tests with AppDB Multi-Cluster topology
-      - e2e_operator_upgrade_appdb_tls
+      - e2e_appdb_tls_operator_upgrade_v1_32_to_mck
       - e2e_om_appdb_flags_and_config
       - e2e_om_appdb_upgrade
       - e2e_om_appdb_monitoring_tls
@@ -942,7 +942,7 @@ task_groups:
       - e2e_om_ops_manager_upgrade
       - e2e_om_update_before_reconciliation
       - e2e_om_feature_controls
-      - e2e_multi_cluster_appdb_state_operator_upgrade_downgrade
+      - e2e_multi_cluster_appdb_upgrade_downgrade_v1_27_to_mck
       - e2e_multi_cluster_sharded_disaster_recovery
     # disabled tests:
     #    - e2e_om_multiple # multi-cluster failures in EVG
@@ -995,7 +995,7 @@ task_groups:
       - e2e_om_ops_manager_backup_kmip
       - e2e_om_ops_manager_scale
       - e2e_om_ops_manager_upgrade
-      - e2e_multi_cluster_appdb_state_operator_upgrade_downgrade
+      - e2e_multi_cluster_appdb_upgrade_downgrade_v1_27_to_mck
       - e2e_om_update_before_reconciliation
       - e2e_om_feature_controls
       - e2e_multi_cluster_sharded_disaster_recovery

docker/mongodb-kubernetes-tests/tests/conftest.py

@@ -61,6 +61,14 @@
 LEGACY_CENTRAL_CLUSTER_NAME: str = "__default"
 LEGACY_DEPLOYMENT_STATE_VERSION: str = "1.27.0"
 
+# Helm charts
+LEGACY_OPERATOR_CHART = "mongodb/enterprise-operator"
+MCK_HELM_CHART = "mongodb/mongodb-kubernetes"
+LOCAL_HELM_CHART_DIR = "helm_chart"
+
+OFFICIAL_OPERATOR_IMAGE_NAME = "mongodb-kubernetes"
+LEGACY_OPERATOR_IMAGE_NAME = "mongodb-enterprise-operator-ubi"
+
 # Names for operator and RBAC
 OPERATOR_NAME = "mongodb-kubernetes-operator"
 MULTI_CLUSTER_OPERATOR_NAME = OPERATOR_NAME + "-multi-cluster"
@@ -801,7 +809,7 @@ def _install_multi_cluster_operator(
     helm_opts: Dict[str, str],
     central_cluster_name: str,
     operator_name: Optional[str] = MULTI_CLUSTER_OPERATOR_NAME,
-    helm_chart_path: Optional[str] = "helm_chart",
+    helm_chart_path: Optional[str] = LOCAL_HELM_CHART_DIR,
     custom_operator_version: Optional[str] = None,
 ) -> Operator:
     multi_cluster_operator_installation_config.update(helm_opts)
@@ -856,6 +864,9 @@ def official_operator(
         central_cluster_client,
         member_cluster_clients,
         member_cluster_names,
+        helm_chart_path=MCK_HELM_CHART,
+        operator_name=OPERATOR_NAME,
+        operator_image=OFFICIAL_OPERATOR_IMAGE_NAME,
     )
 
 
@@ -877,8 +888,42 @@ def official_meko_operator(
         central_cluster_client,
         member_cluster_clients,
         member_cluster_names,
+        helm_chart_path=LEGACY_OPERATOR_CHART,
+        operator_name=LEGACY_OPERATOR_NAME,
+        operator_image=LEGACY_OPERATOR_IMAGE_NAME,
+    )
+
+
+# In 1.27.0, we changed the way we manage the deployment state. We use this fixture to test the upgrade path from this
+# version, to ensure we do not break the state
+def install_legacy_deployment_state_meko(
+    namespace: str,
+    managed_security_context: str,
+    operator_installation_config: dict[str, str],
+    central_cluster_name=None,  # These 4 fields apply to multi cluster operator only
+    central_cluster_client=None,
+    member_cluster_clients=None,
+    member_cluster_names=None,
+):
+    logger.info(
+        f"Installing the operator from chart {LEGACY_OPERATOR_CHART}, with version {LEGACY_DEPLOYMENT_STATE_VERSION}"
+    )
+    operator = install_official_operator(
+        namespace,
+        managed_security_context,
+        operator_installation_config,
+        central_cluster_name=central_cluster_name,  # These 4 fields apply to multi cluster operator only
+        central_cluster_client=central_cluster_client,
+        member_cluster_clients=member_cluster_clients,
+        member_cluster_names=member_cluster_names,
+        custom_operator_version=LEGACY_DEPLOYMENT_STATE_VERSION,
+        helm_chart_path=LEGACY_OPERATOR_CHART,  # We are testing the upgrade from legacy state management, introduced in MEKO
         operator_name=LEGACY_OPERATOR_NAME,
+        operator_image=LEGACY_OPERATOR_IMAGE_NAME,
     )
+    operator.assert_is_running()
+    # Dumping deployments in logs ensures we are using the correct operator version
+    log_deployments_info(namespace)
 
 
 def install_official_operator(
@@ -890,7 +935,9 @@ def install_official_operator(
     member_cluster_clients: Optional[List[MultiClusterClient]],
     member_cluster_names: Optional[List[str]],
     custom_operator_version: Optional[str] = None,
+    helm_chart_path: Optional[str] = MCK_HELM_CHART,
     operator_name: Optional[str] = OPERATOR_NAME,
+    operator_image: Optional[str] = OFFICIAL_OPERATOR_IMAGE_NAME,
 ) -> Operator:
     """
     Installs the Operator from the official Helm Chart.
@@ -912,7 +959,7 @@ def install_official_operator(
     # Note, that we don't intend to install the official Operator to standalone clusters (kops/openshift) as we want to
     # avoid damaged CRDs. But we may need to install the "openshift like" environment to Kind instead of the "ubi"
     # images are used for installing the dev Operator
-    helm_args["operator.operator_image_name"] = "{}-ubi".format(operator_name)
+    helm_args["operator.operator_image_name"] = operator_image
 
     # Note:
     # We might want in the future to install CRDs when performing upgrade/downgrade tests, the helm install only takes
@@ -966,7 +1013,7 @@ def install_official_operator(
             member_cluster_clients,
             helm_opts=helm_args,
             central_cluster_name=get_central_cluster_name(),
-            helm_chart_path="mongodb/enterprise-operator",
+            helm_chart_path=helm_chart_path,
             custom_operator_version=custom_operator_version,
             operator_name=operator_name,
         )
@@ -976,7 +1023,7 @@ def install_official_operator(
         return Operator(
             namespace=namespace,
             helm_args=helm_args,
-            helm_chart_path="mongodb/enterprise-operator",
+            helm_chart_path=helm_chart_path,
             name=operator_name,
         ).install(custom_operator_version=custom_operator_version)
 

docker/mongodb-kubernetes-tests/tests/multicluster_appdb/multicluster_appdb_state_operator_upgrade_downgrade.py renamed to docker/mongodb-kubernetes-tests/tests/multicluster_appdb/multicluster_appdb_upgrade_downgrade_v1_27_to_mck.py

@@ -10,14 +10,12 @@
 from pytest import fixture, mark
 from tests import test_logger
 from tests.conftest import (
-    LEGACY_DEPLOYMENT_STATE_VERSION,
     LEGACY_MULTI_CLUSTER_OPERATOR_NAME,
-    LEGACY_OPERATOR_NAME,
     MULTI_CLUSTER_OPERATOR_NAME,
     create_appdb_certs,
     get_central_cluster_name,
     get_custom_appdb_version,
-    install_official_operator,
+    install_legacy_deployment_state_meko,
     log_deployments_info,
 )
 from tests.multicluster.conftest import cluster_spec_list
@@ -42,6 +40,8 @@
 At each step, we verify that the state is correct
 """
 
+# TODO CLOUDP-318100: this test should eventually be updated and not pinned to 1.27 anymore
+
 
 def assert_cm_expected_data(
     name: str, namespace: str, expected_data: Optional[Dict], central_cluster_client: kubernetes.client.ApiClient
@@ -168,7 +168,7 @@ def ops_manager(
     return resource
 
 
-@mark.e2e_multi_cluster_appdb_state_operator_upgrade_downgrade
+@mark.e2e_multi_cluster_appdb_upgrade_downgrade_v1_27_to_mck
 class TestOpsManagerCreation:
     """
     Ensure correct deployment and state of AppDB, with operator version 1.27 installed.
@@ -186,23 +186,15 @@ def test_install_legacy_state_official_operator(
         member_cluster_clients,
         member_cluster_names,
     ):
-        logger.info(
-            f"Installing the official operator from helm charts, with version {LEGACY_DEPLOYMENT_STATE_VERSION}"
-        )
-        operator = install_official_operator(
-            namespace,
-            managed_security_context,
-            operator_installation_config,
-            central_cluster_name,
-            central_cluster_client,
-            member_cluster_clients,
-            member_cluster_names,
-            custom_operator_version=LEGACY_DEPLOYMENT_STATE_VERSION,
-            operator_name=LEGACY_OPERATOR_NAME,
+        install_legacy_deployment_state_meko(
+            namespace=namespace,
+            managed_security_context=managed_security_context,
+            operator_installation_config=operator_installation_config,
+            central_cluster_name=central_cluster_name,
+            central_cluster_client=central_cluster_client,
+            member_cluster_clients=member_cluster_clients,
+            member_cluster_names=member_cluster_names,
         )
-        operator.assert_is_running()
-        # Dumping deployments in logs ensure we are using the correct operator version
-        log_deployments_info(namespace)
 
     def test_create_appdb_certs_secret(
         self,
@@ -240,15 +232,15 @@ def test_state_correctness(
         assert_cm_expected_data(configmap_name, namespace, expected_data, central_cluster_client)
 
 
-@mark.e2e_multi_cluster_appdb_state_operator_upgrade_downgrade
+@mark.e2e_multi_cluster_appdb_upgrade_downgrade_v1_27_to_mck
 class TestOperatorUpgrade:
     """
     Upgrade the operator to latest dev version, scale AppDB, and ensure state correctness.
     """
 
     def test_downscale_latest_official_operator(self, namespace: str):
-        # Scale down the existing operator deployment to 0. This is needed as long as the
-        # `official_operator` fixture installs the MEKO operator.
+        # Scale down the existing operator deployment to 0. This is needed as we are initially installing MEKO
+        # and replacing it with MCK
         downscale_operator_deployment(deployment_name=LEGACY_MULTI_CLUSTER_OPERATOR_NAME, namespace=namespace)
 
     def test_install_default_operator(self, namespace: str, multi_cluster_operator: Operator):
@@ -285,15 +277,13 @@ def test_old_state_still_exists(
         )
 
 
-@mark.e2e_multi_cluster_appdb_state_operator_upgrade_downgrade
+@mark.e2e_multi_cluster_appdb_upgrade_downgrade_v1_27_to_mck
 class TestOperatorDowngrade:
     """
     Downgrade the Operator to 1.27, scale AppDB and ensure state correctness.
     """
 
     def test_downscale_default_operator(self, namespace: str):
-        # Scale down the existing operator deployment to 0. This is needed as long as the
-        # `official_operator` fixture installs the MEKO operator.
         downscale_operator_deployment(deployment_name=MULTI_CLUSTER_OPERATOR_NAME, namespace=namespace)
 
     def test_install_legacy_state_official_operator(
@@ -306,21 +296,15 @@ def test_install_legacy_state_official_operator(
         member_cluster_clients,
         member_cluster_names,
     ):
-        logger.info(f"Downgrading the operator to version {LEGACY_DEPLOYMENT_STATE_VERSION}, from helm chart release")
-        operator = install_official_operator(
-            namespace,
-            managed_security_context,
-            operator_installation_config,
-            central_cluster_name,
-            central_cluster_client,
-            member_cluster_clients,
-            member_cluster_names,
-            custom_operator_version=LEGACY_DEPLOYMENT_STATE_VERSION,
-            operator_name=LEGACY_OPERATOR_NAME,
+        install_legacy_deployment_state_meko(
+            namespace=namespace,
+            managed_security_context=managed_security_context,
+            operator_installation_config=operator_installation_config,
+            central_cluster_name=central_cluster_name,
+            central_cluster_client=central_cluster_client,
+            member_cluster_clients=member_cluster_clients,
+            member_cluster_names=member_cluster_names,
         )
-        operator.assert_is_running()
-        # Dumping deployments in logs ensure we are using the correct operator version
-        log_deployments_info(namespace)
 
     def test_om_running_after_downgrade(self, ops_manager: MongoDBOpsManager):
         ops_manager.load()