Feat: adding configurable parameter for custom role label in multiple resources

charts/postgres-operator/crds/operatorconfigurations.yaml

@@ -327,6 +327,9 @@ spec:
                   pod_role_label:
                     type: string
                     default: "spilo-role"
+                  pod_leader_label_value:
+                    type: string
+                    default: "master"
                   pod_service_account_definition:
                     type: string
                     default: ""

charts/postgres-operator/values.yaml

@@ -194,6 +194,7 @@ configKubernetes:
   pod_management_policy: "ordered_ready"
   # label assigned to the Postgres pods (and services/endpoints)
   pod_role_label: spilo-role
+  pod_leader_label_value: master
   # service account definition as JSON/YAML string to be used by postgres cluster pods
   # pod_service_account_definition: ""
 

docs/reference/operator_parameters.md

@@ -434,6 +434,10 @@ configuration they are grouped under the `kubernetes` key.
   name of the label assigned to the Postgres pods (and services/endpoints) by
   the operator. The default is `spilo-role`.
 
+* **pod_leader_label_value**
+  value of the pod label if Postgres role is primary when running on Kubernetes.
+  The default is 'master'.
+
 * **cluster_labels**
   list of `name:value` pairs for additional labels assigned to the cluster
   objects. The default is `application:spilo`.

e2e/tests/constants.py

@@ -0,0 +1,4 @@
+LEADER_LABEL_VALUE = "master" # value should be the same as in the configmap: pod_leader_label_value
+SPILO_CURRENT = "registry.opensource.zalan.do/acid/spilo-17-e2e:0.3"
+SPILO_LAZY = "registry.opensource.zalan.do/acid/spilo-17-e2e:0.4"
+SPILO_FULL_IMAGE = "ghcr.io/zalando/spilo-17:4.0-p2"

e2e/tests/k8s_api.py

@@ -6,6 +6,7 @@
 from kubernetes import client, config
 from kubernetes.client.rest import ApiException
 
+from tests.constants import LEADER_LABEL_VALUE
 
 def to_selector(labels):
     return ",".join(["=".join(lbl) for lbl in labels.items()])
@@ -47,7 +48,7 @@ def get_pg_nodes(self, pg_cluster_name, namespace='default'):
         replica_pod_nodes = []
         podsList = self.api.core_v1.list_namespaced_pod(namespace, label_selector=pg_cluster_name)
         for pod in podsList.items:
-            if pod.metadata.labels.get('spilo-role') == 'master':
+            if pod.metadata.labels.get('spilo-role') == LEADER_LABEL_VALUE:
                 master_pod_node = pod.spec.node_name
             elif pod.metadata.labels.get('spilo-role') == 'replica':
                 replica_pod_nodes.append(pod.spec.node_name)
@@ -59,7 +60,7 @@ def get_cluster_nodes(self, cluster_labels='application=spilo,cluster-name=acid-
         r = []
         podsList = self.api.core_v1.list_namespaced_pod(namespace, label_selector=cluster_labels)
         for pod in podsList.items:
-            if pod.metadata.labels.get('spilo-role') == 'master' and pod.status.phase == 'Running':
+            if pod.metadata.labels.get('spilo-role') == LEADER_LABEL_VALUE and pod.status.phase == 'Running':
                 m.append(pod.spec.node_name)
             elif pod.metadata.labels.get('spilo-role') == 'replica' and pod.status.phase == 'Running':
                 r.append(pod.spec.node_name)
@@ -351,7 +352,7 @@ def get_cluster_pod(self, role, labels='application=spilo,cluster-name=acid-mini
             return pods[0]
 
     def get_cluster_leader_pod(self, labels='application=spilo,cluster-name=acid-minimal-cluster', namespace='default'):
-        return self.get_cluster_pod('master', labels, namespace)
+        return self.get_cluster_pod(LEADER_LABEL_VALUE, labels, namespace)
 
     def get_cluster_replica_pod(self, labels='application=spilo,cluster-name=acid-minimal-cluster', namespace='default'):
         return self.get_cluster_pod('replica', labels, namespace)
@@ -383,7 +384,7 @@ def get_pg_nodes(self, pg_cluster_labels='cluster-name=acid-minimal-cluster', na
         replica_pod_nodes = []
         podsList = self.api.core_v1.list_namespaced_pod(namespace, label_selector=pg_cluster_labels)
         for pod in podsList.items:
-            if pod.metadata.labels.get('spilo-role') == 'master':
+            if pod.metadata.labels.get('spilo-role') == LEADER_LABEL_VALUE:
                 master_pod_node = pod.spec.node_name
             elif pod.metadata.labels.get('spilo-role') == 'replica':
                 replica_pod_nodes.append(pod.spec.node_name)
@@ -395,7 +396,7 @@ def get_cluster_nodes(self, cluster_labels='cluster-name=acid-minimal-cluster',
         r = []
         podsList = self.api.core_v1.list_namespaced_pod(namespace, label_selector=cluster_labels)
         for pod in podsList.items:
-            if pod.metadata.labels.get('spilo-role') == 'master' and pod.status.phase == 'Running':
+            if pod.metadata.labels.get('spilo-role') == LEADER_LABEL_VALUE and pod.status.phase == 'Running':
                 m.append(pod.spec.node_name)
             elif pod.metadata.labels.get('spilo-role') == 'replica' and pod.status.phase == 'Running':
                 r.append(pod.spec.node_name)
@@ -622,7 +623,7 @@ def get_pg_nodes(self):
         replica_pod_nodes = []
         podsList = self.api.core_v1.list_namespaced_pod(self.namespace, label_selector=self.labels)
         for pod in podsList.items:
-            if pod.metadata.labels.get('spilo-role') == 'master':
+            if pod.metadata.labels.get('spilo-role') == LEADER_LABEL_VALUE:
                 master_pod_node = pod.spec.node_name
             elif pod.metadata.labels.get('spilo-role') == 'replica':
                 replica_pod_nodes.append(pod.spec.node_name)

e2e/tests/test_e2e.py

@@ -8,13 +8,11 @@
 
 from datetime import datetime, date, timedelta
 from kubernetes import client
+from kubernetes.client.rest import ApiException
 
 from tests.k8s_api import K8s
-from kubernetes.client.rest import ApiException
+from tests.constants import SPILO_CURRENT, SPILO_FULL_IMAGE, SPILO_LAZY, LEADER_LABEL_VALUE
 
-SPILO_CURRENT = "registry.opensource.zalan.do/acid/spilo-17-e2e:0.3"
-SPILO_LAZY = "registry.opensource.zalan.do/acid/spilo-17-e2e:0.4"
-SPILO_FULL_IMAGE = "ghcr.io/zalando/spilo-17:4.0-p2"
 
 def to_selector(labels):
     return ",".join(["=".join(lbl) for lbl in labels.items()])
@@ -154,7 +152,7 @@ def setUpClass(cls):
         result = k8s.create_with_kubectl("manifests/minimal-postgres-manifest.yaml")
         print('stdout: {}, stderr: {}'.format(result.stdout, result.stderr))
         try:
-            k8s.wait_for_pod_start('spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_start('spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
         except timeout_decorator.TimeoutError:
             print('Operator log: {}'.format(k8s.get_operator_log()))
@@ -223,7 +221,7 @@ def test_additional_pod_capabilities(self):
             k8s.update_config(patch_capabilities)
 
             # changed security context of postgres container should trigger a rolling update
-            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
             self.eventuallyEqual(lambda: k8s.get_operator_state(), {"0": "idle"}, "Operator does not get in sync")
@@ -657,7 +655,7 @@ def test_custom_ssl_certificate(self):
                 "acid.zalan.do", "v1", "default", "postgresqls", "acid-minimal-cluster", pg_patch_tls)
 
             # wait for switched over
-            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
             self.eventuallyEqual(lambda: k8s.count_pods_with_env_variable("SSL_CERTIFICATE_FILE", cluster_label), 2, "TLS env variable SSL_CERTIFICATE_FILE missing in Spilo pods")
@@ -860,7 +858,7 @@ def test_enable_load_balancer(self):
         k8s = self.k8s
         cluster_label = 'application=spilo,cluster-name=acid-minimal-cluster,spilo-role={}'
 
-        self.eventuallyEqual(lambda: k8s.get_service_type(cluster_label.format("master")),
+        self.eventuallyEqual(lambda: k8s.get_service_type(cluster_label.format(LEADER_LABEL_VALUE)),
                              'ClusterIP',
                              "Expected ClusterIP type initially, found {}")
 
@@ -875,7 +873,7 @@ def test_enable_load_balancer(self):
             k8s.api.custom_objects_api.patch_namespaced_custom_object(
                 "acid.zalan.do", "v1", "default", "postgresqls", "acid-minimal-cluster", pg_patch_enable_lbs)
 
-            self.eventuallyEqual(lambda: k8s.get_service_type(cluster_label.format("master")),
+            self.eventuallyEqual(lambda: k8s.get_service_type(cluster_label.format(LEADER_LABEL_VALUE)),
                                  'LoadBalancer',
                                  "Expected LoadBalancer service type for master, found {}")
 
@@ -893,7 +891,7 @@ def test_enable_load_balancer(self):
             k8s.api.custom_objects_api.patch_namespaced_custom_object(
                 "acid.zalan.do", "v1", "default", "postgresqls", "acid-minimal-cluster", pg_patch_disable_lbs)
 
-            self.eventuallyEqual(lambda: k8s.get_service_type(cluster_label.format("master")),
+            self.eventuallyEqual(lambda: k8s.get_service_type(cluster_label.format(LEADER_LABEL_VALUE)),
                                  'ClusterIP',
                                  "Expected LoadBalancer service type for master, found {}")
 
@@ -1226,7 +1224,7 @@ def get_annotations():
         self.eventuallyEqual(lambda: k8s.get_operator_state(), {"0": "idle"}, "Operator does not get in sync")
 
         k8s.wait_for_pod_failover(master_nodes, 'spilo-role=replica,' + cluster_label)
-        k8s.wait_for_pod_start('spilo-role=master,' + cluster_label)
+        k8s.wait_for_pod_start('spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
         k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
         self.eventuallyEqual(check_version, 14, "Version should be upgraded from 13 to 14")
 
@@ -1251,8 +1249,8 @@ def get_annotations():
             "acid.zalan.do", "v1", "default", "postgresqls", "acid-upgrade-test", pg_patch_version_15)
         self.eventuallyEqual(lambda: k8s.get_operator_state(), {"0": "idle"}, "Operator does not get in sync")
 
-        k8s.wait_for_pod_failover(master_nodes, 'spilo-role=master,' + cluster_label)
-        k8s.wait_for_pod_start('spilo-role=master,' + cluster_label)
+        k8s.wait_for_pod_failover(master_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
+        k8s.wait_for_pod_start('spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
         k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
         self.eventuallyEqual(check_version, 14, "Version should not be upgraded")
 
@@ -1277,7 +1275,7 @@ def get_annotations():
         self.eventuallyEqual(lambda: k8s.get_operator_state(), {"0": "idle"}, "Operator does not get in sync")
 
         k8s.wait_for_pod_failover(master_nodes, 'spilo-role=replica,' + cluster_label)
-        k8s.wait_for_pod_start('spilo-role=master,' + cluster_label)
+        k8s.wait_for_pod_start('spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
         k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
         self.eventuallyEqual(check_version, 16, "Version should be upgraded from 14 to 16")
 
@@ -1303,8 +1301,8 @@ def get_annotations():
             "acid.zalan.do", "v1", "default", "postgresqls", "acid-upgrade-test", pg_patch_version_17)
         self.eventuallyEqual(lambda: k8s.get_operator_state(), {"0": "idle"}, "Operator does not get in sync")
 
-        k8s.wait_for_pod_failover(master_nodes, 'spilo-role=master,' + cluster_label)
-        k8s.wait_for_pod_start('spilo-role=master,' + cluster_label)
+        k8s.wait_for_pod_failover(master_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
+        k8s.wait_for_pod_start('spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
         k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
         self.eventuallyEqual(check_version, 16, "Version should not be upgraded because annotation for last upgrade's failure is set")
 
@@ -1314,7 +1312,7 @@ def get_annotations():
         self.eventuallyEqual(lambda: k8s.get_operator_state(), {"0": "idle"}, "Operator does not get in sync")
 
         k8s.wait_for_pod_failover(master_nodes, 'spilo-role=replica,' + cluster_label)
-        k8s.wait_for_pod_start('spilo-role=master,' + cluster_label)
+        k8s.wait_for_pod_start('spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
         k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
         fourth_annotations = get_annotations()
@@ -1432,7 +1430,7 @@ def test_resource_generation(self):
                              "Operator does not get in sync")
 
         # wait for switched over
-        k8s.wait_for_pod_failover(replica_nodes, 'spilo-role=master,' + cluster_label)
+        k8s.wait_for_pod_failover(replica_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
         k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
         def verify_pod_resources():
@@ -1464,7 +1462,7 @@ def test_multi_namespace_support(self):
 
         try:
             k8s.create_with_kubectl("manifests/complete-postgres-manifest.yaml")
-            k8s.wait_for_pod_start("spilo-role=master", self.test_namespace)
+            k8s.wait_for_pod_start("spilo-role={}".format(LEADER_LABEL_VALUE), self.test_namespace)
             k8s.wait_for_pod_start("spilo-role=replica", self.test_namespace)
             self.assert_master_is_unique(self.test_namespace, "acid-test-cluster")
             # acid-test-cluster will be deleted in test_owner_references test
@@ -1539,7 +1537,7 @@ def test_node_affinity(self):
             k8s.wait_for_pod_failover(master_nodes, 'spilo-role=replica,' + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
             # next master will be switched over and pod needs to be replaced as well to finish the rolling update
-            k8s.wait_for_pod_failover(master_nodes, 'spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_failover(master_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
             podsList = k8s.api.core_v1.list_namespaced_pod('default', label_selector=cluster_label)
@@ -1572,7 +1570,7 @@ def test_node_affinity(self):
 
             # node affinity change should cause another rolling update and relocation of replica
             k8s.wait_for_pod_failover(master_nodes, 'spilo-role=replica,' + cluster_label)
-            k8s.wait_for_pod_start('spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_start('spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
         except timeout_decorator.TimeoutError:
@@ -1633,7 +1631,7 @@ def test_node_readiness_label(self):
                 k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
             # next master will be switched over and pod needs to be replaced as well to finish the rolling update
-            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
             # patch also node where master ran before
@@ -1921,7 +1919,7 @@ def test_rolling_update_flag(self):
             podsList = k8s.api.core_v1.list_namespaced_pod('default', label_selector=cluster_label)
             for pod in podsList.items:
                 # add flag only to the master to make it appear to the operator as a leftover from a rolling update
-                if pod.metadata.labels.get('spilo-role') == 'master':
+                if pod.metadata.labels.get('spilo-role') == LEADER_LABEL_VALUE:
                     old_creation_timestamp = pod.metadata.creation_timestamp
                     k8s.patch_pod(flag, pod.metadata.name, pod.metadata.namespace)
                 else:
@@ -1932,7 +1930,7 @@ def test_rolling_update_flag(self):
             k8s.delete_operator_pod()
 
             # operator should now recreate the master pod and do a switchover before
-            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
             # check if the former replica is now the new master
@@ -2003,7 +2001,7 @@ def test_rolling_update_label_timeout(self):
             self.eventuallyEqual(lambda: k8s.pg_get_status(), "SyncFailed", "Expected SYNC event to fail")
 
             # wait for next sync, replica should be running normally by now and be ready for switchover
-            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role=master,' + cluster_label)
+            k8s.wait_for_pod_failover(replica_nodes, 'spilo-role={},'.format(LEADER_LABEL_VALUE) + cluster_label)
             k8s.wait_for_pod_start('spilo-role=replica,' + cluster_label)
 
             # check if the former replica is now the new master
@@ -2078,7 +2076,7 @@ def test_service_annotations(self):
             "alice": "bob"
         }
 
-        self.eventuallyTrue(lambda: k8s.check_service_annotations("cluster-name=acid-minimal-cluster,spilo-role=master", annotations), "Wrong annotations")
+        self.eventuallyTrue(lambda: k8s.check_service_annotations("cluster-name=acid-minimal-cluster,spilo-role={}".format(LEADER_LABEL_VALUE), annotations), "Wrong annotations")
         self.eventuallyTrue(lambda: k8s.check_service_annotations("cluster-name=acid-minimal-cluster,spilo-role=replica", annotations), "Wrong annotations")
 
         # clean up
@@ -2150,7 +2148,7 @@ def test_standby_cluster(self):
 
         try:
             k8s.create_with_kubectl("manifests/standby-manifest.yaml")
-            k8s.wait_for_pod_start("spilo-role=master," + cluster_label)
+            k8s.wait_for_pod_start("spilo-role={},".format(LEADER_LABEL_VALUE) + cluster_label)
 
         except timeout_decorator.TimeoutError:
             print('Operator log: {}'.format(k8s.get_operator_log()))
@@ -2456,11 +2454,11 @@ def test_zz_cluster_deletion(self):
 
     def assert_master_is_unique(self, namespace='default', clusterName="acid-minimal-cluster"):
         '''
-           Check that there is a single pod in the k8s cluster with the label "spilo-role=master"
+           Check that there is a single pod in the k8s cluster with the label "spilo-role=primary" or "spilo-role=master"
            To be called manually after operations that affect pods
         '''
         k8s = self.k8s
-        labels = 'spilo-role=master,cluster-name=' + clusterName
+        labels = 'spilo-role={},cluster-name='.format(LEADER_LABEL_VALUE) + clusterName
 
         num_of_master_pods = k8s.count_pods_with_label(labels, namespace)
         self.assertEqual(num_of_master_pods, 1, "Expected 1 master pod, found {}".format(num_of_master_pods))

manifests/configmap.yaml

@@ -134,6 +134,7 @@ data:
   pod_management_policy: "ordered_ready"
   # pod_priority_class_name: "postgres-pod-priority"
   pod_role_label: spilo-role
+  pod_leader_label_value: master
   pod_service_account_definition: ""
   pod_service_account_name: "postgres-pod"
   pod_service_account_role_binding_definition: ""

manifests/operatorconfiguration.crd.yaml

@@ -325,6 +325,9 @@ spec:
                   pod_role_label:
                     type: string
                     default: "spilo-role"
+                  pod_leader_label_value:
+                    type: string
+                    default: "master"
                   pod_service_account_definition:
                     type: string
                     default: ""

manifests/postgresql-operator-default-configuration.yaml

@@ -99,6 +99,7 @@ configuration:
     pod_management_policy: "ordered_ready"
     # pod_priority_class_name: "postgres-pod-priority"
     pod_role_label: spilo-role
+    pod_leader_label_value: master
     # pod_service_account_definition: ""
     pod_service_account_name: postgres-pod
     # pod_service_account_role_binding_definition: ""