Make startup probes optional (#646)

Co-authored-by: Yevhen Ivantsov <[email protected]>

src/main/charts/bamboo/README.md

@@ -85,6 +85,7 @@ Kubernetes: `>=1.21.x-0`
 | bamboo.setPermissions | bool | `true` | Boolean to define whether to set local home directory permissions on startup of Bamboo container. Set to 'false' to disable this behaviour.  |
 | bamboo.shutdown.command | string | `"/shutdown-wait.sh"` | By default pods will be stopped via a [preStop hook](https://kubernetes.io/docs/concepts/containers/container-lifecycle-hooks/), using a script supplied by the Docker image. If any other shutdown behaviour is needed it can be achieved by overriding this value. Note that the shutdown command needs to wait for the application shutdown completely before exiting; see [the default command](https://bitbucket.org/atlassian-docker/docker-bamboo-server/src/master/shutdown-wait.sh) for details.  |
 | bamboo.shutdown.terminationGracePeriodSeconds | int | `30` | The termination grace period for pods during shutdown. This should be set to the internal grace period, plus a small buffer to allow the JVM to fully terminate.  |
+| bamboo.startupProbe.enabled | bool | `true` | Whether to apply the startupProbe check to pod.  |
 | bamboo.startupProbe.failureThreshold | int | `120` | The number of consecutive failures of the Bamboo container startup probe before the pod fails startup checks.  |
 | bamboo.startupProbe.periodSeconds | int | `5` | How often (in seconds) the Bamboo container startup probe will run  |
 | bamboo.sysadminCredentials | object | `{"displayNameSecretKey":"displayName","emailAddressSecretKey":"emailAddress","passwordSecretKey":"password","secretName":null,"usernameSecretKey":"username"}` | The admin user configuration, and credentials that Bamboo should use. If supplied here the admin configuration will be skipped in the setup wizard.  |

src/main/charts/bamboo/templates/statefulset.yaml

@@ -178,11 +178,13 @@ spec:
             failureThreshold: {{ .Values.bamboo.readinessProbe.failureThreshold }}
             {{- end }}
           {{- end }}
+          {{- if .Values.bamboo.startupProbe.enabled }}
           startupProbe:
             tcpSocket:
               port: {{ .Values.bamboo.ports.http }}
             periodSeconds: {{ .Values.bamboo.startupProbe.periodSeconds }}
             failureThreshold: {{ .Values.bamboo.startupProbe.failureThreshold }}
+          {{- end }}
           {{- if .Values.bamboo.livenessProbe.enabled }}
           livenessProbe:
             tcpSocket:

src/main/charts/bamboo/values.yaml

@@ -630,6 +630,10 @@ bamboo:
   #
   startupProbe:
 
+    # -- Whether to apply the startupProbe check to pod.
+    #
+    enabled: true
+
     # -- How often (in seconds) the Bamboo container startup probe will run
     #
     periodSeconds: 5

src/main/charts/bitbucket/README.md

@@ -120,6 +120,7 @@ Kubernetes: `>=1.21.x-0`
 | bitbucket.sshService.loadBalancerIP | string | `nil` | Use specific loadBalancerIP. Only applies to service type LoadBalancer.  |
 | bitbucket.sshService.port | int | `22` | Port to expose the SSH service on.  |
 | bitbucket.sshService.type | string | `"LoadBalancer"` | SSH Service type  |
+| bitbucket.startupProbe.enabled | bool | `true` | Whether to apply the startupProbe check to pod.  |
 | bitbucket.startupProbe.failureThreshold | int | `120` | The number of consecutive failures of the Bitbucket container startup probe before the pod fails startup checks.  |
 | bitbucket.startupProbe.periodSeconds | int | `5` | How often (in seconds) the Bitbucket container startup probe will run  |
 | bitbucket.sysadminCredentials.displayNameSecretKey | string | `"displayName"` | The key in the Kubernetes Secret that contains the sysadmin display name  |

src/main/charts/bitbucket/templates/statefulset.yaml

@@ -100,11 +100,13 @@ spec:
             failureThreshold: {{ .Values.bitbucket.readinessProbe.failureThreshold }}
             {{- end }}
           {{- end }}
+          {{- if .Values.bitbucket.startupProbe.enabled }}
           startupProbe:
             tcpSocket:
               port: {{ .Values.bitbucket.ports.http }}
             periodSeconds: {{ .Values.bitbucket.startupProbe.periodSeconds }}
             failureThreshold: {{ .Values.bitbucket.startupProbe.failureThreshold }}
+          {{- end }}
           {{- if .Values.bitbucket.livenessProbe.enabled }}
           livenessProbe:
             tcpSocket:

src/main/charts/bitbucket/values.yaml

@@ -696,6 +696,10 @@ bitbucket:
   #
   startupProbe:
 
+    # -- Whether to apply the startupProbe check to pod.
+    #
+    enabled: true
+
     # -- How often (in seconds) the Bitbucket container startup probe will run
     #
     periodSeconds: 5

src/main/charts/confluence/README.md

@@ -83,6 +83,7 @@ Kubernetes: `>=1.21.x-0`
 | confluence.setPermissions | bool | `true` | Boolean to define whether to set local home directory permissions on startup of Confluence container. Set to 'false' to disable this behaviour.  |
 | confluence.shutdown.command | string | `"/shutdown-wait.sh"` | By default pods will be stopped via a [preStop hook](https://kubernetes.io/docs/concepts/containers/container-lifecycle-hooks/), using a script supplied by the Docker image. If any other shutdown behaviour is needed it can be achieved by overriding this value. Note that the shutdown command needs to wait for the application shutdown completely before exiting; see [the default command](https://bitbucket.org/atlassian-docker/docker-atlassian-confluence-server/src/master/shutdown-wait.sh) for details.  |
 | confluence.shutdown.terminationGracePeriodSeconds | int | `25` | The termination grace period for pods during shutdown. This should be set to the Confluence internal grace period (default 20 seconds), plus a small buffer to allow the JVM to fully terminate.  |
+| confluence.startupProbe.enabled | bool | `true` | Whether to apply the startupProbe check to pod.  |
 | confluence.startupProbe.failureThreshold | int | `120` | The number of consecutive failures of the Confluence container startup probe before the pod fails startup checks.  |
 | confluence.startupProbe.periodSeconds | int | `5` | How often (in seconds) the Confluence container startup probe will run  |
 | confluence.tomcatConfig | object | `{"acceptCount":"10","connectionTimeout":"20000","customServerXml":"","debug":"0","enableLookups":"false","generateByHelm":false,"maxHttpHeaderSize":"8192","maxThreads":"100","mgmtPort":"8000","minSpareThreads":"10","port":"8090","protocol":"org.apache.coyote.http11.Http11NioProtocol","proxyInternalIps":null,"proxyName":null,"proxyPort":null,"redirectPort":"8443","scheme":null,"secure":null,"uriEncoding":"UTF-8"}` | By default Tomcat's server.xml is generated in the container entrypoint from a template shipped with an official Confluence image. However, server.xml generation may fail if container is not run as root, which is a common case if Confluence is deployed to OpenShift.  |

src/main/charts/confluence/templates/statefulset.yaml

@@ -89,11 +89,13 @@ spec:
             failureThreshold: {{ .Values.confluence.readinessProbe.failureThreshold }}
             {{- end }}
           {{- end }}
+          {{- if .Values.confluence.startupProbe.enabled }}
           startupProbe:
             tcpSocket:
               port: {{ .Values.confluence.ports.http }}
             periodSeconds: {{ .Values.confluence.startupProbe.periodSeconds }}
             failureThreshold: {{ .Values.confluence.startupProbe.failureThreshold }}
+          {{- end }}
           {{- if .Values.confluence.livenessProbe.enabled }}
           livenessProbe:
             tcpSocket:

src/main/charts/confluence/values.yaml

@@ -628,6 +628,10 @@ confluence:
   #
   startupProbe:
 
+    # -- Whether to apply the startupProbe check to pod.
+    #
+    enabled: true
+
     # -- How often (in seconds) the Confluence container startup probe will run
     #
     periodSeconds: 5

src/main/charts/crowd/README.md

@@ -69,6 +69,7 @@ Kubernetes: `>=1.21.x-0`
 | crowd.setPermissions | bool | `true` | Boolean to define whether to set local home directory permissions on startup of Crowd container. Set to 'false' to disable this behaviour.  |
 | crowd.shutdown.command | string | `"/shutdown-wait.sh"` | By default pods will be stopped via a [preStop hook](https://kubernetes.io/docs/concepts/containers/container-lifecycle-hooks/), using a script supplied by the Docker image. If any other shutdown behaviour is needed it can be achieved by overriding this value. Note that the shutdown command needs to wait for the application shutdown completely before exiting; see [the default command](https://bitbucket.org/atlassian-docker/docker-atlassian-crowd/src/master/shutdown-wait.sh) for details.  |
 | crowd.shutdown.terminationGracePeriodSeconds | int | `30` | The termination grace period for pods during shutdown. This should be set to the internal grace period, plus a small buffer to allow the JVM to fully terminate.  |
+| crowd.startupProbe.enabled | bool | `true` | Whether to apply the startupProbe check to pod.  |
 | crowd.startupProbe.failureThreshold | int | `120` | The number of consecutive failures of the Crowd container startup probe before the pod fails startup checks.  |
 | crowd.startupProbe.periodSeconds | int | `5` | How often (in seconds) the Crowd container startup probe will run  |
 | crowd.topologySpreadConstraints | list | `[]` | Defines topology spread constraints for Crowd pods. See details: https://kubernetes.io/docs/concepts/workloads/pods/pod-topology-spread-constraints/  |

src/main/charts/crowd/templates/statefulset.yaml

@@ -86,11 +86,13 @@ spec:
             failureThreshold: {{ .Values.crowd.readinessProbe.failureThreshold }}
             {{- end }}
           {{- end }}
+          {{- if .Values.crowd.startupProbe.enabled }}
           startupProbe:
             tcpSocket:
               port: {{ .Values.crowd.ports.http }}
             periodSeconds: {{ .Values.crowd.startupProbe.periodSeconds }}
             failureThreshold: {{ .Values.crowd.startupProbe.failureThreshold }}
+          {{- end }}
           {{- if .Values.crowd.livenessProbe.enabled }}
           livenessProbe:
             tcpSocket:

src/main/charts/crowd/values.yaml

@@ -197,6 +197,10 @@ crowd:
 
   startupProbe:
 
+    # -- Whether to apply the startupProbe check to pod.
+    #
+    enabled: true
+
     # -- How often (in seconds) the Crowd container startup probe will run
     #
     periodSeconds: 5

src/main/charts/jira/README.md

@@ -111,6 +111,7 @@ Kubernetes: `>=1.21.x-0`
 | jira.setPermissions | bool | `true` | Boolean to define whether to set local home directory permissions on startup of Jira container. Set to 'false' to disable this behaviour.  |
 | jira.shutdown.command | string | `"/shutdown-wait.sh"` | By default pods will be stopped via a [preStop hook](https://kubernetes.io/docs/concepts/containers/container-lifecycle-hooks/), using a script supplied by the Docker image. If any other shutdown behaviour is needed it can be achieved by overriding this value. Note that the shutdown command needs to wait for the application shutdown completely before exiting; see [the default command](https://bitbucket.org/atlassian-docker/docker-atlassian-jira/src/master/shutdown-wait.sh) for details.  |
 | jira.shutdown.terminationGracePeriodSeconds | int | `30` | The termination grace period for pods during shutdown. This should be set to the internal grace period, plus a small buffer to allow the JVM to fully terminate.  |
+| jira.startupProbe.enabled | bool | `true` | Whether to apply the startupProbe check to pod.  |
 | jira.startupProbe.failureThreshold | int | `120` | The number of consecutive failures of the Jira container startup probe before the pod fails startup checks.  |
 | jira.startupProbe.periodSeconds | int | `5` | How often (in seconds) the Jira container startup probe will run  |
 | jira.tomcatConfig | object | `{"acceptCount":"10","connectionTimeout":"20000","customServerXml":"","enableLookups":"false","generateByHelm":false,"maxHttpHeaderSize":"8192","maxThreads":"100","mgmtPort":"8005","minSpareThreads":"10","port":"8080","protocol":"HTTP/1.1","proxyName":null,"proxyPort":null,"redirectPort":"8443","scheme":null,"secure":null}` | By default Tomcat's server.xml is generated in the container entrypoint from a template shipped with an official Jira image. However, server.xml generation may fail if container is not run as root, which is a common case if Jira is deployed to OpenShift.  |

src/main/charts/jira/templates/statefulset.yaml

@@ -139,11 +139,13 @@ spec:
             failureThreshold: {{ .Values.jira.readinessProbe.failureThreshold }}
             {{- end }}
           {{- end }}
+          {{- if .Values.jira.startupProbe.enabled }}
           startupProbe:
             tcpSocket:
               port: {{ .Values.jira.ports.http }}
             periodSeconds: {{ .Values.jira.startupProbe.periodSeconds }}
             failureThreshold: {{ .Values.jira.startupProbe.failureThreshold }}
+          {{- end }}
           {{- if .Values.jira.livenessProbe.enabled }}
           livenessProbe:
             tcpSocket:

src/main/charts/jira/values.yaml

@@ -516,6 +516,10 @@ jira:
   #
   startupProbe:
 
+    # -- Whether to apply the startupProbe check to pod.
+    #
+    enabled: true
+
     # -- How often (in seconds) the Jira container startup probe will run
     #
     periodSeconds: 5
@@ -848,7 +852,7 @@ monitoring:
     #
     dashboardLabels: {}
       # grafana_dashboard: dc_monitoring
-    
+
     # -- Annotations added to Grafana dashboards ConfigMaps. See: https://github.com/kiwigrid/k8s-sidecar#usage
     #
     dashboardAnnotations: {}

src/test/java/test/ReadinessLivenessProbesTest.java

@@ -119,4 +119,27 @@ void test_readiness_probe_custom_probe(Product product) throws Exception {
         assertEquals("bar", resources.getStatefulSet(
                 product.getHelmReleaseName()).getContainer().get("readinessProbe").get("foo").asText());
     }
+
+    @ParameterizedTest
+    @EnumSource(value = Product.class, names = {"bamboo_agent"}, mode = EnumSource.Mode.EXCLUDE)
+    void test_startup_probe_disabled(Product product) throws Exception {
+        final var resources = helm.captureKubeResourcesFromHelmChart(product, Map.of(
+                product + ".startupProbe.enabled", "false"));
+
+        assertThat(resources.getStatefulSet(
+                product.getHelmReleaseName()).getContainer().get("startupProbe")).isEmpty();
+    }
+
+    @ParameterizedTest
+    @EnumSource(value = Product.class, names = {"bamboo_agent"}, mode = EnumSource.Mode.EXCLUDE)
+    void test_startup_probe_customized(Product product) throws Exception {
+        final var resources = helm.captureKubeResourcesFromHelmChart(product, Map.of(
+                product + ".startupProbe.failureThreshold", "1200",
+                product + ".startupProbe.periodSeconds", "14"));
+
+        assertEquals("1200", resources.getStatefulSet(
+                product.getHelmReleaseName()).getContainer().get("startupProbe").get("failureThreshold").asText());
+        assertEquals("14", resources.getStatefulSet(
+                product.getHelmReleaseName()).getContainer().get("startupProbe").get("periodSeconds").asText());
+    }
 }