docs: Document ttlSecondsAfterFinished (#126) (#147)

* docs: Document ttlSecondsAfterFinished (#126) Signed-off-by: Paolo Chila <[email protected]> * fixup! docs: Document ttlSecondsAfterFinished (#126)
keptn-contrib · Jan 11, 2022 · d93645c · d93645c
1 parent 022cdfe
commit d93645c
Show file tree

Hide file tree

Showing 3 changed files with 152 additions and 0 deletions.
diff --git a/README.md b/README.md
@@ -22,6 +22,7 @@
         - [Send start/finished event if the job config.yaml can't be found](#send-startfinished-event-if-the-job-configyaml-cant-be-found)
         - [Additional Event Data](#additional-event-data)
         - [Remote Control Plane](#remote-control-plane)
+        - [Job clean-up](#job-clean-up)
     - [How to validate a job configuration](#how-to-validate-a-job-configuration)
     - [Endless Possibilities](#endless-possibilities)
     - [Credits](#credits)
@@ -567,6 +568,38 @@ If you are using the service in a remote control plane setup make sure the distr
 events used in the `job/config.yaml`. Just edit the `PUBSUB_TOPIC` environment variable in the distributor deployment
 configuration to fit your needs.
 
+### Job clean-up
+
+Jobs objects are kept in kubernetes after completion to allow checking for status or logs inspections/retrieval.
+This is not always desirable so kubernetes allows for [automatic clean-up of finished jobs](https://kubernetes.io/docs/concepts/workloads/controllers/ttlafterfinished/)
+using `ttlSecondsAfterFinished` property in the job spec.
+
+Jobs created by the executor service will still be available for a time after (successful or failed) completion.
+The default value of the time-to-live (TTL) for completed jobs is `21600` seconds (6 hours).
+
+In order to set a different TTL for jobs add the `ttlSecondsAfterFinished` property in the task definition, for example:
+
+```yaml
+tasks:
+  - name: "Run locust tests"
+    files:
+      - locust/basic.py
+      - locust/import.py
+      - locust/locust.conf
+    image: "locustio/locust"
+    cmd:
+      - locust
+    args:
+      - '--config'
+      - /keptn/locust/locust.conf
+      - '-f'
+      - /keptn/locust/basic.py
+      - '--host'
+      - $(HOST)
+    # the corresponding job for this task will be cleaned up 10 minutes (600 seconds) after completion
+    ttlSecondsAfterFinished: 600
+```
+
 ## How to validate a job configuration
 
 `job-lint` is a simple cli tool that validates any given job configuration file and shows possible errors.

diff --git a/pkg/config/config_test.go b/pkg/config/config_test.go
@@ -346,4 +346,50 @@ actions:
 	assert.Check(t, cmp.Equal(config.Actions[0].Tasks[1].ImagePullPolicy, "Always"))
 	assert.Check(t, cmp.Equal(config.Actions[0].Tasks[2].ImagePullPolicy, "Never"))
 	assert.Check(t, cmp.Equal(config.Actions[0].Tasks[3].ImagePullPolicy, "IfNotPresent"))
+}
+
+func TestTTLSecondsAfterFinished(t *testing.T) {
+	configYaml := `
+apiVersion: v2
+actions:
+  - name: "Run some fancy job with various ttl after job completes"
+    events:
+      - name: "sh.keptn.event.test.triggered"
+        jsonpath:
+          property: "$.test.teststrategy"
+          match: "health"
+    tasks:
+      - name: "task1-noinputttl"
+        workingDir: "/bin"
+        image: "somefancyimage"
+        cmd:
+          - cmd
+        args:
+          - arg1
+          - arg2
+      - name: "task2-ttl10mins"
+        workingDir: "/bin"
+        image: "somefancyimage"
+        cmd:
+          - cmd
+        ttlSecondsAfterFinished: 600
+      - name: "task3-zerosecondsttl"
+        workingDir: "/bin"
+        image: "somefancyimage"
+        cmd:
+          - cmd
+        ttlSecondsAfterFinished: 0
+    `
+
+	config, err := NewConfig([]byte(configYaml))
+
+	assert.NilError(t, err)
+
+	assert.Equal(t, len(config.Actions), 1)
+	assert.Equal(t, len(config.Actions[0].Tasks), 3)
+	assert.Assert(t, config.Actions[0].Tasks[0].TTLSecondsAfterFinished == nil)
+	assert.Assert(t, config.Actions[0].Tasks[1].TTLSecondsAfterFinished != nil)
+	assert.Check(t, cmp.Equal(*config.Actions[0].Tasks[1].TTLSecondsAfterFinished, int32(600)))
+	assert.Assert(t, config.Actions[0].Tasks[2].TTLSecondsAfterFinished != nil)
+	assert.Check(t, cmp.Equal(*config.Actions[0].Tasks[2].TTLSecondsAfterFinished, int32(0)))
 }
diff --git a/pkg/k8sutils/job_test.go b/pkg/k8sutils/job_test.go
@@ -520,6 +520,79 @@ func TestImagePullPolicy(t *testing.T) {
 
 }
 
+func TestTTLSecondsAfterFinished(t *testing.T) {
+
+	var DefaultTTLAfterFinished int32 = 21600
+	var TenMinutesTTLAfterFinished int32 = 600
+	var ImmediatedlyDeletableTTLAfterFinished int32 = 0
+
+	tests := []struct {
+		name string
+		ttlSecondsAfterFinished *int32
+		expectedTTLSecondsAfterFinished int32
+	}{
+		{
+			name:               "No ttl specified in input - we should have the default job executor of 21600",
+			ttlSecondsAfterFinished: nil,
+			expectedTTLSecondsAfterFinished: DefaultTTLAfterFinished,
+		},
+		{
+			name:               "10 mins ttl specified in input",
+			ttlSecondsAfterFinished: &TenMinutesTTLAfterFinished,
+			expectedTTLSecondsAfterFinished: TenMinutesTTLAfterFinished,
+		},
+		{
+			name:               "0 seconds ttl specified in input - job eligible for deletion immediately",
+			ttlSecondsAfterFinished: &ImmediatedlyDeletableTTLAfterFinished,
+			expectedTTLSecondsAfterFinished: ImmediatedlyDeletableTTLAfterFinished,
+		},
+	}
+	for i, test := range tests {
+		t.Run(
+			test.name, func(t *testing.T) {
+
+				k8sClientSet := k8sfake.NewSimpleClientset()
+				k8s := k8sImpl{clientset: k8sClientSet}
+
+				jobName:= fmt.Sprintf("ipp-job-%d", i)
+
+				task := config.Task{
+					Name: fmt.Sprintf("noIppTask-%d", i),
+					Image: "someImage:someversion",
+					Cmd: []string{"someCmd"},
+					TTLSecondsAfterFinished: test.ttlSecondsAfterFinished,
+				}
+
+				eventData := keptnv2.EventData{
+					Project: "keptnproject",
+					Stage:   "dev",
+					Service: "keptnservice",
+				}
+
+				namespace := "test-namespace"
+
+				err := k8s.CreateK8sJob(
+					jobName, &config.Action{Name: jobName}, task, &eventData, JobSettings{
+						JobNamespace: namespace,
+						DefaultResourceRequirements: &corev1.ResourceRequirements{
+							Limits:   make(corev1.ResourceList),
+							Requests: make(corev1.ResourceList),
+						},
+					}, "", namespace,
+				)
+				assert.NilError(t, err)
+
+				job, err := k8sClientSet.BatchV1().Jobs(namespace).Get(context.TODO(), jobName, metav1.GetOptions{})
+				assert.NilError(t, err)
+
+				assert.Assert(t, job.Spec.TTLSecondsAfterFinished != nil)
+				assert.Equal(t, *job.Spec.TTLSecondsAfterFinished, test.expectedTTLSecondsAfterFinished)
+			},
+		)
+	}
+
+}
+
 func createK8sSecretObj(name string, namespace string, data map[string][]byte) *corev1.Secret {
 	return &corev1.Secret{
 		TypeMeta: metav1.TypeMeta{