[release-0.8] fix: handle nil Completions in coordinator validation (#935)

* fix: handle nil Completions in coordinator validation

* fixup! fix: handle nil Completions in coordinator validation

* fixup! fix: handle nil Completions in coordinator validation

Author: Ladicle

pkg/webhooks/jobset_webhook.go

@@ -375,9 +375,9 @@ func validateCoordinator(js *jobset.JobSet) error {
 		return fmt.Errorf("coordinator job index %d is invalid for replicatedJob %s", js.Spec.Coordinator.JobIndex, replicatedJob.Name)
 	}
 
-	// Validate job is using indexed completion mode.
-	if replicatedJob.Template.Spec.CompletionMode == nil || *replicatedJob.Template.Spec.CompletionMode != batchv1.IndexedCompletion {
-		return fmt.Errorf("job for coordinator pod must be indexed completion mode")
+	// Validate job is using indexed completion mode and completions number is set.
+	if replicatedJob.Template.Spec.CompletionMode == nil || replicatedJob.Template.Spec.Completions == nil || *replicatedJob.Template.Spec.CompletionMode != batchv1.IndexedCompletion {
+		return fmt.Errorf("job for coordinator pod must be indexed completion mode, and completions number must be set")
 	}
 
 	// Validate Pod index.

pkg/webhooks/jobset_webhook_test.go

@@ -1403,6 +1403,46 @@ func TestValidateCreate(t *testing.T) {
 				fmt.Errorf("coordinator replicatedJob fake-rjob does not exist"),
 			),
 		},
+		{
+			name: "coordinator replicated job missing completions",
+			js: &jobset.JobSet{
+				ObjectMeta: validObjectMeta,
+				Spec: jobset.JobSetSpec{
+					Coordinator: &jobset.Coordinator{
+						ReplicatedJob: "replicatedjob-a",
+						JobIndex:      0,
+						PodIndex:      0,
+					},
+					ReplicatedJobs: []jobset.ReplicatedJob{
+						{
+							Name:     "replicatedjob-a",
+							Replicas: 2,
+							Template: batchv1.JobTemplateSpec{
+								Spec: batchv1.JobSpec{
+									CompletionMode: ptr.To(batchv1.IndexedCompletion),
+									Parallelism:    ptr.To(int32(2)),
+								},
+							},
+						},
+						{
+							Name:     "replicatedjob-b",
+							Replicas: 2,
+							Template: batchv1.JobTemplateSpec{
+								Spec: batchv1.JobSpec{
+									CompletionMode: ptr.To(batchv1.IndexedCompletion),
+									Completions:    ptr.To(int32(2)),
+									Parallelism:    ptr.To(int32(2)),
+								},
+							},
+						},
+					},
+					SuccessPolicy: &jobset.SuccessPolicy{},
+				},
+			},
+			want: errors.Join(
+				fmt.Errorf("job for coordinator pod must be indexed completion mode, and completions number must be set"),
+			),
+		},
 		{
 			name: "coordinator job index invalid",
 			js: &jobset.JobSet{