[RayService] Trim Redis Cleanup job less than 63 chars (#2846)

* redis cleanup

* small comments change

* 1. trim labels value like here:  https://github.com/ray-project/kuberay/blob/master/ray-operator/controllers/ray/common/service.go#L27

2. interface function TrimJobName for clarity

3. Keep job for 3 minutes after completion for debugging

* Update ray-operator/controllers/ray/raycluster_controller.go

Co-authored-by: Rueian <[email protected]>
Signed-off-by: Aviad Shimoni <[email protected]>

* Update ray-operator/controllers/ray/raycluster_controller.go

Co-authored-by: Rueian <[email protected]>
Signed-off-by: Aviad Shimoni <[email protected]>

* remove empty line

---------

Signed-off-by: Aviad Shimoni <[email protected]>
Co-authored-by: Rueian <[email protected]>

Author: aviadshimoni

ray-operator/controllers/ray/raycluster_controller.go

@@ -1143,14 +1143,15 @@ func (r *RayClusterReconciler) buildWorkerPod(ctx context.Context, instance rayv
 func (r *RayClusterReconciler) buildRedisCleanupJob(ctx context.Context, instance rayv1.RayCluster) batchv1.Job {
 	logger := ctrl.LoggerFrom(ctx)
 
+	// Build the head pod
 	pod := r.buildHeadPod(ctx, instance)
 	pod.Labels[utils.RayNodeTypeLabelKey] = string(rayv1.RedisCleanupNode)
 
 	// Only keep the Ray container in the Redis cleanup Job.
 	pod.Spec.Containers = []corev1.Container{pod.Spec.Containers[utils.RayContainerIndex]}
 	pod.Spec.Containers[utils.RayContainerIndex].Command = []string{"/bin/bash", "-lc", "--"}
 	pod.Spec.Containers[utils.RayContainerIndex].Args = []string{
-		"echo \"To get more information about manually delete the storage namespace in Redis and remove the RayCluster's finalizer, please check https://docs.ray.io/en/master/cluster/kubernetes/user-guides/kuberay-gcs-ft.html for more details.\" && " +
+		"echo \"To get more information about manually deleting the storage namespace in Redis and removing the RayCluster's finalizer, please check https://docs.ray.io/en/master/cluster/kubernetes/user-guides/kuberay-gcs-ft.html for more details.\" && " +
 			"python -c " +
 			"\"from ray._private.gcs_utils import cleanup_redis_storage; " +
 			"from urllib.parse import urlparse; " +
@@ -1180,8 +1181,8 @@ func (r *RayClusterReconciler) buildRedisCleanupJob(ctx context.Context, instanc
 		Value: "500",
 	})
 
-	// The container's resource consumption remains constant. so hard-coding the resources is acceptable.
-	// In addition, avoid using the GPU for the Redis cleanup Job.
+	// The container's resource consumption remains constant. Hard-coding the resources is acceptable.
+	// Avoid using the GPU for the Redis cleanup Job.
 	pod.Spec.Containers[utils.RayContainerIndex].Resources = corev1.ResourceRequirements{
 		Limits: corev1.ResourceList{
 			corev1.ResourceCPU:    resource.MustParse("200m"),
@@ -1196,9 +1197,12 @@ func (r *RayClusterReconciler) buildRedisCleanupJob(ctx context.Context, instanc
 	// For Kubernetes Job, the valid values for Pod's `RestartPolicy` are `Never` and `OnFailure`.
 	pod.Spec.RestartPolicy = corev1.RestartPolicyNever
 
+	// Trim the job name to ensure it is within the 63-character limit.
+	jobName := utils.TrimJobName(fmt.Sprintf("%s-%s", instance.Name, "redis-cleanup"))
+
 	redisCleanupJob := batchv1.Job{
 		ObjectMeta: metav1.ObjectMeta{
-			Name:        fmt.Sprintf("%s-%s", instance.Name, "redis-cleanup"),
+			Name:        jobName,
 			Namespace:   instance.Namespace,
 			Labels:      pod.Labels,
 			Annotations: pod.Annotations,
@@ -1209,7 +1213,7 @@ func (r *RayClusterReconciler) buildRedisCleanupJob(ctx context.Context, instanc
 				ObjectMeta: pod.ObjectMeta,
 				Spec:       pod.Spec,
 			},
-			// make this job be best-effort only for 5 minutes.
+			// Make this job best-effort only for 5 minutes.
 			ActiveDeadlineSeconds: ptr.To[int64](300),
 		},
 	}

ray-operator/controllers/ray/utils/util.go

@@ -205,6 +205,11 @@ func CheckName(s string) string {
 	return s
 }
 
+// TrimJobName uses CheckLabel to trim Kubernetes job to constrains
+func TrimJobName(jobName string) string {
+	return CheckLabel(jobName)
+}
+
 // CheckLabel makes sure the label value does not start with a punctuation and the total length is < 63 char
 func CheckLabel(s string) string {
 	maxLenght := 63

ray-operator/test/e2e/raycluster_gcsft_test.go

@@ -89,6 +89,20 @@ func TestGcsFaultToleranceOptions(t *testing.T) {
 			},
 			createSecret: true,
 		},
+		{
+			name:          "Long RayCluster Name",
+			redisPassword: "",
+			rayClusterFn: func(namespace string) *rayv1ac.RayClusterApplyConfiguration {
+				// Intentionally using a long name to test job name trimming
+				return rayv1ac.RayCluster("raycluster-with-a-very-long-name-exceeding-k8s-limit", namespace).WithSpec(
+					newRayClusterSpec().WithGcsFaultToleranceOptions(
+						rayv1ac.GcsFaultToleranceOptions().
+							WithRedisAddress("redis:6379"),
+					),
+				)
+			},
+			createSecret: false,
+		},
 	}
 
 	for _, tc := range testCases {