diff --git a/pkg/controller/mpi_job_controller.go b/pkg/controller/mpi_job_controller.go
index 02026944..042ed53e 100644
--- a/pkg/controller/mpi_job_controller.go
+++ b/pkg/controller/mpi_job_controller.go
@@ -111,10 +111,6 @@ const (
 	// From: k8s.io/kubernetes/pkg/apis/core/validation/events.go
 	eventMessageLimit = 1024
 
-	// jobBackoffLimitExceededReason is the reason that the k8s job controller
-	// uses when the backoff limit is exceeded.
-	jobBackoffLimitExceededReason = "BackoffLimitExceeded"
-
 	openMPISlotsEnv  = "OMPI_MCA_orte_set_default_slots"
 	intelMPISlotsEnv = "I_MPI_PERHOST"
 )
@@ -1149,7 +1145,7 @@ func (c *MPIJobController) updateMPIJobFailedStatus(mpiJob *kubeflow.MPIJob, lau
 	if msg == "" {
 		msg = fmt.Sprintf("MPIJob %s/%s has failed", mpiJob.Namespace, mpiJob.Name)
 	}
-	if reason == jobBackoffLimitExceededReason {
+	if reason == batchv1.JobReasonBackoffLimitExceeded {
 		// Concatenate the reason and message from the last failed Pod.
 		var lastFailedPod *corev1.Pod
 		for _, p := range launcherPods {
diff --git a/pkg/controller/mpi_job_controller_test.go b/pkg/controller/mpi_job_controller_test.go
index 3594d354..cf51131f 100644
--- a/pkg/controller/mpi_job_controller_test.go
+++ b/pkg/controller/mpi_job_controller_test.go
@@ -635,7 +635,7 @@ func TestLauncherFailed(t *testing.T) {
 	launcher.Status.Conditions = append(launcher.Status.Conditions, batchv1.JobCondition{
 		Type:    batchv1.JobFailed,
 		Status:  corev1.ConditionTrue,
-		Reason:  jobBackoffLimitExceededReason,
+		Reason:  batchv1.JobReasonBackoffLimitExceeded,
 		Message: "Job has reached the specified backoff limit",
 	})
 	launcher.Status.Failed = 2
@@ -668,7 +668,7 @@ func TestLauncherFailed(t *testing.T) {
 	msg := fmt.Sprintf("MPIJob %s/%s is created.", mpiJob.Namespace, mpiJob.Name)
 	updateMPIJobConditions(mpiJobCopy, kubeflow.JobCreated, corev1.ConditionTrue, mpiJobCreatedReason, msg)
 	msg = "Job has reached the specified backoff limit: second message"
-	updateMPIJobConditions(mpiJobCopy, kubeflow.JobFailed, corev1.ConditionTrue, jobBackoffLimitExceededReason+"/FailedReason2", msg)
+	updateMPIJobConditions(mpiJobCopy, kubeflow.JobFailed, corev1.ConditionTrue, batchv1.JobReasonBackoffLimitExceeded+"/FailedReason2", msg)
 
 	f.expectUpdateMPIJobStatusAction(mpiJobCopy)