Set worker 0 completed if pod's phase goto succeeded (#1042)

kubeflow · Jul 16, 2019 · 4ee67e2 · 4ee67e2
1 parent 21b0f38
commit 4ee67e2
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 2 deletions.
diff --git a/pkg/controller.v1/tensorflow/pod.go b/pkg/controller.v1/tensorflow/pod.go
@@ -120,7 +120,8 @@ func (tc *TFController) reconcilePods(
 			}
 
 			// Check whether worker 0 is exited without error.
-			if rtype == tfv1.TFReplicaTypeWorker && index == 0 && exitCode == 0 {
+			if rtype == tfv1.TFReplicaTypeWorker && index == 0 &&
+				exitCode == 0 && pod.Status.Phase == v1.PodSucceeded {
 				worker0Completed = true
 			}
 			updateTFJobReplicaStatuses(tfjob, rtype, pod)

diff --git a/pkg/controller.v1beta2/tensorflow/pod.go b/pkg/controller.v1beta2/tensorflow/pod.go
@@ -120,7 +120,8 @@ func (tc *TFController) reconcilePods(
 			}
 
 			// Check whether worker 0 is exited without error.
-			if rtype == tfv1beta2.TFReplicaTypeWorker && index == 0 && exitCode == 0 {
+			if rtype == tfv1beta2.TFReplicaTypeWorker && index == 0 &&
+				exitCode == 0 && pod.Status.Phase == v1.PodSucceeded {
 				worker0Completed = true
 			}
 			updateTFJobReplicaStatuses(tfjob, rtype, pod)