Kill infra pod when a container is restarted.

Revert 0e20f7d736.
Fixes #4759.

pkg/kubelet/kubelet.go

@@ -1081,7 +1081,6 @@ func (kl *Kubelet) syncPod(pod *api.BoundPod, dockerContainers dockertools.Docke
 		glog.Errorf("Unable to get pod with name %q and uid %q info with error(%v)", podFullName, uid, err)
 	}
 
-	podChanged := false
 	for _, container := range pod.Spec.Containers {
 		expectedHash := dockertools.HashContainer(&container)
 		dockerContainerName := dockertools.BuildDockerName(uid, podFullName, &container)
@@ -1090,8 +1089,8 @@ func (kl *Kubelet) syncPod(pod *api.BoundPod, dockerContainers dockertools.Docke
 			glog.V(3).Infof("pod %q container %q exists as %v", podFullName, container.Name, containerID)
 
 			// look for changes in the container.
-			containerChanged := hash != 0 && hash != expectedHash
+			podChanged := hash != 0 && hash != expectedHash
-			if !containerChanged {
+			if !podChanged {
 				// TODO: This should probably be separated out into a separate goroutine.
 				// If the container's liveness probe is unsuccessful, set readiness to false. If liveness is succesful, do a readiness check and set
 				// readiness accordingly. If the initalDelay since container creation on liveness probe has not passed the probe will return Success.
@@ -1123,7 +1122,6 @@ func (kl *Kubelet) syncPod(pod *api.BoundPod, dockerContainers dockertools.Docke
 				}
 				glog.Infof("pod %q container %q is unhealthy. Container will be killed and re-created.", podFullName, container.Name, live)
 			} else {
-				podChanged = true
 				glog.Infof("pod %q container %q hash changed (%d vs %d). Container will be killed and re-created.", podFullName, container.Name, hash, expectedHash)
 			}
 			if err := kl.killContainer(dockerContainer); err != nil {
@@ -1131,6 +1129,16 @@ func (kl *Kubelet) syncPod(pod *api.BoundPod, dockerContainers dockertools.Docke
 				continue
 			}
 			killedContainers[containerID] = empty{}
+
+			if podChanged {
+				// Also kill associated pod infra container if the pod changed.
+				if podInfraContainer, found, _ := dockerContainers.FindPodContainer(podFullName, uid, dockertools.PodInfraContainerName); found {
+					if err := kl.killContainer(podInfraContainer); err != nil {
+						glog.V(1).Infof("Failed to kill pod infra container %q: %v", podInfraContainer.ID, err)
+						continue
+					}
+				}
+			}
 		}
 
 		// Check RestartPolicy for container
@@ -1193,20 +1201,6 @@ func (kl *Kubelet) syncPod(pod *api.BoundPod, dockerContainers dockertools.Docke
 		containersToKeep[containerID] = empty{}
 	}
 
-	if podChanged {
-		// Also kill associated pod infra container if the pod changed.
-		if podInfraContainer, found, _ := dockerContainers.FindPodContainer(podFullName, uid, dockertools.PodInfraContainerName); found {
-			if err := kl.killContainer(podInfraContainer); err != nil {
-				glog.V(1).Infof("Failed to kill pod infra container %q: %v", podInfraContainer.ID, err)
-			}
-		}
-		podInfraContainerID, err = kl.createPodInfraContainer(pod)
-		if err != nil {
-			glog.Errorf("Failed to recreate pod infra container: %v for pod %q", err, podFullName)
-		}
-		containersToKeep[podInfraContainerID] = empty{}
-	}
-
 	// Kill any containers in this pod which were not identified above (guards against duplicates).
 	for id, container := range dockerContainers {
 		curPodFullName, curUUID, _, _ := dockertools.ParseDockerName(container.Names[0])

pkg/kubelet/kubelet_test.go

@@ -934,7 +934,8 @@ func TestSyncPodBadHash(t *testing.T) {
 		t.Errorf("unexpected error: %v", err)
 	}
 
-	verifyCalls(t, fakeDocker, []string{"list", "stop", "list", "create", "start", "stop", "create", "start", "inspect_container"})
+	//verifyCalls(t, fakeDocker, []string{"list", "stop", "list", "create", "start", "stop", "create", "start", "inspect_container"})
+	verifyCalls(t, fakeDocker, []string{"list", "stop", "stop", "list", "create", "start"})
 
 	// A map interation is used to delete containers, so must not depend on
 	// order here.