Merge pull request #36592 from andrewsykim/36273-set-all-node-conditions-unknown-when-node-unreachable

Automatic merge from submit-queue (batch tested with PRs 40917, 41181, 41123, 36592, 41183)

Set all node conditions to Unknown when node is unreachable

**What this PR does / why we need it**:
Sets all node conditions to Unknown when node does not report status/unreachable

**Which issue this PR fixes** 
fixes https://github.com/kubernetes/kubernetes/issues/36273

pkg/controller/node/nodecontroller.go

@@ -742,35 +742,37 @@ func (nc *NodeController) tryUpdateNodeStatus(node *v1.Node) (time.Duration, v1.
 			if observedReadyCondition.Status != v1.ConditionUnknown {
 				currentReadyCondition.Status = v1.ConditionUnknown
 				currentReadyCondition.Reason = "NodeStatusUnknown"
-				currentReadyCondition.Message = fmt.Sprintf("Kubelet stopped posting node status.")
+				currentReadyCondition.Message = "Kubelet stopped posting node status."
 				// LastProbeTime is the last time we heard from kubelet.
 				currentReadyCondition.LastHeartbeatTime = observedReadyCondition.LastHeartbeatTime
 				currentReadyCondition.LastTransitionTime = nc.now()
 			}
 		}
 
-		// Like NodeReady condition, NodeOutOfDisk was last set longer ago than gracePeriod, so update
+		// remaining node conditions should also be set to Unknown
-		// it to Unknown (regardless of its current value) in the master.
+		remainingNodeConditionTypes := []v1.NodeConditionType{v1.NodeOutOfDisk, v1.NodeMemoryPressure, v1.NodeDiskPressure}
-		// TODO(madhusudancs): Refactor this with readyCondition to remove duplicated code.
+		nowTimestamp := nc.now()
-		_, oodCondition := v1.GetNodeCondition(&node.Status, v1.NodeOutOfDisk)
+		for _, nodeConditionType := range remainingNodeConditionTypes {
-		if oodCondition == nil {
+			_, currentCondition := v1.GetNodeCondition(&node.Status, nodeConditionType)
-			glog.V(2).Infof("Out of disk condition of node %v is never updated by kubelet", node.Name)
+			if currentCondition == nil {
+				glog.V(2).Infof("Condition %v of node %v was never updated by kubelet", nodeConditionType, node.Name)
 				node.Status.Conditions = append(node.Status.Conditions, v1.NodeCondition{
-				Type:               v1.NodeOutOfDisk,
+					Type:               nodeConditionType,
 					Status:             v1.ConditionUnknown,
 					Reason:             "NodeStatusNeverUpdated",
-				Message:            fmt.Sprintf("Kubelet never posted node status."),
+					Message:            "Kubelet never posted node status.",
 					LastHeartbeatTime:  node.CreationTimestamp,
-				LastTransitionTime: nc.now(),
+					LastTransitionTime: nowTimestamp,
 				})
 			} else {
-			glog.V(4).Infof("node %v hasn't been updated for %+v. Last out of disk condition is: %+v",
+				glog.V(4).Infof("node %v hasn't been updated for %+v. Last %v is: %+v",
-				node.Name, nc.now().Time.Sub(savedNodeStatus.probeTimestamp.Time), oodCondition)
+					node.Name, nc.now().Time.Sub(savedNodeStatus.probeTimestamp.Time), nodeConditionType, currentCondition)
-			if oodCondition.Status != v1.ConditionUnknown {
+				if currentCondition.Status != v1.ConditionUnknown {
-				oodCondition.Status = v1.ConditionUnknown
+					currentCondition.Status = v1.ConditionUnknown
-				oodCondition.Reason = "NodeStatusUnknown"
+					currentCondition.Reason = "NodeStatusUnknown"
-				oodCondition.Message = fmt.Sprintf("Kubelet stopped posting node status.")
+					currentCondition.Message = "Kubelet stopped posting node status."
-				oodCondition.LastTransitionTime = nc.now()
+					currentCondition.LastTransitionTime = nowTimestamp
+				}
 			}
 		}
 

pkg/controller/node/nodecontroller_test.go

@@ -1374,6 +1374,22 @@ func TestMonitorNodeStatusUpdateStatus(t *testing.T) {
 								LastHeartbeatTime:  metav1.Date(2012, 1, 1, 0, 0, 0, 0, time.UTC),
 								LastTransitionTime: fakeNow,
 							},
+							{
+								Type:               v1.NodeMemoryPressure,
+								Status:             v1.ConditionUnknown,
+								Reason:             "NodeStatusNeverUpdated",
+								Message:            "Kubelet never posted node status.",
+								LastHeartbeatTime:  metav1.Date(2012, 1, 1, 0, 0, 0, 0, time.UTC),
+								LastTransitionTime: fakeNow,
+							},
+							{
+								Type:               v1.NodeDiskPressure,
+								Status:             v1.ConditionUnknown,
+								Reason:             "NodeStatusNeverUpdated",
+								Message:            "Kubelet never posted node status.",
+								LastHeartbeatTime:  metav1.Date(2012, 1, 1, 0, 0, 0, 0, time.UTC),
+								LastTransitionTime: fakeNow,
+							},
 						},
 					},
 				},
@@ -1483,6 +1499,22 @@ func TestMonitorNodeStatusUpdateStatus(t *testing.T) {
 								LastHeartbeatTime:  metav1.Date(2015, 1, 1, 12, 0, 0, 0, time.UTC),
 								LastTransitionTime: metav1.Time{Time: metav1.Date(2015, 1, 1, 12, 0, 0, 0, time.UTC).Add(time.Hour)},
 							},
+							{
+								Type:               v1.NodeMemoryPressure,
+								Status:             v1.ConditionUnknown,
+								Reason:             "NodeStatusNeverUpdated",
+								Message:            "Kubelet never posted node status.",
+								LastHeartbeatTime:  metav1.Date(2012, 1, 1, 0, 0, 0, 0, time.UTC), // should default to node creation time if condition was never updated
+								LastTransitionTime: metav1.Time{Time: metav1.Date(2015, 1, 1, 12, 0, 0, 0, time.UTC).Add(time.Hour)},
+							},
+							{
+								Type:               v1.NodeDiskPressure,
+								Status:             v1.ConditionUnknown,
+								Reason:             "NodeStatusNeverUpdated",
+								Message:            "Kubelet never posted node status.",
+								LastHeartbeatTime:  metav1.Date(2012, 1, 1, 0, 0, 0, 0, time.UTC), // should default to node creation time if condition was never updated
+								LastTransitionTime: metav1.Time{Time: metav1.Date(2015, 1, 1, 12, 0, 0, 0, time.UTC).Add(time.Hour)},
+							},
 						},
 						Capacity: v1.ResourceList{
 							v1.ResourceName(v1.ResourceCPU):    resource.MustParse("10"),