Reconcile NoExecute Taint

Author: kaiyuechen

pkg/controller/nodelifecycle/node_lifecycle_controller.go

@@ -872,7 +872,7 @@ func (nc *Controller) processTaintBaseEviction(node *v1.Node, observedReadyCondi
 			if !nodeutil.SwapNodeControllerTaint(nc.kubeClient, []*v1.Taint{&taintToAdd}, []*v1.Taint{UnreachableTaintTemplate}, node) {
 				klog.Errorf("Failed to instantly swap UnreachableTaint to NotReadyTaint. Will try again in the next cycle.")
 			}
-		} else if nc.markNodeForTainting(node) {
+		} else if nc.markNodeForTainting(node, v1.ConditionFalse) {
 			klog.V(2).Infof("Node %v is NotReady as of %v. Adding it to the Taint queue.",
 				node.Name,
 				decisionTimestamp,
@@ -885,7 +885,7 @@ func (nc *Controller) processTaintBaseEviction(node *v1.Node, observedReadyCondi
 			if !nodeutil.SwapNodeControllerTaint(nc.kubeClient, []*v1.Taint{&taintToAdd}, []*v1.Taint{NotReadyTaintTemplate}, node) {
 				klog.Errorf("Failed to instantly swap NotReadyTaint to UnreachableTaint. Will try again in the next cycle.")
 			}
-		} else if nc.markNodeForTainting(node) {
+		} else if nc.markNodeForTainting(node, v1.ConditionUnknown) {
 			klog.V(2).Infof("Node %v is unresponsive as of %v. Adding it to the Taint queue.",
 				node.Name,
 				decisionTimestamp,
@@ -1476,9 +1476,21 @@ func (nc *Controller) evictPods(node *v1.Node, pods []*v1.Pod) (bool, error) {
 	return nc.zonePodEvictor[utilnode.GetZoneKey(node)].Add(node.Name, string(node.UID)), nil
 }
 
-func (nc *Controller) markNodeForTainting(node *v1.Node) bool {
+func (nc *Controller) markNodeForTainting(node *v1.Node, status v1.ConditionStatus) bool {
 	nc.evictorLock.Lock()
 	defer nc.evictorLock.Unlock()
+	if status == v1.ConditionFalse {
+		if !taintutils.TaintExists(node.Spec.Taints, NotReadyTaintTemplate) {
+			nc.zoneNoExecuteTainter[utilnode.GetZoneKey(node)].SetRemove(node.Name)
+		}
+	}
+
+	if status == v1.ConditionUnknown {
+		if !taintutils.TaintExists(node.Spec.Taints, UnreachableTaintTemplate) {
+			nc.zoneNoExecuteTainter[utilnode.GetZoneKey(node)].SetRemove(node.Name)
+		}
+	}
+
 	return nc.zoneNoExecuteTainter[utilnode.GetZoneKey(node)].Add(node.Name, string(node.UID))
 }
 

pkg/controller/nodelifecycle/scheduler/rate_limited_queue.go

@@ -194,6 +194,15 @@ func (q *UniqueQueue) Clear() {
 	}
 }
 
+// SetRemove remove value from the set if value existed
+func (q *UniqueQueue) SetRemove(value string) {
+	q.lock.Lock()
+	defer q.lock.Unlock()
+	if q.set.Has(value) {
+		q.set.Delete(value)
+	}
+}
+
 // RateLimitedTimedQueue is a unique item priority queue ordered by
 // the expected next time of execution. It is also rate limited.
 type RateLimitedTimedQueue struct {
@@ -280,6 +289,11 @@ func (q *RateLimitedTimedQueue) Clear() {
 	q.queue.Clear()
 }
 
+// SetRemove remove value from the set of the queue
+func (q *RateLimitedTimedQueue) SetRemove(value string) {
+	q.queue.SetRemove(value)
+}
+
 // SwapLimiter safely swaps current limiter for this queue with the
 // passed one if capacities or qps's differ.
 func (q *RateLimitedTimedQueue) SwapLimiter(newQPS float32) {

pkg/controller/nodelifecycle/scheduler/rate_limited_queue_test.go

@@ -282,6 +282,17 @@ func TestClear(t *testing.T) {
 	}
 }
 
+func TestSetRemove(t *testing.T) {
+	evictor := NewRateLimitedTimedQueue(flowcontrol.NewFakeAlwaysRateLimiter())
+	evictor.Add("first", "11111")
+
+	evictor.SetRemove("first")
+
+	if evictor.queue.set.Len() != 0 {
+		t.Fatalf("SetRemove should remove element from the set.")
+	}
+}
+
 func TestSwapLimiter(t *testing.T) {
 	evictor := NewRateLimitedTimedQueue(flowcontrol.NewFakeAlwaysRateLimiter())
 	fakeAlways := flowcontrol.NewFakeAlwaysRateLimiter()