From e2b78ddadcb74eee16a262d64fbd073084337025 Mon Sep 17 00:00:00 2001
From: gmarek <gmarek@google.com>
Date: Fri, 14 Oct 2016 12:38:39 +0200
Subject: [PATCH] NodeController waits for informer sync before doing anything

---
 pkg/controller/node/nodecontroller.go | 25 ++++++++++++++++++++++---
 1 file changed, 22 insertions(+), 3 deletions(-)

diff --git a/pkg/controller/node/nodecontroller.go b/pkg/controller/node/nodecontroller.go
index da61e24773a..ba0700f27c8 100644
--- a/pkg/controller/node/nodecontroller.go
+++ b/pkg/controller/node/nodecontroller.go
@@ -131,9 +131,13 @@ type NodeController struct {
 	// The maximum duration before a pod evicted from a node can be forcefully terminated.
 	maximumGracePeriod time.Duration
 	recorder           record.EventRecorder
-	podStore           cache.StoreToPodLister
-	nodeStore          cache.StoreToNodeLister
-	daemonSetStore     cache.StoreToDaemonSetLister
+	podInformer        informers.PodInformer
+	nodeInformer       informers.NodeInformer
+	daemonSetInformer  informers.DaemonSetInformer
+
+	podStore       cache.StoreToPodLister
+	nodeStore      cache.StoreToNodeLister
+	daemonSetStore cache.StoreToDaemonSetLister
 	// allocate/recycle CIDRs for node if allocateNodeCIDRs == true
 	cidrAllocator CIDRAllocator
 
@@ -228,6 +232,9 @@ func NewNodeController(
 		largeClusterThreshold:       largeClusterThreshold,
 		unhealthyZoneThreshold:      unhealthyZoneThreshold,
 		zoneStates:                  make(map[string]zoneState),
+		podInformer:                 podInformer,
+		nodeInformer:                nodeInformer,
+		daemonSetInformer:           daemonSetInformer,
 	}
 	nc.enterPartialDisruptionFunc = nc.ReducedQPSFunc
 	nc.enterFullDisruptionFunc = nc.HealthyQPSFunc
@@ -351,6 +358,10 @@ func NewNodeController(
 func (nc *NodeController) Run() {
 	// Incorporate the results of node status pushed from kubelet to master.
 	go wait.Until(func() {
+		if !cache.WaitForCacheSync(wait.NeverStop, nc.nodeInformer.Informer().HasSynced, nc.podInformer.Informer().HasSynced, nc.daemonSetInformer.Informer().HasSynced) {
+			glog.Errorf("NodeController timed out while waiting for informers to sync...")
+			return
+		}
 		if err := nc.monitorNodeStatus(); err != nil {
 			glog.Errorf("Error monitoring node status: %v", err)
 		}
@@ -369,6 +380,10 @@ func (nc *NodeController) Run() {
 	//    c. If there are pods still terminating, wait for their estimated completion
 	//       before retrying
 	go wait.Until(func() {
+		if !cache.WaitForCacheSync(wait.NeverStop, nc.nodeInformer.Informer().HasSynced, nc.podInformer.Informer().HasSynced, nc.daemonSetInformer.Informer().HasSynced) {
+			glog.Errorf("NodeController timed out while waiting for informers to sync...")
+			return
+		}
 		nc.evictorLock.Lock()
 		defer nc.evictorLock.Unlock()
 		for k := range nc.zonePodEvictor {
@@ -402,6 +417,10 @@ func (nc *NodeController) Run() {
 	// TODO: replace with a controller that ensures pods that are terminating complete
 	// in a particular time period
 	go wait.Until(func() {
+		if !cache.WaitForCacheSync(wait.NeverStop, nc.nodeInformer.Informer().HasSynced, nc.podInformer.Informer().HasSynced, nc.daemonSetInformer.Informer().HasSynced) {
+			glog.Errorf("NodeController timed out while waiting for informers to sync...")
+			return
+		}
 		nc.evictorLock.Lock()
 		defer nc.evictorLock.Unlock()
 		for k := range nc.zoneTerminationEvictor {