vendor update for CSI 0.3.0

vendor/k8s.io/kubernetes/pkg/controller/nodeipam/ipam/BUILD

@@ -9,6 +9,7 @@ load(
 go_test(
     name = "go_default_test",
     srcs = [
+        "cloud_cidr_allocator_test.go",
         "controller_test.go",
         "range_allocator_test.go",
         "timeout_test.go",
@@ -48,7 +49,9 @@ go_library(
         "//pkg/controller/nodeipam/ipam/cidrset:go_default_library",
         "//pkg/controller/nodeipam/ipam/sync:go_default_library",
         "//pkg/controller/util/node:go_default_library",
+        "//pkg/scheduler/algorithm:go_default_library",
         "//pkg/util/node:go_default_library",
+        "//pkg/util/taints:go_default_library",
         "//vendor/github.com/golang/glog:go_default_library",
         "//vendor/k8s.io/api/core/v1:go_default_library",
         "//vendor/k8s.io/apimachinery/pkg/api/errors:go_default_library",

vendor/k8s.io/kubernetes/pkg/controller/nodeipam/ipam/adapter.go

@@ -52,7 +52,7 @@ func newAdapter(k8s clientset.Interface, cloud *gce.GCECloud) *adapter {
 	ret.recorder = broadcaster.NewRecorder(scheme.Scheme, v1.EventSource{Component: "cloudCIDRAllocator"})
 	glog.V(0).Infof("Sending events to api server.")
 	broadcaster.StartRecordingToSink(&v1core.EventSinkImpl{
-		Interface: v1core.New(k8s.CoreV1().RESTClient()).Events(""),
+		Interface: k8s.CoreV1().Events(""),
 	})
 
 	return ret

vendor/k8s.io/kubernetes/pkg/controller/nodeipam/ipam/cidr_allocator.go

@@ -69,6 +69,12 @@ const (
 
 	// cidrUpdateRetries is the no. of times a NodeSpec update will be retried before dropping it.
 	cidrUpdateRetries = 3
+
+	// updateRetryTimeout is the time to wait before requeing a failed node for retry
+	updateRetryTimeout = 100 * time.Millisecond
+
+	// updateMaxRetries is the max retries for a failed node
+	updateMaxRetries = 10
 )
 
 // CIDRAllocator is an interface implemented by things that know how

vendor/k8s.io/kubernetes/pkg/controller/nodeipam/ipam/cloud_cidr_allocator.go

@@ -20,13 +20,14 @@ import (
 	"fmt"
 	"net"
 	"sync"
+	"time"
 
 	"github.com/golang/glog"
 
+	"k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/types"
 	utilruntime "k8s.io/apimachinery/pkg/util/runtime"
-	"k8s.io/apimachinery/pkg/util/sets"
 	informers "k8s.io/client-go/informers/core/v1"
 	corelisters "k8s.io/client-go/listers/core/v1"
 	"k8s.io/client-go/tools/cache"
@@ -41,9 +42,16 @@ import (
 	"k8s.io/kubernetes/pkg/cloudprovider/providers/gce"
 	"k8s.io/kubernetes/pkg/controller"
 	nodeutil "k8s.io/kubernetes/pkg/controller/util/node"
+	"k8s.io/kubernetes/pkg/scheduler/algorithm"
 	utilnode "k8s.io/kubernetes/pkg/util/node"
+	utiltaints "k8s.io/kubernetes/pkg/util/taints"
 )
 
+// nodeProcessingInfo tracks information related to current nodes in processing
+type nodeProcessingInfo struct {
+	retries int
+}
+
 // cloudCIDRAllocator allocates node CIDRs according to IP address aliases
 // assigned by the cloud provider. In this case, the allocation and
 // deallocation is delegated to the external provider, and the controller
@@ -67,7 +75,7 @@ type cloudCIDRAllocator struct {
 
 	// Keep a set of nodes that are currectly being processed to avoid races in CIDR allocation
 	lock              sync.Mutex
-	nodesInProcessing sets.String
+	nodesInProcessing map[string]*nodeProcessingInfo
 }
 
 var _ CIDRAllocator = (*cloudCIDRAllocator)(nil)
@@ -82,7 +90,7 @@ func NewCloudCIDRAllocator(client clientset.Interface, cloud cloudprovider.Inter
 	recorder := eventBroadcaster.NewRecorder(scheme.Scheme, v1.EventSource{Component: "cidrAllocator"})
 	eventBroadcaster.StartLogging(glog.Infof)
 	glog.V(0).Infof("Sending events to api server.")
-	eventBroadcaster.StartRecordingToSink(&v1core.EventSinkImpl{Interface: v1core.New(client.CoreV1().RESTClient()).Events("")})
+	eventBroadcaster.StartRecordingToSink(&v1core.EventSinkImpl{Interface: client.CoreV1().Events("")})
 
 	gceCloud, ok := cloud.(*gce.GCECloud)
 	if !ok {
@@ -97,7 +105,7 @@ func NewCloudCIDRAllocator(client clientset.Interface, cloud cloudprovider.Inter
 		nodesSynced:       nodeInformer.Informer().HasSynced,
 		nodeUpdateChannel: make(chan string, cidrUpdateQueueSize),
 		recorder:          recorder,
-		nodesInProcessing: sets.NewString(),
+		nodesInProcessing: map[string]*nodeProcessingInfo{},
 	}
 
 	nodeInformer.Informer().AddEventHandler(cache.ResourceEventHandlerFuncs{
@@ -108,8 +116,9 @@ func NewCloudCIDRAllocator(client clientset.Interface, cloud cloudprovider.Inter
 			}
 			// Even if PodCIDR is assigned, but NetworkUnavailable condition is
 			// set to true, we need to process the node to set the condition.
+			networkUnavailableTaint := &v1.Taint{Key: algorithm.TaintNodeNetworkUnavailable, Effect: v1.TaintEffectNoSchedule}
 			_, cond := v1node.GetNodeCondition(&newNode.Status, v1.NodeNetworkUnavailable)
-			if cond == nil || cond.Status != v1.ConditionFalse {
+			if cond == nil || cond.Status != v1.ConditionFalse || utiltaints.TaintExists(newNode.Spec.Taints, networkUnavailableTaint) {
 				return ca.AllocateOrOccupyCIDR(newNode)
 			}
 			return nil
@@ -147,9 +156,15 @@ func (ca *cloudCIDRAllocator) worker(stopChan <-chan struct{}) {
 				return
 			}
 			if err := ca.updateCIDRAllocation(workItem); err != nil {
-				// Requeue the failed node for update again.
-				ca.nodeUpdateChannel <- workItem
+				if ca.canRetry(workItem) {
+					time.AfterFunc(updateRetryTimeout, func() {
+						// Requeue the failed node for update again.
+						ca.nodeUpdateChannel <- workItem
+					})
+					continue
+				}
 			}
+			ca.removeNodeFromProcessing(workItem)
 		case <-stopChan:
 			return
 		}
@@ -159,17 +174,28 @@ func (ca *cloudCIDRAllocator) worker(stopChan <-chan struct{}) {
 func (ca *cloudCIDRAllocator) insertNodeToProcessing(nodeName string) bool {
 	ca.lock.Lock()
 	defer ca.lock.Unlock()
-	if ca.nodesInProcessing.Has(nodeName) {
+	if _, found := ca.nodesInProcessing[nodeName]; found {
 		return false
 	}
-	ca.nodesInProcessing.Insert(nodeName)
+	ca.nodesInProcessing[nodeName] = &nodeProcessingInfo{}
+	return true
+}
+
+func (ca *cloudCIDRAllocator) canRetry(nodeName string) bool {
+	ca.lock.Lock()
+	defer ca.lock.Unlock()
+	count := ca.nodesInProcessing[nodeName].retries + 1
+	if count > updateMaxRetries {
+		return false
+	}
+	ca.nodesInProcessing[nodeName].retries = count
 	return true
 }
 
 func (ca *cloudCIDRAllocator) removeNodeFromProcessing(nodeName string) {
 	ca.lock.Lock()
 	defer ca.lock.Unlock()
-	ca.nodesInProcessing.Delete(nodeName)
+	delete(ca.nodesInProcessing, nodeName)
 }
 
 // WARNING: If you're adding any return calls or defer any more work from this
@@ -191,10 +217,11 @@ func (ca *cloudCIDRAllocator) AllocateOrOccupyCIDR(node *v1.Node) error {
 
 // updateCIDRAllocation assigns CIDR to Node and sends an update to the API server.
 func (ca *cloudCIDRAllocator) updateCIDRAllocation(nodeName string) error {
-	defer ca.removeNodeFromProcessing(nodeName)
-
 	node, err := ca.nodeLister.Get(nodeName)
 	if err != nil {
+		if errors.IsNotFound(err) {
+			return nil // node no longer available, skip processing
+		}
 		glog.Errorf("Failed while getting node %v for updating Node.Spec.PodCIDR: %v", nodeName, err)
 		return err
 	}

vendor/k8s.io/kubernetes/pkg/controller/nodeipam/ipam/cloud_cidr_allocator_test.go

@@ -0,0 +1,59 @@
+/*
+Copyright 2018 The Kubernetes Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package ipam
+
+import (
+	"testing"
+	"time"
+
+	"k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/client-go/informers"
+	"k8s.io/client-go/kubernetes/fake"
+)
+
+func hasNodeInProcessing(ca *cloudCIDRAllocator, name string) bool {
+	ca.lock.Lock()
+	defer ca.lock.Unlock()
+
+	_, found := ca.nodesInProcessing[name]
+	return found
+}
+
+func TestBoundedRetries(t *testing.T) {
+	clientSet := fake.NewSimpleClientset()
+	updateChan := make(chan string, 1) // need to buffer as we are using only on go routine
+	stopChan := make(chan struct{})
+	sharedInfomer := informers.NewSharedInformerFactory(clientSet, 1*time.Hour)
+	ca := &cloudCIDRAllocator{
+		client:            clientSet,
+		nodeUpdateChannel: updateChan,
+		nodeLister:        sharedInfomer.Core().V1().Nodes().Lister(),
+		nodesSynced:       sharedInfomer.Core().V1().Nodes().Informer().HasSynced,
+		nodesInProcessing: map[string]*nodeProcessingInfo{},
+	}
+	go ca.worker(stopChan)
+	nodeName := "testNode"
+	ca.AllocateOrOccupyCIDR(&v1.Node{
+		ObjectMeta: metav1.ObjectMeta{
+			Name: nodeName,
+		},
+	})
+	for hasNodeInProcessing(ca, nodeName) {
+		// wait for node to finish processing (should terminate and not time out)
+	}
+}

vendor/k8s.io/kubernetes/pkg/controller/nodeipam/ipam/controller.go

@@ -174,14 +174,15 @@ func (c *Controller) onAdd(node *v1.Node) error {
 	c.lock.Lock()
 	defer c.lock.Unlock()
 
-	if syncer, ok := c.syncers[node.Name]; !ok {
+	syncer, ok := c.syncers[node.Name]
+	if !ok {
 		syncer = c.newSyncer(node.Name)
 		c.syncers[node.Name] = syncer
 		go syncer.Loop(nil)
 	} else {
 		glog.Warningf("Add for node %q that already exists", node.Name)
-		syncer.Update(node)
 	}
+	syncer.Update(node)
 
 	return nil
 }

vendor/k8s.io/kubernetes/pkg/controller/nodeipam/ipam/range_allocator.go

@@ -76,7 +76,7 @@ func NewCIDRRangeAllocator(client clientset.Interface, nodeInformer informers.No
 	recorder := eventBroadcaster.NewRecorder(scheme.Scheme, v1.EventSource{Component: "cidrAllocator"})
 	eventBroadcaster.StartLogging(glog.Infof)
 	glog.V(0).Infof("Sending events to api server.")
-	eventBroadcaster.StartRecordingToSink(&v1core.EventSinkImpl{Interface: v1core.New(client.CoreV1().RESTClient()).Events("")})
+	eventBroadcaster.StartRecordingToSink(&v1core.EventSinkImpl{Interface: client.CoreV1().Events("")})
 
 	set, err := cidrset.NewCIDRSet(clusterCIDR, subNetMaskSize)
 	if err != nil {