CA: Debugging snapshot adding a new field for TemplateNode. This capt…

…ures all the templates for nodegroups present

cluster-autoscaler/core/static_autoscaler.go

@@ -291,6 +291,8 @@ func (a *StaticAutoscaler) RunOnce(currentTime time.Time) errors.AutoscalerError
 		return autoscalerError.AddPrefix("failed to build node infos for node groups: ")
 	}
 
+	a.DebuggingSnapshotter.SetTemplateNodes(nodeInfosForGroups)
+
 	nodeInfosForGroups, err = a.processors.NodeInfoProcessor.Process(autoscalingContext, nodeInfosForGroups)
 	if err != nil {
 		klog.Errorf("Failed to process nodeInfos: %v", err)
@@ -416,9 +418,9 @@ func (a *StaticAutoscaler) RunOnce(currentTime time.Time) errors.AutoscalerError
 
 	l, err := a.ClusterSnapshot.NodeInfos().List()
 	if err != nil {
-		klog.Errorf("Unable to fetch NodeInfo List for Debugging Snapshot, %v", err)
+		klog.Errorf("Unable to fetch ClusterNode List for Debugging Snapshot, %v", err)
 	} else {
-		a.AutoscalingContext.DebuggingSnapshotter.SetNodeGroupInfo(l)
+		a.AutoscalingContext.DebuggingSnapshotter.SetClusterNodes(l)
 	}
 
 	unschedulablePodsToHelp, _ := a.processors.PodListProcessor.Process(a.AutoscalingContext, unschedulablePods)

cluster-autoscaler/debuggingsnapshot/debugging_snapshot.go

@@ -25,20 +25,23 @@ import (
 	"k8s.io/kubernetes/pkg/scheduler/framework"
 )
 
-// NodeInfo captures a single entity of nodeInfo. i.e. Node specs and all the pods on that node.
-type NodeInfo struct {
-	Node *v1.Node             `json:"Node"`
-	Pods []*framework.PodInfo `json:"Pods"`
+// ClusterNode captures a single entity of nodeInfo. i.e. Node specs and all the pods on that node.
+type ClusterNode struct {
+	Node *v1.Node  `json:"Node"`
+	Pods []*v1.Pod `json:"Pods"`
 }
 
 // DebuggingSnapshot is the interface used to define any debugging snapshot
 // implementation, incl. any custom impl. to be used by DebuggingSnapshotter
 type DebuggingSnapshot interface {
-	// SetNodeGroupInfo is a setter to capture all the NodeInfo
-	SetNodeGroupInfo([]*framework.NodeInfo)
-	// SetUnscheduledPodsCanBeScheduled is a setter for all pods which are unscheduled
+	// SetClusterNodes is a setter to capture all the ClusterNode
+	SetClusterNodes([]*framework.NodeInfo)
+	// SetUnscheduledPodsCanBeScheduled is a setter for all pods which are unscheduled,
 	// but they can be scheduled. i.e. pods which aren't triggering scale-up
 	SetUnscheduledPodsCanBeScheduled([]*v1.Pod)
+	// SetTemplateNodes is a setter for all the TemplateNodes present in the cluster
+	// incl. templates for which there are no nodes
+	SetTemplateNodes(map[string]*framework.NodeInfo)
 	// SetErrorMessage sets the error message in the snapshot
 	SetErrorMessage(string)
 	// SetEndTimestamp sets the timestamp in the snapshot,
@@ -58,11 +61,12 @@ type DebuggingSnapshot interface {
 // Please add all new output fields in this struct. This is to make the data
 // encoding/decoding easier as the single object going into the decoder
 type DebuggingSnapshotImpl struct {
-	NodeInfo                      []*NodeInfo `json:"NodeList"`
-	UnscheduledPodsCanBeScheduled []*v1.Pod   `json:"UnscheduledPodsCanBeScheduled"`
-	Error                         string      `json:"Error,omitempty"`
-	StartTimestamp                time.Time   `json:"StartTimestamp"`
-	EndTimestamp                  time.Time   `json:"EndTimestamp"`
+	NodeList                      []*ClusterNode          `json:"NodeList"`
+	UnscheduledPodsCanBeScheduled []*v1.Pod               `json:"UnscheduledPodsCanBeScheduled"`
+	Error                         string                  `json:"Error,omitempty"`
+	StartTimestamp                time.Time               `json:"StartTimestamp"`
+	EndTimestamp                  time.Time               `json:"EndTimestamp"`
+	TemplateNodes                 map[string]*ClusterNode `json:"TemplateNodes"`
 }
 
 // SetUnscheduledPodsCanBeScheduled is the setter for UnscheduledPodsCanBeScheduled
@@ -73,31 +77,48 @@ func (s *DebuggingSnapshotImpl) SetUnscheduledPodsCanBeScheduled(podList []*v1.P
 
 	s.UnscheduledPodsCanBeScheduled = nil
 	for _, pod := range podList {
-		s.UnscheduledPodsCanBeScheduled = append(s.UnscheduledPodsCanBeScheduled, pod)
+		s.UnscheduledPodsCanBeScheduled = append(s.UnscheduledPodsCanBeScheduled, pod.DeepCopy())
 	}
 }
 
-// SetNodeGroupInfo is the setter for Node Group Info
+// SetTemplateNodes is the setter for TemplateNodes
+func (s *DebuggingSnapshotImpl) SetTemplateNodes(templates map[string]*framework.NodeInfo) {
+	if templates == nil {
+		return
+	}
+
+	s.TemplateNodes = make(map[string]*ClusterNode)
+	for ng, template := range templates {
+		s.TemplateNodes[ng] = GetClusterNodeCopy(template)
+	}
+}
+
+// GetClusterNodeCopy is an util func to copy template node and filter values
+func GetClusterNodeCopy(template *framework.NodeInfo) *ClusterNode {
+	cNode := &ClusterNode{}
+	cNode.Node = template.Node().DeepCopy()
+	var pods []*v1.Pod
+	for _, p := range template.Pods {
+		pods = append(pods, p.Pod.DeepCopy())
+	}
+	cNode.Pods = pods
+	return cNode
+}
+
+// SetClusterNodes is the setter for Node Group Info
 // All filtering/prettifying of data should be done here.
-func (s *DebuggingSnapshotImpl) SetNodeGroupInfo(nodeInfos []*framework.NodeInfo) {
+func (s *DebuggingSnapshotImpl) SetClusterNodes(nodeInfos []*framework.NodeInfo) {
 	if nodeInfos == nil {
 		return
 	}
 
-	var NodeInfoList []*NodeInfo
+	var NodeInfoList []*ClusterNode
 
 	for _, n := range nodeInfos {
-		nClone := n.Clone()
-		node := nClone.Node()
-
-		nodeInfo := &NodeInfo{
-			Node: node,
-			Pods: nClone.Pods,
-		}
-
-		NodeInfoList = append(NodeInfoList, nodeInfo)
+		clusterNode := GetClusterNodeCopy(n)
+		NodeInfoList = append(NodeInfoList, clusterNode)
 	}
-	s.NodeInfo = NodeInfoList
+	s.NodeList = NodeInfoList
 }
 
 // SetEndTimestamp is the setter for end timestamp

cluster-autoscaler/debuggingsnapshot/debugging_snapshot_test.go

@@ -59,7 +59,7 @@ func TestBasicSetterWorkflow(t *testing.T) {
 	nodeGroups = append(nodeGroups, nodeInfo)
 	nodeGroups[0].SetNode(node)
 	timestamp := time.Now().In(time.UTC)
-	snapshot.SetNodeGroupInfo(nodeGroups)
+	snapshot.SetClusterNodes(nodeGroups)
 	snapshot.SetEndTimestamp(timestamp)
 	op, err := snapshot.GetOutputBytes()
 	assert.False(t, err)
@@ -87,9 +87,7 @@ func TestBasicSetterWorkflow(t *testing.T) {
 	assert.IsType(t, JSONList{}, pNodeInfo["Pods"])
 	assert.Greater(t, len(pNodeInfo["Pods"].([]interface{})), 0)
 	assert.IsType(t, JSONMap{}, pNodeInfo["Pods"].([]interface{})[0])
-	pPodInfo := pNodeInfo["Pods"].([]interface{})[0].(map[string]interface{})
-	assert.IsType(t, JSONMap{}, pPodInfo["Pod"])
-	pPod := pPodInfo["Pod"].(map[string]interface{})
+	pPod := pNodeInfo["Pods"].([]interface{})[0].(map[string]interface{})
 	assert.IsType(t, JSONMap{}, pPod["metadata"])
 	pPodMeta := pPod["metadata"].(map[string]interface{})
 	assert.IsType(t, String, pPodMeta["name"])

cluster-autoscaler/debuggingsnapshot/debugging_snapshotter.go

@@ -74,11 +74,14 @@ type DebuggingSnapshotter interface {
 	// StartDataCollection will check the State(s) and enable data
 	// collection for the loop if applicable
 	StartDataCollection()
-	// SetNodeGroupInfo is a setter to capture all the NodeInfo
-	SetNodeGroupInfo([]*framework.NodeInfo)
+	// SetClusterNodes is a setter to capture all the ClusterNode
+	SetClusterNodes([]*framework.NodeInfo)
 	// SetUnscheduledPodsCanBeScheduled is a setter for all pods which are unscheduled
 	// but they can be scheduled. i.e. pods which aren't triggering scale-up
 	SetUnscheduledPodsCanBeScheduled([]*v1.Pod)
+	// SetTemplateNodes is a setter for all the TemplateNodes present in the cluster
+	// incl. templates for which there are no nodes
+	SetTemplateNodes(map[string]*framework.NodeInfo)
 	// ResponseHandler is the http response handler to manage incoming requests
 	ResponseHandler(http.ResponseWriter, *http.Request)
 	// IsDataCollectionAllowed checks the internal State of the snapshotter
@@ -205,16 +208,16 @@ func (d *DebuggingSnapshotterImpl) Flush() {
 	}
 }
 
-// SetNodeGroupInfo is the setter for Node Group Info
+// SetClusterNodes is the setter for Node Group Info
 // All filtering/prettifying of data should be done here.
-func (d *DebuggingSnapshotterImpl) SetNodeGroupInfo(nodeInfos []*framework.NodeInfo) {
+func (d *DebuggingSnapshotterImpl) SetClusterNodes(nodeInfos []*framework.NodeInfo) {
 	if !d.IsDataCollectionAllowed() {
 		return
 	}
 	d.Mutex.Lock()
 	defer d.Mutex.Unlock()
-	klog.Infof("NodeGroupInfo is being set for the debugging snapshot")
-	d.DebuggingSnapshot.SetNodeGroupInfo(nodeInfos)
+	klog.V(4).Infof("NodeGroupInfo is being set for the debugging snapshot")
+	d.DebuggingSnapshot.SetClusterNodes(nodeInfos)
 	*d.State = DATA_COLLECTED
 }
 
@@ -225,11 +228,22 @@ func (d *DebuggingSnapshotterImpl) SetUnscheduledPodsCanBeScheduled(podList []*v
 	}
 	d.Mutex.Lock()
 	defer d.Mutex.Unlock()
-	klog.Infof("UnscheduledPodsCanBeScheduled is being set for the debugging snapshot")
+	klog.V(4).Infof("UnscheduledPodsCanBeScheduled is being set for the debugging snapshot")
 	d.DebuggingSnapshot.SetUnscheduledPodsCanBeScheduled(podList)
 	*d.State = DATA_COLLECTED
 }
 
+// SetTemplateNodes is the setter for TemplateNodes
+func (d *DebuggingSnapshotterImpl) SetTemplateNodes(templates map[string]*framework.NodeInfo) {
+	if !d.IsDataCollectionAllowed() {
+		return
+	}
+	klog.V(4).Infof("TemplateNodes is being set for the debugging snapshot")
+	d.Mutex.Lock()
+	defer d.Mutex.Unlock()
+	d.DebuggingSnapshot.SetTemplateNodes(templates)
+}
+
 // Cleanup clears the internal data sets of the cluster
 func (d *DebuggingSnapshotterImpl) Cleanup() {
 	if d.CancelRequest != nil {

cluster-autoscaler/debuggingsnapshot/debugging_snapshotter_test.go

@@ -74,7 +74,7 @@ func TestBasicSnapshotRequest(t *testing.T) {
 	for !snapshotter.IsDataCollectionAllowed() {
 		snapshotter.StartDataCollection()
 	}
-	snapshotter.SetNodeGroupInfo(nodeGroups)
+	snapshotter.SetClusterNodes(nodeGroups)
 	snapshotter.Flush()
 
 	wg.Wait()
@@ -152,7 +152,7 @@ func TestRejectParallelRequest(t *testing.T) {
 	snapshotter.ResponseHandler(w1, req1)
 	assert.Equal(t, http.StatusTooManyRequests, w1.Code)
 
-	snapshotter.SetNodeGroupInfo(nil)
+	snapshotter.SetClusterNodes(nil)
 	snapshotter.Flush()
 	wg.Wait()