Merge pull request #77491 from andyzhangx/azuredisk-metrics

staging/src/k8s.io/legacy-cloud-providers/azure/azure_backoff.go

@@ -550,25 +550,13 @@ func (az *Cloud) deleteRouteWithRetry(routeName string) error {
 	})
 }
 
-// CreateOrUpdateVMWithRetry invokes az.VirtualMachinesClient.CreateOrUpdate with exponential backoff retry
-func (az *Cloud) CreateOrUpdateVMWithRetry(resourceGroup, vmName string, newVM compute.VirtualMachine) error {
-	return wait.ExponentialBackoff(az.requestBackoff(), func() (bool, error) {
-		ctx, cancel := getContextWithCancel()
-		defer cancel()
-
-		resp, err := az.VirtualMachinesClient.CreateOrUpdate(ctx, resourceGroup, vmName, newVM)
-		klog.V(10).Infof("VirtualMachinesClient.CreateOrUpdate(%s): end", vmName)
-		return az.processHTTPRetryResponse(nil, "", resp, err)
-	})
-}
-
 // UpdateVmssVMWithRetry invokes az.VirtualMachineScaleSetVMsClient.Update with exponential backoff retry
-func (az *Cloud) UpdateVmssVMWithRetry(resourceGroupName string, VMScaleSetName string, instanceID string, parameters compute.VirtualMachineScaleSetVM) error {
+func (az *Cloud) UpdateVmssVMWithRetry(resourceGroupName string, VMScaleSetName string, instanceID string, parameters compute.VirtualMachineScaleSetVM, source string) error {
 	return wait.ExponentialBackoff(az.requestBackoff(), func() (bool, error) {
 		ctx, cancel := getContextWithCancel()
 		defer cancel()
 
-		resp, err := az.VirtualMachineScaleSetVMsClient.Update(ctx, resourceGroupName, VMScaleSetName, instanceID, parameters)
+		resp, err := az.VirtualMachineScaleSetVMsClient.Update(ctx, resourceGroupName, VMScaleSetName, instanceID, parameters, source)
 		klog.V(10).Infof("VirtualMachinesClient.CreateOrUpdate(%s,%s): end", VMScaleSetName, instanceID)
 		return az.processHTTPRetryResponse(nil, "", resp, err)
 	})

staging/src/k8s.io/legacy-cloud-providers/azure/azure_controller_standard.go

@@ -84,7 +84,7 @@ func (as *availabilitySet) AttachDisk(isManagedDisk bool, diskName, diskURI stri
 	// Invalidate the cache right after updating
 	defer as.cloud.vmCache.Delete(vmName)
 
-	_, err = as.VirtualMachinesClient.CreateOrUpdate(ctx, nodeResourceGroup, vmName, newVM)
+	_, err = as.VirtualMachinesClient.CreateOrUpdate(ctx, nodeResourceGroup, vmName, newVM, "attach_disk")
 	if err != nil {
 		klog.Errorf("azureDisk - attach disk(%s, %s) failed, err: %v", diskName, diskURI, err)
 		detail := err.Error()
@@ -151,7 +151,7 @@ func (as *availabilitySet) DetachDisk(diskName, diskURI string, nodeName types.N
 	// Invalidate the cache right after updating
 	defer as.cloud.vmCache.Delete(vmName)
 
-	return as.VirtualMachinesClient.CreateOrUpdate(ctx, nodeResourceGroup, vmName, newVM)
+	return as.VirtualMachinesClient.CreateOrUpdate(ctx, nodeResourceGroup, vmName, newVM, "detach_disk")
 }
 
 // GetDataDisks gets a list of data disks attached to the node.

staging/src/k8s.io/legacy-cloud-providers/azure/azure_controller_vmss.go

@@ -89,7 +89,7 @@ func (ss *scaleSet) AttachDisk(isManagedDisk bool, diskName, diskURI string, nod
 	defer ss.vmssVMCache.Delete(key)
 
 	klog.V(2).Infof("azureDisk - update(%s): vm(%s) - attach disk(%s, %s)", nodeResourceGroup, nodeName, diskName, diskURI)
-	_, err = ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM)
+	_, err = ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM, "attach_disk")
 	if err != nil {
 		detail := err.Error()
 		if strings.Contains(detail, errLeaseFailed) || strings.Contains(detail, errDiskBlobNotFound) {
@@ -159,7 +159,7 @@ func (ss *scaleSet) DetachDisk(diskName, diskURI string, nodeName types.NodeName
 	defer ss.vmssVMCache.Delete(key)
 
 	klog.V(2).Infof("azureDisk - update(%s): vm(%s) - detach disk(%s, %s)", nodeResourceGroup, nodeName, diskName, diskURI)
-	return ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM)
+	return ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM, "detach_disk")
 }
 
 // GetDataDisks gets a list of data disks attached to the node.

staging/src/k8s.io/legacy-cloud-providers/azure/azure_fakes.go

@@ -290,7 +290,7 @@ func newFakeAzureVirtualMachinesClient() *fakeAzureVirtualMachinesClient {
 	return fVMC
 }
 
-func (fVMC *fakeAzureVirtualMachinesClient) CreateOrUpdate(ctx context.Context, resourceGroupName string, VMName string, parameters compute.VirtualMachine) (resp *http.Response, err error) {
+func (fVMC *fakeAzureVirtualMachinesClient) CreateOrUpdate(ctx context.Context, resourceGroupName string, VMName string, parameters compute.VirtualMachine, source string) (resp *http.Response, err error) {
 	fVMC.mutex.Lock()
 	defer fVMC.mutex.Unlock()
 
@@ -550,7 +550,7 @@ func (fVMC *fakeVirtualMachineScaleSetVMsClient) GetInstanceView(ctx context.Con
 	return result, nil
 }
 
-func (fVMC *fakeVirtualMachineScaleSetVMsClient) Update(ctx context.Context, resourceGroupName string, VMScaleSetName string, instanceID string, parameters compute.VirtualMachineScaleSetVM) (resp *http.Response, err error) {
+func (fVMC *fakeVirtualMachineScaleSetVMsClient) Update(ctx context.Context, resourceGroupName string, VMScaleSetName string, instanceID string, parameters compute.VirtualMachineScaleSetVM, source string) (resp *http.Response, err error) {
 	fVMC.mutex.Lock()
 	defer fVMC.mutex.Unlock()
 

staging/src/k8s.io/legacy-cloud-providers/azure/azure_metrics.go

@@ -33,6 +33,7 @@ var (
 		"request",         // API function that is being invoked
 		"resource_group",  // Resource group of the resource being monitored
 		"subscription_id", // Subscription ID of the resource being monitored
+		"source",          // Oeration source(optional)
 	}
 
 	apiMetrics = registerAPIMetrics(metricLabels...)
@@ -43,10 +44,10 @@ type metricContext struct {
 	attributes []string
 }
 
-func newMetricContext(prefix, request, resourceGroup, subscriptionID string) *metricContext {
+func newMetricContext(prefix, request, resourceGroup, subscriptionID, source string) *metricContext {
 	return &metricContext{
 		start:      time.Now(),
-		attributes: []string{prefix + "_" + request, strings.ToLower(resourceGroup), subscriptionID},
+		attributes: []string{prefix + "_" + request, strings.ToLower(resourceGroup), subscriptionID, source},
 	}
 }
 

staging/src/k8s.io/legacy-cloud-providers/azure/azure_metrics_test.go

@@ -23,12 +23,12 @@ import (
 )
 
 func TestAzureMetricLabelCardinality(t *testing.T) {
-	mc := newMetricContext("test", "create", "resource_group", "subscription_id")
+	mc := newMetricContext("test", "create", "resource_group", "subscription_id", "source")
 	assert.Len(t, mc.attributes, len(metricLabels), "cardinalities of labels and values must match")
 }
 
 func TestAzureMetricLabelPrefix(t *testing.T) {
-	mc := newMetricContext("prefix", "request", "resource_group", "subscription_id")
+	mc := newMetricContext("prefix", "request", "resource_group", "subscription_id", "source")
 	found := false
 	for _, attribute := range mc.attributes {
 		if attribute == "prefix_request" {

staging/src/k8s.io/legacy-cloud-providers/azure/azure_test.go

@@ -1108,7 +1108,7 @@ func getClusterResources(az *Cloud, vmCount int, availabilitySetCount int) (clus
 
 		vmCtx, vmCancel := getContextWithCancel()
 		defer vmCancel()
-		_, err := az.VirtualMachinesClient.CreateOrUpdate(vmCtx, az.Config.ResourceGroup, vmName, newVM)
+		_, err := az.VirtualMachinesClient.CreateOrUpdate(vmCtx, az.Config.ResourceGroup, vmName, newVM, "")
 		if err != nil {
 		}
 		// add to kubernetes

staging/src/k8s.io/legacy-cloud-providers/azure/azure_vmss.go

@@ -712,10 +712,10 @@ func (ss *scaleSet) EnsureHostInPool(service *v1.Service, nodeName types.NodeNam
 	ctx, cancel := getContextWithCancel()
 	defer cancel()
 	klog.V(2).Infof("EnsureHostInPool begins to update vmssVM(%s) with new backendPoolID %s", vmName, backendPoolID)
-	resp, err := ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM)
+	resp, err := ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM, "network_update")
 	if ss.CloudProviderBackoff && shouldRetryHTTPRequest(resp, err) {
 		klog.V(2).Infof("EnsureHostInPool update backing off vmssVM(%s) with new backendPoolID %s, err: %v", vmName, backendPoolID, err)
-		retryErr := ss.UpdateVmssVMWithRetry(nodeResourceGroup, ssName, instanceID, newVM)
+		retryErr := ss.UpdateVmssVMWithRetry(nodeResourceGroup, ssName, instanceID, newVM, "network_update")
 		if retryErr != nil {
 			err = retryErr
 			klog.Errorf("EnsureHostInPool update abort backoff vmssVM(%s) with new backendPoolID %s, err: %v", vmName, backendPoolID, err)
@@ -841,10 +841,10 @@ func (ss *scaleSet) ensureBackendPoolDeletedFromNode(service *v1.Service, nodeNa
 	ctx, cancel := getContextWithCancel()
 	defer cancel()
 	klog.V(2).Infof("ensureBackendPoolDeletedFromNode begins to update vmssVM(%s) with backendPoolID %s", nodeName, backendPoolID)
-	resp, err := ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM)
+	resp, err := ss.VirtualMachineScaleSetVMsClient.Update(ctx, nodeResourceGroup, ssName, instanceID, newVM, "network_update")
 	if ss.CloudProviderBackoff && shouldRetryHTTPRequest(resp, err) {
 		klog.V(2).Infof("ensureBackendPoolDeletedFromNode update backing off vmssVM(%s) with backendPoolID %s, err: %v", nodeName, backendPoolID, err)
-		retryErr := ss.UpdateVmssVMWithRetry(nodeResourceGroup, ssName, instanceID, newVM)
+		retryErr := ss.UpdateVmssVMWithRetry(nodeResourceGroup, ssName, instanceID, newVM, "network_update")
 		if retryErr != nil {
 			err = retryErr
 			klog.Errorf("ensureBackendPoolDeletedFromNode update abort backoff vmssVM(%s) with backendPoolID %s, err: %v", nodeName, backendPoolID, err)