kubernetes
/
autoscaler
mirror of https://github.com/kubernetes/autoscaler.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Set PodsTriggeredScaleUp field for failed scale ups.

This commit is contained in:
Aleksandra Gacek 2021-10-14 15:39:24 +02:00
parent c7c14381f5
commit a7660c1f5a
1 changed files with 19 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
cluster-autoscaler/core/scale_up.go
View File

@ -478,9 +478,8 @@ func ScaleUp(context *context.AutoscalingContext, processors *ca_processors.Auto
newNodes = context.MaxNodesTotal - len(nodes) - len(upcomingNodes) newNodes = context.MaxNodesTotal - len(nodes) - len(upcomingNodes)
context.LogRecorder.Eventf(apiv1.EventTypeWarning, "MaxNodesTotalReached", "Max total nodes in cluster reached: %v", context.MaxNodesTotal) context.LogRecorder.Eventf(apiv1.EventTypeWarning, "MaxNodesTotalReached", "Max total nodes in cluster reached: %v", context.MaxNodesTotal)
if newNodes < 1 { if newNodes < 1 {
return scaleUpError(&status.ScaleUpStatus{}, errors.NewAutoscalerError( return scaleUpError(&status.ScaleUpStatus{PodsTriggeredScaleUp: bestOption.Pods},
errors.TransientError, errors.NewAutoscalerError(errors.TransientError, "max node total count already reached"))
"max node total count already reached"))
} }
} }
@ -490,7 +489,7 @@ func ScaleUp(context *context.AutoscalingContext, processors *ca_processors.Auto
createNodeGroupResult, err := processors.NodeGroupManager.CreateNodeGroup(context, bestOption.NodeGroup) createNodeGroupResult, err := processors.NodeGroupManager.CreateNodeGroup(context, bestOption.NodeGroup)
if err != nil { if err != nil {
return scaleUpError( return scaleUpError(
&status.ScaleUpStatus{FailedCreationNodeGroups: []cloudprovider.NodeGroup{bestOption.NodeGroup}}, &status.ScaleUpStatus{FailedCreationNodeGroups: []cloudprovider.NodeGroup{bestOption.NodeGroup}, PodsTriggeredScaleUp: bestOption.Pods},
err) err)
} }
createNodeGroupResults = append(createNodeGroupResults, createNodeGroupResult) createNodeGroupResults = append(createNodeGroupResults, createNodeGroupResult)
@ -522,7 +521,7 @@ func ScaleUp(context *context.AutoscalingContext, processors *ca_processors.Auto
option, err2 := computeExpansionOption(context, podEquivalenceGroups, nodeGroup, nodeInfo, upcomingNodes) option, err2 := computeExpansionOption(context, podEquivalenceGroups, nodeGroup, nodeInfo, upcomingNodes)
if err2 != nil { if err2 != nil {
return scaleUpError(&status.ScaleUpStatus{}, errors.ToAutoscalerError(errors.InternalError, err)) return scaleUpError(&status.ScaleUpStatus{PodsTriggeredScaleUp: bestOption.Pods}, errors.ToAutoscalerError(errors.InternalError, err))
} }
if len(option.Pods) > 0 && option.NodeCount > 0 { if len(option.Pods) > 0 && option.NodeCount > 0 {
@ -541,7 +540,9 @@ func ScaleUp(context *context.AutoscalingContext, processors *ca_processors.Auto
// This should never happen, as we already should have retrieved // This should never happen, as we already should have retrieved
// nodeInfo for any considered nodegroup. // nodeInfo for any considered nodegroup.
klog.Errorf("No node info for: %s", bestOption.NodeGroup.Id()) klog.Errorf("No node info for: %s", bestOption.NodeGroup.Id())
return scaleUpError(&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults}, errors.NewAutoscalerError( return scaleUpError(
&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults, PodsTriggeredScaleUp: bestOption.Pods},
errors.NewAutoscalerError(
errors.CloudProviderError, errors.CloudProviderError,
"No node info for best expansion option!")) "No node info for best expansion option!"))
} }
@ -549,14 +550,18 @@ func ScaleUp(context *context.AutoscalingContext, processors *ca_processors.Auto
// apply upper limits for CPU and memory // apply upper limits for CPU and memory
newNodes, err = applyScaleUpResourcesLimits(context, processors, newNodes, scaleUpResourcesLeft, nodeInfo, bestOption.NodeGroup, resourceLimiter) newNodes, err = applyScaleUpResourcesLimits(context, processors, newNodes, scaleUpResourcesLeft, nodeInfo, bestOption.NodeGroup, resourceLimiter)
if err != nil { if err != nil {
return scaleUpError(&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults}, err) return scaleUpError(
&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults, PodsTriggeredScaleUp: bestOption.Pods},
err)
} }
targetNodeGroups := []cloudprovider.NodeGroup{bestOption.NodeGroup} targetNodeGroups := []cloudprovider.NodeGroup{bestOption.NodeGroup}
if context.BalanceSimilarNodeGroups { if context.BalanceSimilarNodeGroups {
similarNodeGroups, typedErr := processors.NodeGroupSetProcessor.FindSimilarNodeGroups(context, bestOption.NodeGroup, nodeInfos) similarNodeGroups, typedErr := processors.NodeGroupSetProcessor.FindSimilarNodeGroups(context, bestOption.NodeGroup, nodeInfos)
if typedErr != nil { if typedErr != nil {
return scaleUpError(&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults}, typedErr.AddPrefix("Failed to find matching node groups: ")) return scaleUpError(
&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults, PodsTriggeredScaleUp: bestOption.Pods},
typedErr.AddPrefix("Failed to find matching node groups: "))
} }
similarNodeGroups = filterNodeGroupsByPods(similarNodeGroups, bestOption.Pods, expansionOptions) similarNodeGroups = filterNodeGroupsByPods(similarNodeGroups, bestOption.Pods, expansionOptions)
for _, ng := range similarNodeGroups { for _, ng := range similarNodeGroups {
@ -583,7 +588,9 @@ func ScaleUp(context *context.AutoscalingContext, processors *ca_processors.Auto
scaleUpInfos, typedErr := processors.NodeGroupSetProcessor.BalanceScaleUpBetweenGroups( scaleUpInfos, typedErr := processors.NodeGroupSetProcessor.BalanceScaleUpBetweenGroups(
context, targetNodeGroups, newNodes) context, targetNodeGroups, newNodes)
if typedErr != nil { if typedErr != nil {
return scaleUpError(&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults}, typedErr) return scaleUpError(
&status.ScaleUpStatus{CreateNodeGroupResults: createNodeGroupResults, PodsTriggeredScaleUp: bestOption.Pods},
typedErr)
} }
klog.V(1).Infof("Final scale-up plan: %v", scaleUpInfos) klog.V(1).Infof("Final scale-up plan: %v", scaleUpInfos)
for _, info := range scaleUpInfos { for _, info := range scaleUpInfos {
@ -593,6 +600,7 @@ func ScaleUp(context *context.AutoscalingContext, processors *ca_processors.Auto
&status.ScaleUpStatus{ &status.ScaleUpStatus{
CreateNodeGroupResults: createNodeGroupResults, CreateNodeGroupResults: createNodeGroupResults,
FailedResizeNodeGroups: []cloudprovider.NodeGroup{info.Group}, FailedResizeNodeGroups: []cloudprovider.NodeGroup{info.Group},
PodsTriggeredScaleUp: bestOption.Pods,
}, },
typedErr, typedErr,
) )