Cherry-pick AWS SDK go updates

cluster-autoscaler/cloudprovider/aws/README.md

@@ -426,6 +426,8 @@ To refresh static list, please run `go run ec2_instance_types/gen.go` under
 
 ## Using the AWS SDK vendored in the AWS cloudprovider
 
+### v1
+
 If you want to use a newer version of the AWS SDK than the version currently vendored as a direct dependency by Cluster Autoscaler, then you can use the version vendored under this AWS cloudprovider.
 
 The current version vendored is `v1.48.7`.
@@ -437,6 +439,60 @@ If you want to update the vendored AWS SDK to a newer version, please make sure
 3. Update the import statements within the newly-copied AWS SDK to reference the new paths (e.g., `github.com/aws/aws-sdk-go/aws/awsutil` -> `k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go/aws/awsutil`). You can use this command from the aws-sdk-go folder `find . -type f -exec sed -i ‘s#github.com/aws/aws-sdk-go#k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go#’ {} \;`
 4. Update the version number above to indicate the new vendored version.
 
+### v2
+If you want to use a newer version of the AWS SDK than the version currently vendored as a direct dependency by Cluster Autoscaler, then you can use the version vendored under this AWS cloudprovider.
+
+The current version vendored is `2025-06-17`.
+
+The aws-sdk-go-v2 also depends on aws/smithy-go; this is similarly vendored; check the go.mod of the aws-sdk-go-v2 package for the correct version of this package.
+
+The current version vendored is `v1.22.4`
+
+You can use the commands below to update the vendored copy of these modules to a newer version
+
+```shell
+export SDK_VERSION="2025-06-17"
+export SMITHY_VERSION="1.22.4"
+
+# Prepare clean smithy-go folder
+## Clean up old release
+rm -rf smithy-go/
+
+## Download and extract specified release
+curl -L "https://github.com/aws/smithy-go/archive/refs/tags/v${SMITHY_VERSION}.tar.gz" -o smithy-go.tar.gz
+tar xf smithy-go.tar.gz
+mv "smithy-go-${SMITHY_VERSION}" smithy-go
+rm smithy-go.tar.gz
+
+# Adjust for vendoring within cluster-autoscaler 
+## Remove unneeded files, directories, and tests to reduce size
+rm -r ./smithy-go/.github ./smithy-go/codegen ./smithy-go/.travis.yml
+find ./smithy-go \( -name "*_test.go" -o -name 'go.mod' -o -name 'go.sum' \) -exec rm {} \+
+
+## Update paths for local vendoring
+find ./smithy-go -name '*.go' -type f -exec sed -i '' 's#github.com/aws/smithy-go#k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/smithy-go#' {} \+
+
+# Prepare clean sdk-v2 folder
+## Clean up old release
+rm -rf aws-sdk-go-v2/
+
+## Download and extract specified release
+curl -L "https://github.com/aws/aws-sdk-go-v2/archive/refs/tags/release-${SDK_VERSION}.tar.gz" -o aws-sdk-go-v2.tar.gz
+tar xf aws-sdk-go-v2.tar.gz
+mv "aws-sdk-go-v2-release-${SDK_VERSION}" aws-sdk-go-v2
+rm aws-sdk-go-v2.tar.gz
+
+# Adjust for vendoring within cluster-autoscaler
+## Remove unneeded files, directories, and tests to reduce size
+rm ./aws-sdk-go-v2/.travis.yml ./aws-sdk-go-v2/.golangci.toml ./aws-sdk-go-v2/buildspec.yml ./aws-sdk-go-v2/ci-find-smithy-go.sh ./aws-sdk-go-v2/local-mod-replace.sh ./aws-sdk-go-v2/modman.toml
+rm -r ./aws-sdk-go-v2/.github ./aws-sdk-go-v2/codegen ./aws-sdk-go-v2/example ./aws-sdk-go-v2/internal/codegen ./aws-sdk-go-v2/internal/repotools
+find ./aws-sdk-go-v2 \( -name "*_test.go" -o -name "*.go.snap" -o -name 'go.mod' -o -name 'go.sum' \) -exec rm {} \+
+
+## Update imports to local path
+find ./aws-sdk-go-v2 -name '*.go' -type f -exec sed -i '' 's#github.com/aws/aws-sdk-go-v2#k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go-v2#' {} \+
+find ./aws-sdk-go-v2 -name '*.go' -type f -exec sed -i '' 's#github.com/aws/smithy-go#k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/smithy-go#' {} \+
+```
+
 ## Using cloud config with helm
 
 If you want to use custom AWS cloud config e.g. endpoint urls

cluster-autoscaler/cloudprovider/aws/auto_scaling_groups.go

@@ -17,15 +17,17 @@ limitations under the License.
 package aws
 
 import (
+	"context"
 	"fmt"
 	"reflect"
 	"strings"
 	"sync"
 	"time"
 
-	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go/aws"
-	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go/service/autoscaling"
-	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go/service/ec2"
+	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go-v2/aws"
+	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go-v2/service/autoscaling"
+	autoscalingtypes "k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go-v2/service/autoscaling/types"
+	ec2types "k8s.io/autoscaler/cluster-autoscaler/cloudprovider/aws/aws-sdk-go-v2/service/ec2/types"
 	"k8s.io/autoscaler/cluster-autoscaler/config/dynamic"
 	klog "k8s.io/klog/v2"
 )
@@ -41,7 +43,7 @@ type asgCache struct {
 	asgToInstances       map[AwsRef][]AwsInstanceRef
 	instanceToAsg        map[AwsInstanceRef]*asg
 	instanceStatus       map[AwsInstanceRef]*string
-	instanceLifecycle    map[AwsInstanceRef]*string
+	instanceLifecycle    map[AwsInstanceRef]autoscalingtypes.LifecycleState
 	asgInstanceTypeCache *instanceTypeExpirationStore
 	mutex                sync.Mutex
 	awsService           *awsWrapper
@@ -60,8 +62,8 @@ type launchTemplate struct {
 type mixedInstancesPolicy struct {
 	launchTemplate                *launchTemplate
 	instanceTypesOverrides        []string
-	instanceRequirementsOverrides *autoscaling.InstanceRequirements
-	instanceRequirements          *ec2.InstanceRequirements
+	instanceRequirementsOverrides *autoscalingtypes.InstanceRequirements
+	instanceRequirements          *ec2types.InstanceRequirements
 }
 
 type asg struct {
@@ -76,7 +78,7 @@ type asg struct {
 	LaunchConfigurationName string
 	LaunchTemplate          *launchTemplate
 	MixedInstancesPolicy    *mixedInstancesPolicy
-	Tags                    []*autoscaling.TagDescription
+	Tags                    []autoscalingtypes.TagDescription
 }
 
 func newASGCache(awsService *awsWrapper, explicitSpecs []string, autoDiscoverySpecs []asgAutoDiscoveryConfig) (*asgCache, error) {
@@ -86,7 +88,7 @@ func newASGCache(awsService *awsWrapper, explicitSpecs []string, autoDiscoverySp
 		asgToInstances:        make(map[AwsRef][]AwsInstanceRef),
 		instanceToAsg:         make(map[AwsInstanceRef]*asg),
 		instanceStatus:        make(map[AwsInstanceRef]*string),
-		instanceLifecycle:     make(map[AwsInstanceRef]*string),
+		instanceLifecycle:     make(map[AwsInstanceRef]autoscalingtypes.LifecycleState),
 		asgInstanceTypeCache:  newAsgInstanceTypeCache(awsService),
 		interrupt:             make(chan struct{}),
 		asgAutoDiscoverySpecs: autoDiscoverySpecs,
@@ -243,12 +245,12 @@ func (m *asgCache) InstanceStatus(ref AwsInstanceRef) (*string, error) {
 	return nil, fmt.Errorf("could not find instance %v", ref)
 }
 
-func (m *asgCache) findInstanceLifecycle(ref AwsInstanceRef) (*string, error) {
+func (m *asgCache) findInstanceLifecycle(ref AwsInstanceRef) (autoscalingtypes.LifecycleState, error) {
 	if lifecycle, found := m.instanceLifecycle[ref]; found {
 		return lifecycle, nil
 	}
 
-	return nil, fmt.Errorf("could not find instance %v", ref)
+	return "", fmt.Errorf("could not find instance %v", ref)
 }
 
 func (m *asgCache) SetAsgSize(asg *asg, size int) error {
@@ -261,12 +263,12 @@ func (m *asgCache) SetAsgSize(asg *asg, size int) error {
 func (m *asgCache) setAsgSizeNoLock(asg *asg, size int) error {
 	params := &autoscaling.SetDesiredCapacityInput{
 		AutoScalingGroupName: aws.String(asg.Name),
-		DesiredCapacity:      aws.Int64(int64(size)),
+		DesiredCapacity:      aws.Int32(int32(size)),
 		HonorCooldown:        aws.Bool(false),
 	}
 	klog.V(0).Infof("Setting asg %s size to %d", asg.Name, size)
 	start := time.Now()
-	_, err := m.awsService.SetDesiredCapacity(params)
+	_, err := m.awsService.SetDesiredCapacity(context.Background(), params)
 	observeAWSRequest("SetDesiredCapacity", err, start)
 	if err != nil {
 		return err
@@ -356,12 +358,11 @@ func (m *asgCache) DeleteInstances(instances []*AwsInstanceRef) error {
 			return err
 		}
 
-		if lifecycle != nil &&
-			*lifecycle == autoscaling.LifecycleStateTerminated ||
-			*lifecycle == autoscaling.LifecycleStateTerminating ||
-			*lifecycle == autoscaling.LifecycleStateTerminatingWait ||
-			*lifecycle == autoscaling.LifecycleStateTerminatingProceed {
-			klog.V(2).Infof("instance %s is already terminating in state %s, will skip instead", instance.Name, *lifecycle)
+		if lifecycle == autoscalingtypes.LifecycleStateTerminated ||
+			lifecycle == autoscalingtypes.LifecycleStateTerminating ||
+			lifecycle == autoscalingtypes.LifecycleStateTerminatingWait ||
+			lifecycle == autoscalingtypes.LifecycleStateTerminatingProceed {
+			klog.V(2).Infof("instance %s is already terminating in state %s, will skip instead", instance.Name, lifecycle)
 			continue
 		}
 
@@ -370,12 +371,12 @@ func (m *asgCache) DeleteInstances(instances []*AwsInstanceRef) error {
 			ShouldDecrementDesiredCapacity: aws.Bool(true),
 		}
 		start := time.Now()
-		resp, err := m.awsService.TerminateInstanceInAutoScalingGroup(params)
+		resp, err := m.awsService.TerminateInstanceInAutoScalingGroup(context.Background(), params)
 		observeAWSRequest("TerminateInstanceInAutoScalingGroup", err, start)
 		if err != nil {
 			return err
 		}
-		klog.V(4).Infof(*resp.Activity.Description)
+		klog.V(4).Infof("Terminated instance %s in autoscaling group: %s", instance.Name, aws.ToString(resp.Activity.Description))
 
 		// Proactively decrement the size so autoscaler makes better decisions
 		commonAsg.curSize--
@@ -421,7 +422,7 @@ func (m *asgCache) regenerate() error {
 	newInstanceToAsgCache := make(map[AwsInstanceRef]*asg)
 	newAsgToInstancesCache := make(map[AwsRef][]AwsInstanceRef)
 	newInstanceStatusMap := make(map[AwsInstanceRef]*string)
-	newInstanceLifecycleMap := make(map[AwsInstanceRef]*string)
+	newInstanceLifecycleMap := make(map[AwsInstanceRef]autoscalingtypes.LifecycleState)
 
 	// Fetch details of all ASGs
 	refreshNames := m.buildAsgNames()
@@ -448,7 +449,7 @@ func (m *asgCache) regenerate() error {
 	// Register or update ASGs
 	exists := make(map[AwsRef]bool)
 	for _, group := range groups {
-		asg, err := m.buildAsgFromAWS(group)
+		asg, err := m.buildAsgFromAWS(&group)
 		if err != nil {
 			return err
 		}
@@ -497,19 +498,21 @@ func (m *asgCache) regenerate() error {
 	return nil
 }
 
-func (m *asgCache) createPlaceholdersForDesiredNonStartedInstances(groups []*autoscaling.Group) []*autoscaling.Group {
+func (m *asgCache) createPlaceholdersForDesiredNonStartedInstances(groups []autoscalingtypes.AutoScalingGroup) []autoscalingtypes.AutoScalingGroup {
+	var updatedGroups []autoscalingtypes.AutoScalingGroup
 	for _, g := range groups {
 		desired := *g.DesiredCapacity
-		realInstances := int64(len(g.Instances))
+		realInstances := int32(len(g.Instances))
 		if desired <= realInstances {
+			updatedGroups = append(updatedGroups, g)
 			continue
 		}
 
 		klog.V(4).Infof("Instance group %s has only %d instances created while requested count is %d. "+
 			"Creating placeholder instances.", *g.AutoScalingGroupName, realInstances, desired)
 
 		healthStatus := ""
-		isAvailable, err := m.isNodeGroupAvailable(g)
+		isAvailable, err := m.isNodeGroupAvailable(&g)
 		if err != nil {
 			klog.V(4).Infof("Could not check instance availability, creating placeholder node anyways: %v", err)
 		} else if !isAvailable {
@@ -519,23 +522,24 @@ func (m *asgCache) createPlaceholdersForDesiredNonStartedInstances(groups []*aut
 
 		for i := realInstances; i < desired; i++ {
 			id := fmt.Sprintf("%s-%s-%d", placeholderInstanceNamePrefix, *g.AutoScalingGroupName, i)
-			g.Instances = append(g.Instances, &autoscaling.Instance{
+			g.Instances = append(g.Instances, autoscalingtypes.Instance{
 				InstanceId:       &id,
-				AvailabilityZone: g.AvailabilityZones[0],
+				AvailabilityZone: aws.String(g.AvailabilityZones[0]),
 				HealthStatus:     &healthStatus,
 			})
 		}
+		updatedGroups = append(updatedGroups, g)
 	}
-	return groups
+	return updatedGroups
 }
 
-func (m *asgCache) isNodeGroupAvailable(group *autoscaling.Group) (bool, error) {
+func (m *asgCache) isNodeGroupAvailable(group *autoscalingtypes.AutoScalingGroup) (bool, error) {
 	input := &autoscaling.DescribeScalingActivitiesInput{
 		AutoScalingGroupName: group.AutoScalingGroupName,
 	}
 
 	start := time.Now()
-	response, err := m.awsService.DescribeScalingActivities(input)
+	response, err := m.awsService.DescribeScalingActivities(context.Background(), input)
 	observeAWSRequest("DescribeScalingActivities", err, start)
 	if err != nil {
 		return true, err // If we can't describe the scaling activities we assume the node group is available
@@ -547,8 +551,8 @@ func (m *asgCache) isNodeGroupAvailable(group *autoscaling.Group) (bool, error)
 			lut := a.lastUpdateTime
 			if activity.StartTime.Before(lut) {
 				break
-			} else if *activity.StatusCode == "Failed" {
-				klog.Warningf("ASG %s scaling failed with %s", asgRef.Name, *activity)
+			} else if activity.StatusCode == autoscalingtypes.ScalingActivityStatusCodeFailed {
+				klog.Warningf("ASG %s scaling failed with description: %s", asgRef.Name, aws.ToString(activity.Description))
 				return false, nil
 			}
 		} else {
@@ -558,11 +562,11 @@ func (m *asgCache) isNodeGroupAvailable(group *autoscaling.Group) (bool, error)
 	return true, nil
 }
 
-func (m *asgCache) buildAsgFromAWS(g *autoscaling.Group) (*asg, error) {
+func (m *asgCache) buildAsgFromAWS(g *autoscalingtypes.AutoScalingGroup) (*asg, error) {
 	spec := dynamic.NodeGroupSpec{
-		Name:               aws.StringValue(g.AutoScalingGroupName),
-		MinSize:            int(aws.Int64Value(g.MinSize)),
-		MaxSize:            int(aws.Int64Value(g.MaxSize)),
+		Name:               aws.ToString(g.AutoScalingGroupName),
+		MinSize:            int(aws.ToInt32(g.MinSize)),
+		MaxSize:            int(aws.ToInt32(g.MaxSize)),
 		SupportScaleToZero: scaleToZeroSupported,
 	}
 
@@ -575,9 +579,9 @@ func (m *asgCache) buildAsgFromAWS(g *autoscaling.Group) (*asg, error) {
 		minSize: spec.MinSize,
 		maxSize: spec.MaxSize,
 
-		curSize:                 int(aws.Int64Value(g.DesiredCapacity)),
-		AvailabilityZones:       aws.StringValueSlice(g.AvailabilityZones),
-		LaunchConfigurationName: aws.StringValue(g.LaunchConfigurationName),
+		curSize:                 int(aws.ToInt32(g.DesiredCapacity)),
+		AvailabilityZones:       g.AvailabilityZones,
+		LaunchConfigurationName: aws.ToString(g.LaunchConfigurationName),
 		Tags:                    g.Tags,
 	}
 
@@ -586,8 +590,8 @@ func (m *asgCache) buildAsgFromAWS(g *autoscaling.Group) (*asg, error) {
 	}
 
 	if g.MixedInstancesPolicy != nil {
-		getInstanceTypes := func(overrides []*autoscaling.LaunchTemplateOverrides) []string {
-			res := []string{}
+		getInstanceTypes := func(overrides []autoscalingtypes.LaunchTemplateOverrides) []string {
+			var res []string
 			for _, override := range overrides {
 				if override.InstanceType != nil {
 					res = append(res, *override.InstanceType)
@@ -596,7 +600,7 @@ func (m *asgCache) buildAsgFromAWS(g *autoscaling.Group) (*asg, error) {
 			return res
 		}
 
-		getInstanceTypeRequirements := func(overrides []*autoscaling.LaunchTemplateOverrides) *autoscaling.InstanceRequirements {
+		getInstanceTypeRequirements := func(overrides []autoscalingtypes.LaunchTemplateOverrides) *autoscalingtypes.InstanceRequirements {
 			if len(overrides) == 1 && overrides[0].InstanceRequirements != nil {
 				return overrides[0].InstanceRequirements
 			}
@@ -625,8 +629,8 @@ func (m *asgCache) buildAsgFromAWS(g *autoscaling.Group) (*asg, error) {
 	return asg, nil
 }
 
-func (m *asgCache) getInstanceRequirementsFromMixedInstancesPolicy(policy *mixedInstancesPolicy) (*ec2.InstanceRequirements, error) {
-	instanceRequirements := &ec2.InstanceRequirements{}
+func (m *asgCache) getInstanceRequirementsFromMixedInstancesPolicy(policy *mixedInstancesPolicy) (*ec2types.InstanceRequirements, error) {
+	instanceRequirements := &ec2types.InstanceRequirements{}
 	if policy.instanceRequirementsOverrides != nil {
 		var err error
 		instanceRequirements, err = m.awsService.getEC2RequirementsFromAutoscaling(policy.instanceRequirementsOverrides)
@@ -646,11 +650,11 @@ func (m *asgCache) getInstanceRequirementsFromMixedInstancesPolicy(policy *mixed
 	return instanceRequirements, nil
 }
 
-func (m *asgCache) buildInstanceRefFromAWS(instance *autoscaling.Instance) AwsInstanceRef {
-	providerID := fmt.Sprintf("aws:///%s/%s", aws.StringValue(instance.AvailabilityZone), aws.StringValue(instance.InstanceId))
+func (m *asgCache) buildInstanceRefFromAWS(instance autoscalingtypes.Instance) AwsInstanceRef {
+	providerID := fmt.Sprintf("aws:///%s/%s", aws.ToString(instance.AvailabilityZone), aws.ToString(instance.InstanceId))
 	return AwsInstanceRef{
 		ProviderID: providerID,
-		Name:       aws.StringValue(instance.InstanceId),
+		Name:       aws.ToString(instance.InstanceId),
 	}
 }