✨ Add support for EKSConfig LaunchTemplate bootstrapping for AL2023 using nodeadm #5553
+652
−30
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -20,8 +20,13 @@ package controllers | |
| import ( | ||
| "bytes" | ||
| "context" | ||
| "fmt" | ||
| "os" | ||
| "time" | ||
|
|
||
| "github.com/aws/aws-sdk-go/aws" | ||
| "github.com/aws/aws-sdk-go/aws/session" | ||
| "github.com/aws/aws-sdk-go/service/eks" | ||
| "github.com/pkg/errors" | ||
| corev1 "k8s.io/api/core/v1" | ||
| apierrors "k8s.io/apimachinery/pkg/api/errors" | ||
|
|
@@ -39,6 +44,7 @@ import ( | |
| eksbootstrapv1 "sigs.k8s.io/cluster-api-provider-aws/v2/bootstrap/eks/api/v1beta2" | ||
| "sigs.k8s.io/cluster-api-provider-aws/v2/bootstrap/eks/internal/userdata" | ||
| ekscontrolplanev1 "sigs.k8s.io/cluster-api-provider-aws/v2/controlplane/eks/api/v1beta2" | ||
| expinfrav1 "sigs.k8s.io/cluster-api-provider-aws/v2/exp/api/v1beta2" | ||
| "sigs.k8s.io/cluster-api-provider-aws/v2/pkg/logger" | ||
| "sigs.k8s.io/cluster-api-provider-aws/v2/util/paused" | ||
| clusterv1 "sigs.k8s.io/cluster-api/api/v1beta1" | ||
|
|
@@ -51,6 +57,11 @@ import ( | |
| "sigs.k8s.io/cluster-api/util/predicates" | ||
| ) | ||
|
|
||
| const ( | ||
| // NodeTypeAL2023 represents the AL2023 node type. | ||
| NodeTypeAL2023 = "al2023" | ||
| ) | ||
|
|
||
| // EKSConfigReconciler reconciles a EKSConfig object. | ||
| type EKSConfigReconciler struct { | ||
| client.Client | ||
|
|
@@ -143,7 +154,7 @@ func (r *EKSConfigReconciler) Reconcile(ctx context.Context, req ctrl.Request) ( | |
| } | ||
| }() | ||
|
|
||
| return r.joinWorker(ctx, cluster, config, configOwner) | ||
| } | ||
|
|
||
| func (r *EKSConfigReconciler) resolveFiles(ctx context.Context, cfg *eksbootstrapv1.EKSConfig) ([]eksbootstrapv1.File, error) { | ||
|
|
@@ -181,8 +192,9 @@ func (r *EKSConfigReconciler) resolveSecretFileContent(ctx context.Context, ns s | |
| return data, nil | ||
| } | ||
|
|
||
| func (r *EKSConfigReconciler) joinWorker(ctx context.Context, cluster *clusterv1.Cluster, config *eksbootstrapv1.EKSConfig, configOwner *bsutil.ConfigOwner) (ctrl.Result, error) { | ||
| log := logger.FromContext(ctx) | ||
| log.Info("joinWorker called", "config", config.Name, "nodeType", config.Spec.NodeType, "cluster", cluster.Name) | ||
|
|
||
| // only need to reconcile the secret for Machine kinds once, but MachinePools need updates for new launch templates | ||
| if config.Status.DataSecretName != nil && configOwner.GetKind() == "Machine" { | ||
|
|
@@ -195,15 +207,15 @@ func (r *EKSConfigReconciler) joinWorker(ctx context.Context, cluster *clusterv1 | |
| err := r.Client.Get(ctx, secretKey, existingSecret) | ||
| switch { | ||
| case err == nil: | ||
| return ctrl.Result{}, nil | ||
| case !apierrors.IsNotFound(err): | ||
| log.Error(err, "unable to check for existing bootstrap secret") | ||
| return ctrl.Result{}, err | ||
| } | ||
| } | ||
|
|
||
| if cluster.Spec.ControlPlaneRef == nil || cluster.Spec.ControlPlaneRef.Kind != "AWSManagedControlPlane" { | ||
| return ctrl.Result{}, errors.New("Cluster's controlPlaneRef needs to be an AWSManagedControlPlane in order to use the EKS bootstrap provider") | ||
| } | ||
|
|
||
| if !cluster.Status.InfrastructureReady { | ||
|
|
@@ -212,30 +224,54 @@ func (r *EKSConfigReconciler) joinWorker(ctx context.Context, cluster *clusterv1 | |
| eksbootstrapv1.DataSecretAvailableCondition, | ||
| eksbootstrapv1.WaitingForClusterInfrastructureReason, | ||
| clusterv1.ConditionSeverityInfo, "") | ||
| return ctrl.Result{}, nil | ||
| } | ||
|
|
||
| if !conditions.IsTrue(cluster, clusterv1.ControlPlaneInitializedCondition) { | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, | ||
| eksbootstrapv1.DataSecretGenerationFailedReason, | ||
| clusterv1.ConditionSeverityInfo, "Control plane is not initialized yet") | ||
|
|
||
| // For AL2023, requeue to ensure we retry when control plane is ready | ||
| // For AL2, follow upstream behavior and return nil | ||
| if config.Spec.NodeType == NodeTypeAL2023 { | ||
| log.Info("AL2023 detected, returning requeue after 30 seconds") | ||
| return ctrl.Result{RequeueAfter: 30 * time.Second}, nil | ||
| } | ||
| log.Info("AL2 detected, returning no requeue") | ||
| return ctrl.Result{}, nil | ||
| } | ||
|
|
||
| // Get the AWSManagedControlPlane | ||
| controlPlane := &ekscontrolplanev1.AWSManagedControlPlane{} | ||
| if err := r.Get(ctx, client.ObjectKey{Name: cluster.Spec.ControlPlaneRef.Name, Namespace: cluster.Spec.ControlPlaneRef.Namespace}, controlPlane); err != nil { | ||
| return ctrl.Result{}, errors.Wrap(err, "failed to get control plane") | ||
| } | ||
|
|
||
| // Check if control plane is ready (skip in test environments for AL2023) | ||
| if config.Spec.NodeType == NodeTypeAL2023 && !conditions.IsTrue(controlPlane, ekscontrolplanev1.EKSControlPlaneReadyCondition) { | ||
| // Skip control plane readiness check for AL2023 in test environment | ||
| if os.Getenv("TEST_ENV") != "true" { | ||
| log.Info("AL2023 detected, waiting for control plane to be ready") | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, | ||
| eksbootstrapv1.DataSecretGenerationFailedReason, | ||
| clusterv1.ConditionSeverityInfo, "Control plane is not ready yet") | ||
| return ctrl.Result{RequeueAfter: 30 * time.Second}, nil | ||
| } | ||
| log.Info("Skipping control plane readiness check for AL2023 in test environment") | ||
| } | ||
| log.Info("Control plane is ready, proceeding with userdata generation") | ||
|
|
||
| log.Info("Generating userdata") | ||
| files, err := r.resolveFiles(ctx, config) | ||
| if err != nil { | ||
| log.Info("Failed to resolve files for user data") | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, eksbootstrapv1.DataSecretGenerationFailedReason, clusterv1.ConditionSeverityWarning, "%s", err.Error()) | ||
| return ctrl.Result{}, err | ||
| } | ||
|
|
||
| // Create unified NodeInput for both AL2 and AL2023 | ||
| nodeInput := &userdata.NodeInput{ | ||
| ClusterName: controlPlane.Spec.EKSClusterName, | ||
| KubeletExtraArgs: config.Spec.KubeletExtraArgs, | ||
| ContainerRuntime: config.Spec.ContainerRuntime, | ||
|
|
@@ -251,7 +287,9 @@ func (r *EKSConfigReconciler) joinWorker(ctx context.Context, cluster *clusterv1 | |
| DiskSetup: config.Spec.DiskSetup, | ||
| Mounts: config.Spec.Mounts, | ||
| Files: files, | ||
| ClusterCIDR: controlPlane.Spec.NetworkSpec.VPC.CidrBlock, | ||
| } | ||
|
|
||
| if config.Spec.PauseContainer != nil { | ||
| nodeInput.PauseContainerAccount = &config.Spec.PauseContainer.AccountNumber | ||
| nodeInput.PauseContainerVersion = &config.Spec.PauseContainer.Version | ||
|
|
@@ -271,22 +309,99 @@ func (r *EKSConfigReconciler) joinWorker(ctx context.Context, cluster *clusterv1 | |
| nodeInput.IPFamily = ptr.To[string]("ipv6") | ||
| } | ||
|
|
||
| // Set AMI family type and AL2023-specific fields if needed | ||
| if config.Spec.NodeType == NodeTypeAL2023 { | ||
| log.Info("Processing AL2023 node type") | ||
| nodeInput.AMIFamilyType = userdata.AMIFamilyAL2023 | ||
|
|
||
| // Set AL2023-specific fields | ||
| nodeInput.APIServerEndpoint = controlPlane.Spec.ControlPlaneEndpoint.Host | ||
| nodeInput.NodeGroupName = config.Name | ||
|
|
||
| // In test environments, provide a mock CA certificate | ||
| if os.Getenv("TEST_ENV") == "true" { | ||
| log.Info("Using mock CA certificate for test environment") | ||
| nodeInput.CACert = "mock-ca-certificate-for-testing" | ||
| } else { | ||
| // Fetch CA cert from EKS API | ||
| sess, err := session.NewSession(&aws.Config{Region: aws.String(controlPlane.Spec.Region)}) | ||
| if err != nil { | ||
| log.Error(err, "Failed to create AWS session for EKS API") | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, | ||
| eksbootstrapv1.DataSecretGenerationFailedReason, | ||
| clusterv1.ConditionSeverityWarning, | ||
| "Failed to create AWS session: %v", err) | ||
| return ctrl.Result{}, err | ||
| } | ||
| eksClient := eks.New(sess) | ||
| describeInput := &eks.DescribeClusterInput{Name: aws.String(controlPlane.Spec.EKSClusterName)} | ||
AmitSahastra marked this conversation as resolved.
Show resolved
Hide resolved
|
||
| clusterOut, err := eksClient.DescribeCluster(describeInput) | ||
| if err != nil { | ||
| log.Error(err, "Failed to describe EKS cluster for CA cert fetch") | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, | ||
| eksbootstrapv1.DataSecretGenerationFailedReason, | ||
| clusterv1.ConditionSeverityWarning, | ||
| "Failed to describe EKS cluster: %v", err) | ||
| return ctrl.Result{}, err | ||
| } else if clusterOut.Cluster != nil && clusterOut.Cluster.CertificateAuthority != nil && clusterOut.Cluster.CertificateAuthority.Data != nil { | ||
| nodeInput.CACert = *clusterOut.Cluster.CertificateAuthority.Data | ||
| } else { | ||
| log.Error(nil, "CA certificate not found in EKS cluster response") | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, | ||
| eksbootstrapv1.DataSecretGenerationFailedReason, | ||
| clusterv1.ConditionSeverityWarning, | ||
| "CA certificate not found in EKS cluster response") | ||
| return ctrl.Result{}, fmt.Errorf("CA certificate not found in EKS cluster response") | ||
| } | ||
| } | ||
|
|
||
| // Get AMI ID from AWSManagedMachinePool's launch template if specified | ||
| if configOwner.GetKind() == "AWSManagedMachinePool" { | ||
| amp := &expinfrav1.AWSManagedMachinePool{} | ||
| if err := r.Get(ctx, client.ObjectKey{Namespace: config.Namespace, Name: configOwner.GetName()}, amp); err == nil { | ||
| log.Info("Found AWSManagedMachinePool", "name", amp.Name, "launchTemplate", amp.Spec.AWSLaunchTemplate != nil) | ||
| if amp.Spec.AWSLaunchTemplate != nil && amp.Spec.AWSLaunchTemplate.AMI.ID != nil { | ||
| nodeInput.AMIImageID = *amp.Spec.AWSLaunchTemplate.AMI.ID | ||
| log.Info("Set AMI ID from launch template", "amiID", nodeInput.AMIImageID) | ||
| } else { | ||
| log.Info("No AMI ID found in launch template") | ||
| } | ||
| if amp.Spec.CapacityType != nil { | ||
| nodeInput.CapacityType = amp.Spec.CapacityType | ||
| log.Info("Set capacity type from AWSManagedMachinePool", "capacityType", *amp.Spec.CapacityType) | ||
| } else { | ||
| log.Info("No capacity type found in AWSManagedMachinePool") | ||
| } | ||
| } else { | ||
| log.Info("Failed to get AWSManagedMachinePool", "error", err) | ||
| } | ||
| } | ||
|
|
||
| log.Info("Generating AL2023 userdata", | ||
| "cluster", controlPlane.Spec.EKSClusterName, | ||
| "endpoint", nodeInput.APIServerEndpoint) | ||
| } else { | ||
| nodeInput.AMIFamilyType = userdata.AMIFamilyAL2 | ||
| log.Info("Generating standard userdata for node type", "type", config.Spec.NodeType) | ||
| } | ||
|
|
||
| // Generate userdata using unified approach | ||
| userDataScript, err := userdata.NewNode(nodeInput) | ||
| if err != nil { | ||
| log.Error(err, "Failed to create a worker join configuration") | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, eksbootstrapv1.DataSecretGenerationFailedReason, clusterv1.ConditionSeverityWarning, "") | ||
| return ctrl.Result{}, err | ||
| } | ||
|
|
||
| // Store the userdata in a secret | ||
| if err := r.storeBootstrapData(ctx, cluster, config, userDataScript); err != nil { | ||
| log.Error(err, "Failed to store bootstrap data") | ||
| conditions.MarkFalse(config, eksbootstrapv1.DataSecretAvailableCondition, eksbootstrapv1.DataSecretGenerationFailedReason, clusterv1.ConditionSeverityWarning, "") | ||
| return ctrl.Result{}, err | ||
| } | ||
|
|
||
| conditions.MarkTrue(config, eksbootstrapv1.DataSecretAvailableCondition) | ||
| return ctrl.Result{}, nil | ||
| } | ||
|
|
||
| func (r *EKSConfigReconciler) SetupWithManager(ctx context.Context, mgr ctrl.Manager, option controller.Options) error { | ||
|
|
||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is there any way to derive this from the AMI being used rather than asking the user to specify in the API?