feat(eks): support Bottlerocket Nvidia AMIs (#28287)

Adds support for `BOTTLEROCKET_ARM_64_NVIDIA` and `BOTTLEROCKET_x86_64_NVIDIA` AMI types. Closes #28241. ---- *By submitting this pull request, I confirm that my contribution is made under the terms of the Apache-2.0 license*
aws · Dec 8, 2023 · 6aa1b1b · 6aa1b1b
1 parent 020bf18
commit 6aa1b1b
Show file tree

Hide file tree

Showing 3 changed files with 71 additions and 11 deletions.
diff --git a/packages/aws-cdk-lib/aws-eks/README.md b/packages/aws-cdk-lib/aws-eks/README.md
@@ -558,6 +558,17 @@ For example, if the Amazon EKS cluster version is `1.17`, the Bottlerocket AMI v
 
 Please note Bottlerocket does not allow to customize bootstrap options and `bootstrapOptions` properties is not supported when you create the `Bottlerocket` capacity.
 
+To create a Bottlerocket managed nodegroup with Nvidia-based EC2 instance types use the `BOTTLEROCKET_X86_64_NVIDIA` or
+`BOTTLEROCKET_ARM_64_NVIDIA` AMIs:
+
+```ts
+declare const cluster: eks.Cluster;
+cluster.addNodegroupCapacity('BottlerocketNvidiaNG', {
+  amiType: eks.NodegroupAmiType.BOTTLEROCKET_X86_64_NVIDIA,
+  instanceTypes: [new ec2.InstanceType('g4dn.xlarge')],
+});
+```
+
 For more details about Bottlerocket, see [Bottlerocket FAQs](https://aws.amazon.com/bottlerocket/faqs/) and [Bottlerocket Open Source Blog](https://aws.amazon.com/blogs/opensource/announcing-the-general-availability-of-bottlerocket-an-open-source-linux-distribution-purpose-built-to-run-containers/).
 
 ### Endpoint Access

diff --git a/packages/aws-cdk-lib/aws-eks/lib/managed-nodegroup.ts b/packages/aws-cdk-lib/aws-eks/lib/managed-nodegroup.ts
@@ -17,9 +17,13 @@ export interface INodegroup extends IResource {
 }
 
 /**
- * The AMI type for your node group. GPU instance types should use the `AL2_x86_64_GPU` AMI type, which uses the
- * Amazon EKS-optimized Linux AMI with GPU support. Non-GPU instances should use the `AL2_x86_64` AMI type, which
- * uses the Amazon EKS-optimized Linux AMI.
+ * The AMI type for your node group.
+ *
+ * GPU instance types should use the `AL2_x86_64_GPU` AMI type, which uses the
+ * Amazon EKS-optimized Linux AMI with GPU support or the `BOTTLEROCKET_ARM_64_NVIDIA` or `BOTTLEROCKET_X86_64_NVIDIA`
+ * AMI types, which uses the Amazon EKS-optimized Linux AMI with Nvidia-GPU support.
+ *
+ * Non-GPU instances should use the `AL2_x86_64` AMI type, which uses the Amazon EKS-optimized Linux AMI.
  */
 export enum NodegroupAmiType {
   /**
@@ -35,13 +39,21 @@ export enum NodegroupAmiType {
    */
   AL2_ARM_64 = 'AL2_ARM_64',
   /**
-   *  Bottlerocket Linux(ARM-64)
+   *  Bottlerocket Linux (ARM-64)
    */
   BOTTLEROCKET_ARM_64 = 'BOTTLEROCKET_ARM_64',
   /**
-   * Bottlerocket(x86-64)
+   * Bottlerocket (x86-64)
    */
   BOTTLEROCKET_X86_64 = 'BOTTLEROCKET_x86_64',
+  /**
+   *  Bottlerocket Linux with Nvidia-GPU support (ARM-64)
+   */
+  BOTTLEROCKET_ARM_64_NVIDIA = 'BOTTLEROCKET_ARM_64_NVIDIA',
+  /**
+   * Bottlerocket with Nvidia-GPU support (x86-64)
+   */
+  BOTTLEROCKET_X86_64_NVIDIA = 'BOTTLEROCKET_x86_64_NVIDIA',
   /**
    * Windows Core 2019 (x86-64)
    */
@@ -215,7 +227,7 @@ export interface NodegroupOptions {
   /**
    * The instance type to use for your node group. Currently, you can specify a single instance type for a node group.
    * The default value for this parameter is `t3.medium`. If you choose a GPU instance type, be sure to specify the
-   * `AL2_x86_64_GPU` with the amiType parameter.
+   * `AL2_x86_64_GPU`, `BOTTLEROCKET_ARM_64_NVIDIA`, or `BOTTLEROCKET_x86_64_NVIDIA` with the amiType parameter.
    *
    * @default t3.medium
    * @deprecated Use `instanceTypes` instead.
@@ -409,7 +421,7 @@ export class Nodegroup extends Resource implements INodegroup {
 
       // if the user explicitly configured an ami type, make sure it's included in the possibleAmiTypes
       if (props.amiType && !possibleAmiTypes.includes(props.amiType)) {
-        throw new Error(`The specified AMI does not match the instance types architecture, either specify one of ${possibleAmiTypes} or don't specify any`);
+        throw new Error(`The specified AMI does not match the instance types architecture, either specify one of ${possibleAmiTypes.join(', ')} or don't specify any`);
       }
 
       //if the user explicitly configured a Windows ami type, make sure the instanceType is allowed
@@ -550,7 +562,8 @@ const x8664AmiTypes: NodegroupAmiType[] = [NodegroupAmiType.AL2_X86_64, Nodegrou
 const windowsAmiTypes: NodegroupAmiType[] = [NodegroupAmiType.WINDOWS_CORE_2019_X86_64,
   NodegroupAmiType.WINDOWS_CORE_2022_X86_64, NodegroupAmiType.WINDOWS_FULL_2019_X86_64,
   NodegroupAmiType.WINDOWS_FULL_2022_X86_64];
-const gpuAmiTypes: NodegroupAmiType[] = [NodegroupAmiType.AL2_X86_64_GPU];
+const gpuAmiTypes: NodegroupAmiType[] = [NodegroupAmiType.AL2_X86_64_GPU,
+  NodegroupAmiType.BOTTLEROCKET_X86_64_NVIDIA, NodegroupAmiType.BOTTLEROCKET_ARM_64_NVIDIA];
 
 /**
  * This function check if the instanceType is GPU instance.

diff --git a/packages/aws-cdk-lib/aws-eks/test/nodegroup.test.ts b/packages/aws-cdk-lib/aws-eks/test/nodegroup.test.ts
@@ -558,7 +558,7 @@ describe('node group', () => {
         new ec2.InstanceType('p3.large'),
         new ec2.InstanceType('g3.large'),
       ],
-    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64_GPU or don't specify any/);
+    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64_GPU, BOTTLEROCKET_x86_64_NVIDIA, BOTTLEROCKET_ARM_64_NVIDIA or don't specify any/);
   });
 
   /**
@@ -580,7 +580,7 @@ describe('node group', () => {
         new ec2.InstanceType('c5.large'),
         new ec2.InstanceType('m5.large'),
       ],
-    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64,BOTTLEROCKET_x86_64,WINDOWS_CORE_2019_x86_64,WINDOWS_CORE_2022_x86_64,WINDOWS_FULL_2019_x86_64,WINDOWS_FULL_2022_x86_64 or don't specify any/);
+    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64, BOTTLEROCKET_x86_64, WINDOWS_CORE_2019_x86_64, WINDOWS_CORE_2022_x86_64, WINDOWS_FULL_2019_x86_64, WINDOWS_FULL_2022_x86_64 or don't specify any/);
   });
 
   test('throws when AmiType is Windows and forbidden instanceType is selected', () => {
@@ -619,7 +619,43 @@ describe('node group', () => {
         new ec2.InstanceType('c5.large'),
         new ec2.InstanceType('m5.large'),
       ],
-    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64,BOTTLEROCKET_x86_64,WINDOWS_CORE_2019_x86_64,WINDOWS_CORE_2022_x86_64,WINDOWS_FULL_2019_x86_64,WINDOWS_FULL_2022_x86_64 or don't specify any/);
+    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64, BOTTLEROCKET_x86_64, WINDOWS_CORE_2019_x86_64, WINDOWS_CORE_2022_x86_64, WINDOWS_FULL_2019_x86_64, WINDOWS_FULL_2022_x86_64 or don't specify any/);
+  });
+
+  test('throws when LaunchTemplate is undefined, amiType is BOTTLEROCKET_ARM_64_NVIDIA and instanceTypes are not GPU', () => {
+    // GIVEN
+    const { stack, vpc } = testFixture();
+    const cluster = new eks.Cluster(stack, 'Cluster', {
+      vpc,
+      defaultCapacity: 0,
+      version: CLUSTER_VERSION,
+    });
+    // THEN
+    expect(() => cluster.addNodegroupCapacity('ng', {
+      amiType: NodegroupAmiType.BOTTLEROCKET_ARM_64_NVIDIA,
+      instanceTypes: [
+        new ec2.InstanceType('c5.large'),
+        new ec2.InstanceType('m5.large'),
+      ],
+    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64, BOTTLEROCKET_x86_64, WINDOWS_CORE_2019_x86_64, WINDOWS_CORE_2022_x86_64, WINDOWS_FULL_2019_x86_64, WINDOWS_FULL_2022_x86_64 or don't specify any/);
+  });
+
+  test('throws when LaunchTemplate is undefined, amiType is BOTTLEROCKET_X86_64_NVIDIA and instanceTypes are not GPU', () => {
+    // GIVEN
+    const { stack, vpc } = testFixture();
+    const cluster = new eks.Cluster(stack, 'Cluster', {
+      vpc,
+      defaultCapacity: 0,
+      version: CLUSTER_VERSION,
+    });
+    // THEN
+    expect(() => cluster.addNodegroupCapacity('ng', {
+      amiType: NodegroupAmiType.BOTTLEROCKET_X86_64_NVIDIA,
+      instanceTypes: [
+        new ec2.InstanceType('c5.large'),
+        new ec2.InstanceType('m5.large'),
+      ],
+    })).toThrow(/The specified AMI does not match the instance types architecture, either specify one of AL2_x86_64, BOTTLEROCKET_x86_64, WINDOWS_CORE_2019_x86_64, WINDOWS_CORE_2022_x86_64, WINDOWS_FULL_2019_x86_64, WINDOWS_FULL_2022_x86_64 or don't specify any/);
   });
 
   /**