-
Notifications
You must be signed in to change notification settings - Fork 4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
aws-eks: manage nodegroups GPU instance types not up to date #31347
Comments
Your proposed solutions makes good sense to me. Are you interested to submit a PR for that? |
@pahud - definitely interested, I'll put something together over the upcoming days |
Comments on closed issues and PRs are hard for our team to see. |
1 similar comment
Comments on closed issues and PRs are hard for our team to see. |
Describe the bug
As a AWS CDK user for EKS, I want build out a managed node group with a mix of G5/G6/G6e instance types.
This is not possible with the current
isGpuInstanceType
check for managed node groups (here | architecture mapping here). When I specify instance types of both G5 and G6, I receive an error ofinstanceTypes of different architectures is not allowed
due to G6/G6e not existing in theknownGpuInstanceTypes
.Additionally, there is currently no instance class setup currently in the EC2 package for
G6e
instances: https://github.com/aws/aws-cdk/blob/main/packages/aws-cdk-lib/aws-ec2/lib/instance-types.tsRegression Issue
Last Known Working CDK Version
No response
Expected Behavior
I expect to be able to build a managed node group with both G5/G6/G6e instance classes due to the architectures being compatible.
Current Behavior
The error
'instanceTypes of different architectures is not allowed
is being thrown.I cannot reference a G6e instance class.
Reproduction Steps
Create a managed node group with instance types from a mix of G5/G6 instance families and will see the error above.
Try to reference the G6e instance class - it does not exist.
Possible Solution
To fix the G5/G6 issue:
InstanceClass.G6
to the list ofknowGpuInstanceTypes
: https://github.com/aws/aws-cdk/blob/main/packages/aws-cdk-lib/aws-eks/lib/managed-nodegroup.ts#L607To enable G6e instance types and get those compatible as recognized GPU instance types:
InstanceClass.G6E
to the list ofknowGpuInstanceTypes
: https://github.com/aws/aws-cdk/blob/main/packages/aws-cdk-lib/aws-eks/lib/managed-nodegroup.ts#L607Additional Information/Context
Happy to open a PR to handle this - just wanted to get the discussion going
CDK CLI Version
2.156.0
Framework Version
No response
Node.js Version
20.17.0
OS
macOS 14.6.1
Language
TypeScript
Language Version
Version 5.5.4
Other information
No response
The text was updated successfully, but these errors were encountered: