Skip to content

[Bug] Karpenter setting interruptionQueueName renamed to interruptionQueue #7697

Description

@pstast

What were you trying to accomplish?

Use Karpenter with spot instance

What happened?

Instances were abruptly terminated without prior notice. Karpenter was not receiving notifications from AWS events queue about spot interruptions.

How to reproduce it?

Used this Karpenter configuration in ClusterConfig:

karpenter:
  version: '0.35.2'
  createServiceAccount: true
  withSpotInterruptionQueue: true

Cause

eksctl inserts settings.interruptionQueueName value to Helm charts values during Karpenter installation. But this helm value has been renamed to settings.interruptionQueue (without "Name") at the end.

See the current Helm chart values:
https://github.com/aws/karpenter-provider-aws/blob/main/charts/karpenter/values.yaml#L194

The change was introduced in the following pull request merged 3 months ago:
aws/karpenter-provider-aws#4840

As a result, INTERRUPTION_QUEUE environment value is not passed to karpenter pods and the queue is not enabled. After fix (rename value name), INTERRUPTION_QUEUE appears in pod environment variables.

Suggested fix

Rename settings.interruptionQueueName to settings.interruptionQueue in user-provided Helm values when installing Karpenter.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    kind/bugpriority/important-longtermImportant over the long term, but may not be currently staffed and/or may require multiple releases

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions