| title | authors | reviewers | creation-date | last-updated | status | see-also | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Feature Group for Integrating Karpenter with Cluster API |
|
|
2023-10-18 |
2023-10-18 |
proposed |
This document briefly outlines the scope, communication media, and stakeholders for a formal Feature Group dedicated to exploring and defining how Karpenter can be integrated with Cluster API.
As a Cluster API administrator who uses node autoscaling to manage resource in my clusters, I would like to utilize the Karpenter project with Cluster API so that I can gain access to its features and capabilities.
At the time of writing, the Cluster API project only officially supports the Kubernetes Cluster Autoscaler (CAS)for automatically managing node resources. The Karpenter project is an alternative node autoscaler and autoprovisioner which has community requested features, such as pod consolidation, that do not have equivalents in the CAS. We would like to determine how Cluster API users can utilize Karpenter, and if necessary create community supported projects to support this activity.
The scope of this effort will be the following:
- Determine what options are available for a consistent Karpenter on Cluster API experience.
- Agree upon a path forward to guide the Cluster API community with Karpenter usage.
- Work towards a Cluster API Enhancement that describes the Karpenter integration.
- Provide a space and time for discussion of issues related to the Karpenter Cluster API project.
Depending on the outcome of the 2nd item above, we may choose to keep this Feature Group active during the longer term to support Karpenter integration design and work efforts in Cluster API, with the blessing of the larger Cluster API community.
Update: 2024-10-02
After discussion within the feature group and the broader Cluster API community, we are continuing the feature group with a focus on experimentation and working towards a CAEP that describes the integration with Karpenter. Additionally, we will field any project related questions during the feature group.
We will meet on Wednesdays at 11:00 PT (Pacific Time). Convert to your timezone. Meeting notes will be documented in this HackMD document. Meetings will be recorded and posted to the SIG Cluster Lifecycle YouTube playlist.
Regular, summarized updates of group progress will be provided during weekly Cluster API office hours on Wednesdays @ 10:00 PT on Zoom.
Chat with stakeholders on Kubernetes Slack in the cluster-api channel.
Primary Stakeholders are listed below:
- Michael McCune (@elmiko, Red Hat)