Skip to content
This repository has been archived by the owner on Mar 22, 2024. It is now read-only.

Adjust the charts to accomodate Marco's call for StatefulSet support and Deployment support #408

Open
edwbuck opened this issue Jul 26, 2023 · 2 comments

Comments

@edwbuck
Copy link
Contributor

edwbuck commented Jul 26, 2023

During the effort of PR 315, Marco asked for persistence type flags to bring on an additional bit of support for StatefulSet and Deployment support. The idea is that one might choose to deploy using StatefulSets or Deployments, and the Persistence Type (enum) should support both.

After a short review, it seems that this request is worthwhile; however, persistence flags can be delivered without requiring the choice between StatefulSet and Deployments in the current set of Helm Charts. For this reason, in the maintainers meeting of July 25th, 2023, a call was made to promote Marco's suggestion from an attachment "riding the tail coats" of the persistence flag work into a separate issue.

This issue addresses the necessary changes to convert the server deployments to support StatefulSets or Deployments. In the event that one or the other cannot be used due to technical issues, it is approprate to close this issue with a description of why SPIRE is not yet ready to support one of the solutions.

Acceptance criteria:

  • Add a flag to control the SPIRE server deployment to be managed by either a Kubernetes StatefulSet object or a Kubernetes Deployment object
  • Verify functional deployments of both settings
  • If such a flag cannot function correctly, document the reasons why (this may have impacts on certificate suitablility), the conditions under which it fails, and possibly close this issue out without a PR.
@edwbuck
Copy link
Contributor Author

edwbuck commented Jul 26, 2023

@kfox1111 Please read this Issue, to ensure it captures Marco's request. I know he's on vacation, but if it is missing a key component, please update the issue.

Once this issue is fully capturing Marco's request, please proceed on PR 315 as if the request was honored / withdrawn.

@kfox1111
Copy link
Contributor

I think Marco believes the single persistence type flag should determine Deployment over Statefulset if type == emptyDir

My take is it should be a separate flag for type of object used, and could ultimately support StatefulSet, Deployment or DaemonSet. Each has its advantages and disadvantages.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

No branches or pull requests

2 participants