aws / aws/containers-roadmap

[EKS] [request]: deleting Fargate profile should respect pod disruption budgets (PDBs)

Open
#996 1 comment 5 reactions 0 assignees View on GitHub
EKS Fargate Proposed
Dominant language
Shell
Stars
5.4k
Forks
334
PR merge metrics
No merged PRs in 30d

Description

### Community Note

* Please vote on this issue by adding a 👍 [reaction](https://blog.github.com/2016-03-10-add-reactions-to-pull-requests-issues-and-comments/) to the original issue to help the community and maintainers prioritize this request
* Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
* If you are interested in working on this issue or have submitted a pull request, please leave a comment

**Tell us about your request**
Deleting a Fargate profile immediately terminates all pods that are running with this profile. Instead, it should respect PDBs.

**Which service(s) is this request for?**
EKS, Fargate

**Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?**
We want to manage Fargate profiles with Configuration as Code. As Fargate profiles cannot be updated in-place (subnets, namespace and label selector), we need to create new ones and delete the old ones in order to update them. Deleting profiles does not respect PDBs and immediately terminates all pods running with the profile. Also, it's not clear which profile will be chosen if multiple profiles match, so it's not really possible to "drain" a profile gracefully before deleting it.

Also, the documentation doesn't state if the profile deletion is graceful for the pods (i.e. it respects the `terminationGracePeriodSeconds`). Can someone answer this question? If it respects it, what's the maximum value before pods are forcefully terminated?

Contributor guide

Open the contributing guide

Research direction

Start with the EKS Fargate profile deletion behavior and the Kubernetes PDB and terminationGracePeriodSeconds semantics described in the issue. Confirm whether deletion drains pods gracefully, how multiple matching profiles are selected, and the maximum termination grace period; done means the behavior respects PDBs or the documentation clearly states the limits.

Written by the indexing model from the issue text.

Assessment

Tech stack
aws, kubernetes
Domain
cloud, infrastructure
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.