Skip to content
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

scaleDownDelaySeconds: 0 keeps old ReplicaSet running but should scale it down immediately #1992

Open
AndiDog opened this issue Apr 22, 2022 · 12 comments
Labels
bug Something isn't working no-issue-activity

Comments

@AndiDog
Copy link

AndiDog commented Apr 22, 2022

Summary

With scaleDownDelaySeconds: 0 on a blue-green rollout, and clicking Promote-Full in Argo CD UI for faster deployment, the operator kept the old ReplicaSet running without scaling it down or terminating it (and its pods). We would expect the number of seconds to be respected, so zero means "immediately".

In contrast, any non-zero number like scaleDownDelaySeconds: 1 worked fine in our pre-production testing.

There are other reported issues where the meaning of zero was understood differently, so it's important to document it correctly and fix the behavior accordingly. See #1639 and #1841, for example.

Diagnostics

argo-rollouts v1.1.1

(I cannot offer logs right now)


Message from the maintainers:

Impacted by this bug? Give it a 👍. We prioritize the issues with the most 👍.

@AndiDog AndiDog added the bug Something isn't working label Apr 22, 2022
@zachaller
Copy link
Collaborator

So there is some documentation around this here https://argoproj.github.io/argo-rollouts/features/scaledown-aborted-rs/ but it seems to focus around when it is aborted which it might fall into that logic when clicking full promotion. I will try to take a look at this at some point to see if the behavior is correct and also that it matches what is documented.

@kzcPo
Copy link

kzcPo commented May 16, 2022

I think scaleDownDelaySeconds: 0 means permanent retention, and I want to add manual deletion action.

@mitchellhuang
Copy link

We have also experience this and were surprised by it. We assumed by setting it to 0, there would be no delay and the ReplicaSet would be terminated immediately. We also ended up changing it to 1 to get closer towards the desired behavior.

@coolbry95
Copy link

Another option if you want the replica sets to scale down immediately is to use scaleDownDelaySeconds: -1.

This is what we moved to since 0 keeps them indefinitely.

@kzcPo
Copy link

kzcPo commented Aug 5, 2022 via email

@harikrongali harikrongali added this to the v1.4 milestone Oct 20, 2022
@zachaller zachaller removed this from the v1.4 milestone Dec 21, 2022
@github-actions
Copy link
Contributor

This issue is stale because it has been open 60 days with no activity.

@kzcPo
Copy link

kzcPo commented Feb 20, 2023 via email

@github-actions
Copy link
Contributor

This issue is stale because it has been open 60 days with no activity.

@kzcPo
Copy link

kzcPo commented Apr 23, 2023 via email

@github-actions
Copy link
Contributor

This issue is stale because it has been open 60 days with no activity.

@kzcPo
Copy link

kzcPo commented Jun 23, 2023 via email

@github-actions
Copy link
Contributor

This issue is stale because it has been open 60 days with no activity.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working no-issue-activity
Projects
None yet
Development

No branches or pull requests

6 participants