Autoscaler: Adopt Cluster Autoscaler Helm chart

[sc250024]

Description

Since the Helm stable channel will be deprecated eventually, I'm wondering if it's possible that this repository can adopt the stable/cluster-autoscaler Helm chart into this repository?

Other projects are doing the same, notably NGINX Ingress (https://github.com/kubernetes/ingress-nginx) and Kubernetes dashboard (https://github.com/kubernetes/dashboard) as simple examples.

[gjtempleton]

Hi @sc250024 thanks for raising this

We've discussed it at the SIG Autoscaling meeting earlier today and we don't have any objection to adopting the chart going forward as long as there are people willing to maintain it going forward. I'm willing to take this on, but if there are any others who want to contribute to this please let us know.

We'll need to discuss the specifics of how we do this in more detail.

[sc250024]

@gjtempleton I would be willing to maintain it. I can open a PR for it soon.

I've noticed that the current CI process for the autoscaler is using Travis. Do you mind if the Helm chart uses GitHub actions? The reason is because there are already automated actions which handle this quite well.

• https://github.com/helm/chart-testing-action
• https://github.com/helm/chart-releaser-action

Other notable Kubernetes projects (dashboard, nginx-ingress) use these actions as well.

[gjtempleton]

I don't have an objection to Github actions given they're being used elsewhere in the k8s org already, and as you point out those seem to be well supported paths for this.

Best to get input from @MaciekPytel as well.

[MaciekPytel]

I don't have any experience with github actions, but based on the comment above I don't see any reason against using them.

[sc250024]

@MaciekPytel Cool. I'll open a PR.

[sc250024]

PR here: https://github.com/kubernetes/autoscaler/pull/3341

[stevehipwell]

We've done the same thing on the descheduler project and had the same issue around chart naming and CI (waiting on tide to support the testing action).

https://github.com/kubernetes-sigs/descheduler/issues/248

[hakman]

Hi. Quick question in case I am missing something. Should adding the autoscaler repo work?

$ helm version
version.BuildInfo{Version:"v3.2.4", GitCommit:"0ad800ef43d3b826f31a5ad8dfbb4fe05d143688", GitTreeState:"clean", GoVersion:"go1.13.12"}

$ helm repo add autoscaler https://kubernetes.github.io/autoscaler
Error: looks like "https://kubernetes.github.io/autoscaler" is not a valid chart repository or cannot be reached: failed to fetch https://kubernetes.github.io/autoscaler/index.yaml : 404 Not Found

[gjtempleton]

Hey, not yet no, we've run into some issues due to the branch protections etc. with the official chart-publisher action so haven't managed to successfully publish the required index.yaml to the gh-pages branch.

I'm currently trying to find a workaround in #3403

[hakman]

Ah. Thanks for the explanation @gjtempleton :).

[searsaw]

@gjtempleton is the recommendation to just keep using the previous chart in stable then for now until the 404 issue is resolved?

[stevehipwell]

@gjtempleton I think the maintainers on the descheduler project worked around this to get a release out.

[gjtempleton]

@stevehipwell yeah, as far as I'm aware from having had a brief look, that initial release was done via a manual PR pointing at a file generated and uploaded by the chart-releaser action. I was hoping to find a simple solution to allow an automated workflow, but I'll follow their example and publish manually as a first step as this is seemingly causing people issues.

@searsaw / others I'll generate the required PR to manually publish the initial release and try to get it merged as soon as possible.

[lexsys27]

What is the current way of installing the chart? I went through the instructions in the README.md and got an error:

$ helm repo add autoscaler https://kubernetes.github.io/autoscaler
Error: looks like "https://kubernetes.github.io/autoscaler" is not a valid chart repository or cannot be reached: failed to fetch https://kubernetes.github.io/autoscaler/index.yaml : 404 Not Found

[searsaw]

I continued to use the stable/cluster-autoscaler for now.

[lexsys27]

Yep, that still works. Just installed from that chart.

Have not used cluster autoscaler before and a bit confused the docs are referencing repo which is not yet available.

[searsaw]

They switched to using an official deployer action but it isn't working right now.

There is a PR open for manually deploying this chart to the new area that is documented in this repo. We are just waiting for that to be merged in, I believe.

[gjtempleton]

Yeah, I'm waiting on the OK from those I'm adding to the OWNERS file on the gh-pages branch as otherwise I'll end up as a single point of failure on this before mentioning the PR that @searsaw has mentioned.

Sorry again for the inconvenience on this everyone!

[gjtempleton]

3413 has now been merged and I've just validated that I can fetch the chart.

It's not ideal but a manual process will work for now, so I think this issue can now be closed.

[lexsys27]

I confirm the error is solved on my side.

$ helm repo add autoscaler https://kubernetes.github.io/autoscaler
"autoscaler" has been added to your repositories

[searsaw]

Also, I can confirm it worked for me as well. Thank you, @gjtempleton

[yurrriq]

Same. Thanks, everyone! stable/cluster-autoscaler ~will soon be~ has now been marked deprecated by https://github.com/helm/charts/pull/23305

[gjtempleton]

/close

[k8s-ci-robot]

@gjtempleton: Closing this issue.

In response to this:

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[stevehipwell]

The chart releaser action has been updated so it should now be possible to remove the -chart post-fix from the chart name while still having the tags include this.

https://github.com/helm/chart-releaser/pull/74
https://github.com/helm/chart-releaser-action/issues/37