search

MaikuMori / helm-charts

Maiku's Helm charts
https://artifacthub.io/packages/search?user=MaikuMori&sort=relevance
MIT License
18 stars 11 forks source link

Autoscaling api #15

Closed tweiss-mdm closed 1 year ago

tweiss-mdm commented 1 year ago

Hello, When using the helm chart (5.2.0) with autoscaling enabled, the following errors is thrown:

Error: INSTALLATION FAILED: unable to build kubernetes objects from release manifest: unable to recognize "": no matches for kind "HorizontalPodAutoscaler" in version "autoscaling/v2beta1"

I suggest replaceing the api => autoscaling/v2

MaikuMori commented 1 year ago

I'm about to make a new release with gotenberg 7.9.1. I'll include the proposed fix.

I'll be honest, I haven't tested horizontal autoscaling much.

Eventually, I should either add conditionals based on K8S version or document supported versions. Currently, this tracks "recent" K8S versions because that's what I use in GKE.

tweiss-mdm commented 1 year ago

Nice, btw have you ever considered to test the charts ? I did use https://github.com/helm-unittest/helm-unittest a couple of times and liked it a lot

MaikuMori commented 1 year ago

I haven't used that, but I do have Helm install test on CI, it's just that it was using only testing against a single K8S version. I just updated it to use the last 5.

MaikuMori commented 1 year ago

It should be on artifacthub.io in ~30 mins.

MaikuMori commented 1 year ago

@tweiss-mdm I forgot to add, let me know how it works. I'm not convinced that the metrics for autoscaling works well with gotenberg because if you use CPU as the metric, every job is a CPU heavy burst, it would have to average CPU usage over a larger period of time for it to be a decent scaling metric. Memory doesn't seem to be a good metric at all. Then again, maybe what I assume is 15-min rolling average CPU is good enough.

It's also possible to expose advanced configuration options in the char values if needed. I didn't want to break current config options.