bug: autoscaling behavior arg reference

[aliziel]

During one of the recent refactor PRs, a spelling suggestion came up to better align with the HPA spec:

   {{- with .Values.multidim.autoscaling.behavior }}

to match with the spelling in the spec

Originally posted by @sunu in #220 (comment)

---

The eoapi/values.yaml has always used the UK English behaviour, but the HPA templates reference the arg as the US English behavior (even before the refactor):

eoapi-k8s/helm-chart/eoapi/templates/services/hpa.yaml

Lines 16 to 17 in c7176cf

	behavior:
	{{- toYaml (index $v "autoscaling" "behavior") | nindent 4 }}

eoapi-k8s/helm-chart/eoapi/templates/services/stac/hpa.yaml

Lines 15 to 16 in 5cdf18e

	behavior:
	{{- with .Values.stac.autoscaling.behavior }}

eoapi-k8s/helm-chart/eoapi/values.yaml

Lines 353 to 355 in 5cdf18e

	behaviour:
	scaleDown:
	stabilizationWindowSeconds: 60

Options depend on impact and, since we're still getting a sense of usage, these err on the side of caution:

1. The template should be patched to use the current value spelling (UK English)
2. If we want to align with the HPA spec (US English):
   1. Changing at the values level seems like a larger versioning consideration. Include when we cut the next minor/major.
   2. If we know we aren't doing a bigger release anytime soon, we could also patch the template and values file to use either UK/US until switching to just US.
      1. This risks some of the conditional complexity described in Helm Chart Refactoring Proposal: Improving Readability and Maintainability #211, but mentioning because deployed instances have likely always used the HPA defaults.
      2. According to the spec, our behavior.scaleDown.stabilizationWindowSeconds setting is 1/5 of the default — not sure if that's significant enough to justify supporting both. FWIW, its one of only a few values specifically documented:
         

         eoapi-k8s/docs/configuration.md

         Lines 59 to 61 in 5cdf18e

         	#### `autoscaling.behaviour.[scaleDown||scaleUp]`
         	These are normal k8s autoscaling pass throughs. They are stablization windows in seconds to for scaling up or down to prevent flapping from happening. Read more about [the options on the k8s documentation](https://kubernetes.io/docs/tasks/run-application/horizontal-pod-autoscale/#configurable-scaling-behavior)

Wanted to open up the floor for input. Happy to jump on a PR — would likely start with patching the template spelling to UK (1.) if I hear nothing here.

[sunu]

Ah, so sorry -- I totally missed that the actual value definition, and the docs weren’t updated to reflect the spelling in the spec (the US spelling).

Totally agree that we shouldn’t support both variants and introduce unnecessary complexity.

Personally, I’d lean toward updating the spelling everywhere to align with the spec (i.e., using the US spelling), and clearly call it out as a breaking change in the changelog for the next release. Based on our previous discussions around backwards compatibility, this seems like the right time to make such changes -- while we're still doing major refactors and before we lock things down for a more stable 1.0.0 release.

Curious to hear what others think.