Confusing Helm Chart Configs

[fernandonogueira]

Hi! 👋🏼

I tried to change Keptn's resources, but I had some issues.
I'm using the umbrella chart (keptn) with the following values.yaml

lifecycleOperator:
  schedulingGatesEnabled: true
  allowedNamespaces:
    - "staging"
  resources:
    limits:
      cpu: 500m
      memory: 512Mi
    requests:
      cpu: 50m
      memory: 150Mi

However, lifecycle-operator's pods are still using the default resources.

Checking the helm chart, I noticed sometimes it references lifecycle-operator values via .Values.lifecycleOperator.<property>, sometimes via .Values.<property>.

Is this lifecycleOperator property duplicated for some reason, or should it be considered a bug?

I can contribute with a PR, if needed.

For now, I'll use the following values.yaml as a workaround:

lifecycleOperator:
  schedulingGatesEnabled: true
  allowedNamespaces:
    - "staging"
  lifecycleOperator:
    resources:
    limits:
      cpu: 500m
      memory: 512Mi
    requests:
      cpu: 50m
      memory: 150Mi

Thanks in advance 🙇🏼

[mowies]

Hi! Thanks for posting this!

So, if you check the chart readme for the lifecycle operator subchart here, you'll notice that most values are prefixed either with lifecycleOperator. or scheduler.. That's because there's 2 things there that get potentially deployed and we needed to separate the value somehow.
So to set resources for the lifecycle operator from the umbrella chart, your workaround is actually the correct solution. The value should be lifecycleOperator.lifecycleOperator.resources and lifecycleOperator.scheduler.resources respectively (although I noticed. you have the schedulingGatesEnabled value which makes the scheduler values meaningless since it won't get deployed).

We have changes planned to remove the scheduler in the future, since it's already deprecated. As soon as we remove it, also the values here will get simpler.
Hope this helps :)