What's in a good Helm chart?

[rfratto]

We're planning on building an official Helm chart for the Grafana Agent Operator once we complete support for traces, logs, and integrations.

I'm really interested in collecting community feedback on:

• What do you think a "good" Helm chart is?
• Examples of your favorite Helm charts
• For Helm charts that deploy an operator, does it deploy any custom resources? How do you deploy custom resources to your cluster? Normal kubectl apply?

If you don't have any answers to this questions, I'd also love to hear in general what you'd like to see from the Grafana Agent Operator Helm chart!

[rverma-dev]

Istio ones are pretty neat, have a look at https://github.com/istio/istio/tree/master/manifests/charts/istio-operator. Even I like the current grafana ones, apart from cortex. One thing is that the examples/libsonnet are there with raw configs, and in usual cases I would prefer to consume in the same way.
For initial days, its a good idea to provide something with couple of parameters exposed through values.yaml but I usually use that to view the generated config file and then copy it over.

[hjet]

I will try to add more later today, but for the 3rd bullet (CRDs), kube-prometheus uses the crds special dir that Helm installs first (see https://helm.sh/docs/chart_best_practices/custom_resource_definitions/)

[hjet]

prom operator (https://github.com/helm/charts/tree/master/stable/prometheus-operator) and jaeger (https://github.com/jaegertracing/helm-charts/tree/main/charts/jaeger-operator) do this too

[hjet]

however, if you want to go down the rabbit hole: see helm/helm#5871 and helm/helm#7735