Improve configuration management

[keyvaann]

With lot of configuration options available for the stack and constant changes and updates to them managing the changes becomes more difficult and configuration is more likely to get out of sync with the charts or have wrong configuration or format. There should be away to prevent bad configuration applied to the cluster. Current suggestions are:

• Using a linter: This would check the configuration against charts internal values.yaml and some predefined rules to make sure they are in sync.
• Using versions and migrations: In this approach configuration is treated as database schemes and any changes and differences between schemes and actual configuration will be detected by a script and will ask user to migrate the configuration to new format and update it with proper values.

[keyvaann]

Another option is defining the configuration as an CRD which has an schema built in to it and Kubernetes will validate and prevent bad configuration from being applied. It will also be a first step into making Radar-Operator. https://kubernetes.io/docs/tasks/extend-kubernetes/custom-resources/custom-resource-definitions/

[keyvaann]

The easiest and fastest improvement that can be done is defining values.schema.json for charts. Helm will validate given values against defined schema. https://helm.sh/docs/topics/charts/#schema-files https://austindewey.com/2020/06/13/helm-tricks-input-validation-with-values-schema-json/

In order to have a quick start it is possible to use the plugin to create the base schema and then build on top of that. https://github.com/karuppiah7890/helm-schema-gen

Using this schema can make CRD definition minimal or obsolete.