How do you properly test changes to a Helm chart?

Consider using chart tests and the GitHub prehook to validate YAML configurations. You can also perform a dry run. Running lightweight clusters like K3D or KIND can really streamline this process.

Implement pre-upgrade hooks in Helm. For example, I have a job that checks the application’s environment variables to ensure they’re set correctly for production. We follow a pipeline that transitions changes from dev to test to staging and then to production, provided all checks pass. If something fails, the pipeline stops the changes from proceeding.

One effective way to test Helm charts is through unit testing using tools like `helm-unittest`. For a more practical approach, you can set up a local Kubernetes cluster using K3D to deploy your chart and interact with the application in a real environment.

You could try deploying your chart locally with Minikube or K3S. It's also useful to render the chart output or use the Helm diff plugin to see what changes are pending before applying them.

The best practice is to deploy the chart to a test cluster, or at least a test namespace. This way, you can run smoke tests against the actual deployment to ensure everything works as intended.