In talking with @Eevans yesterday it sounds like it is desirable to have a running Cassandra instance for some portion of Kask tests run via the pipeline. The Deployment Pipeline has a step that can be used for integration tests (using helm test) that step deploys kubernetes resources defined in a helm chart (or in the templates/test directory of the chart) and runs helm test against those resources.
There are several ways to achieve the desired outcome of deploying an ephemeral Cassandra instance during the "test deployment" phase of the pipeline -- either having the Cassandra instance be a part of the base chart, or a resource defined for test explicitly. This will help inform best practice as more services have similar needs.
Pieces missing from kask/charts to run integration tests in the deployment pipeline:
- WMF Cassandra image
- Kask chart update
- Update values.yaml
- Kask .pipeline/config.yaml update