Description
For the DAC to work as expected, ordering is important. If a secret is submitted to the API first, it should be there for the DAC to look at when a referencing CouchbaseCluster comes along.
To be technically correct, we do synchronous reads from the Kubernetes API for each CR update. This isn't too bad, as 99% of the time we aren't performing updates to the CR. If the cluster got stuck somehow, and was subjected to high numbers of mutations, then we'd get a corresponding high number of API calls. Now this shouldn't be too onerous, but it would be a good ass coverage to implement caching so state is only updated when it changes, and we avoid a lot of those reads.
Problem is, you now have asynchronous updates to worry about, and your 4th dimension gets more interesting... this is to test:
- Does it work out of the box?
- If not does it work with some retries?
- If so what values should we use?