In case Helm test for a given HelmRelease fails, the only response that can be gathered from the HelmRelease is the name of the test job/pod, but it doesn't tell anything about the testcase that failed inside the pod.

One apporach to work around this is to compile each testcase into a different pod, but that has 2 effects:

1. Pods are executed 1 by 1. Having 10+ test cases can mean that most of the time during a helm install or upgrade is spent with waiting for test pods to be scheduled and started, which can be problematic in case some traffic needs to be diverted away from the application under upgrade.
2. Helm tests stop at first failure. This can give incomplete results of the testing, and might mean multiple retries are needed to resolve the issues 1 by 1.

Would it make sense that in case of a failed test pod, the pod's logs, similarly to helm test --logs collect the logs? Probably it would be too much, and could be tail-ed to the last 100 lines or something similar, as most test runners print a summary at the end of their execution.