Github user ssuchter commented on the issue:
https://github.com/apache/spark/pull/20697
Hm, that wasn't the answer I expected to hear. It leads me to another
question:
If we don't have developers using this script directly, what's the value of
trying to incorporate the running of the Kubernetes integration tests into this
program? Because of the AMPLab setup, we're going to have to have special
Jenkins jobs dedicated to the running of the K8s integration tests. This is
because not all hosts can run these integration tests, so therefore we have
special jobs that are tied to the specific hosts that have the right setup to
run the minikube VMs. So it's just going to be those jobs that call
dev/run-tests.py with the special non-default argument to run these tests.
Those jenkins jobs could just as easily run a different script.
An alternative way to take my argument is that we should delay the
integration into run-tests.py until after **all** of the AMPLab hosts can run
these tests. Since that's not today, it wouldn't need to be in this initial PR,
and could have its own PR.
Thoughts? Thanks, Sean
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
