Is the grinder doing more work than required?

[pshipton]

Describe the bug I started a grinder using "NodesByIterations" on 5 machines. The parent job loads all the repos and downloads the JVM, only to create 5 child jobs that actually runs the testing. It seems loading the repos and downloading the JVM in the parent job is a waste of time. https://openj9-jenkins.osuosl.org/job/Grinder/1292/console

To Reproduce Steps to reproduce the behavior (or Grinder rerun link):

Expected behavior A clear and concise description of what you expected to happen.

Screenshots If applicable, add screenshots to help explain your problem.

Additional context Add any other context about the problem here.

[llxia]

I agree. We only need to download JVM to run TKG when PARALLEL=Dynamic. In other parallel cases (i.e., NodesByIterations, Subdir), we do not need to download JVM and clone test repos.

[llxia]

As @AdamBrousseau pointed out, there is another corner case where PARALLEL=Dynamic and NUM_MACHINES > 1, but the available machine is only 1. In this case, we should run it serially (not launching the child job).