If tlp-stress is built with JDK 11, then (apparently) the compiler optimizes away the code within map:

Run.kt

runners.parallelStream().map {
    it.populate(populate)
}.count()

Better (thanks @sebastiankonieczek) to use somthing like

runners.parallelStream().forEach {
    it.populate(populate)
}

,
otherwise the population of Cassandra tables does not take place.

PS_ building the tool with JDK8 works.

PPS_ thanks for the tool.