Major Description
It would be great to introduce some sort of infrastructure in our Java test to expose data races (something similar to TSAN for C++ tests). As it turns out, there might be many bugs lurking in the code for years: for example, take a look at KUDU-3277.
It might make sense to start with something really simple, but allowing to trigger scheduling anomalies in the JVM or such (e.g., something similar to --stress_cpu_threads in C++ tests).
This might be relevant in this context: https://openjdk.java.net/jeps/8208520.