How many crashes do we do in each run? Do they happen approximately every 7500 messages? If so, it may make sense to use 15000 keys so that there are we can hit the "insert during crash case" atleast twice per run, and then are guaranteed to test the update case the rest of the time.
I think there should be a chance of hitting this type of bug on every run. Then, over time, a test will fail and we can have a chance of finding the bug at that time. [ Full content available at: https://github.com/apache/kafka/pull/5640 ] This message was relayed via gitbox.apache.org for [email protected]
