Github user ghajos commented on the issue: https://github.com/apache/storm/pull/2660 @raghavgautam Thanks for the quick reaction! Unfortunately the newly introduced test passes without #2643, so it does not test the original issue. Since in this modification both the server and the client is closed gracefully the port won't stuck in TIME_WAIT state, no need to use SO_REUSEADDR to restart new instances. To test #2643 the server port should stuck in TIME_WAIT state and then a new server instance should be started successfully. I still think that it is out of scope to test an operating system level TCP setting in a Storm test. cc: @HeartSaVioR
---