Github user srdo commented on a diff in the pull request: https://github.com/apache/storm/pull/1924#discussion_r106687191 --- Diff: external/storm-kafka-client/src/main/java/org/apache/storm/kafka/spout/KafkaSpoutRetryExponentialBackoff.java --- @@ -54,21 +55,28 @@ private static class RetryEntryTimeStampComparator implements Serializable, Comparator<RetrySchedule> { @Override public int compare(RetrySchedule entry1, RetrySchedule entry2) { - return Long.valueOf(entry1.nextRetryTimeNanos()).compareTo(entry2.nextRetryTimeNanos()); + int result = Long.valueOf(entry1.nextRetryTimeNanos()).compareTo(entry2.nextRetryTimeNanos()); + + if(result == 0) { + //TreeSet uses compareTo instead of equals() for the Set contract + //Ensure that we can save two retry schedules with the same timestamp --- End diff -- It's probably impossible in real life if System.nanoTime has sufficiently high resolution (not guaranteed to be any better than System.currentTimeMillis), but it can happen in tests that use simulated time. There's still potential for collisions if two retry schedules happen to have the same hashCode and timestamp, but that's very unlikely. I mostly made the change for the benefit of the tests, but it doesn't hurt to reduce the chance of collision in general.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---