vigyasharma commented on PR #633: URL: https://github.com/apache/lucene/pull/633#issuecomment-1140109855
@mikemccand Thanks for beasting it and uncovering this issue! I tried to do the same on my box but haven't had any luck so far. Any ideas on how I can make it more likely to happen? How many cores (and threads) did you run this on? Side Note: it might be helpful to have a gradle parameter for thread count for concurrency tests. Maybe something that goes with the [Concurrent](https://issues.apache.org/jira/browse/LUCENE-10566) test group? I'm not sure if there's one for this already (the test seems hardcoded to `int numThreads = TEST_NIGHTLY ? 5 : 2;`. I tried reading through code to figure this out. I suspect this is because the `synchronized AddIndexesMergeSource::getNextMerge()` function is really locking on the inner class object, and not on IndexWriter. This works fine for `pendingAddIndexesMerges` because that is an inner class object, but probably hits race conditions for `runningMerges` which belongs to `IndexWriter`. I could either wrap the runningMerges update with a `synchronized (IndexWriter.this) {}`, or make `runningMerges` a `synchronizedSet`. I like the second approach as it automatically fixes this at all other places. What do you think? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
