msfroh commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-731354457
> Maybe we could add a simple unit test here that keeps re-indexing the same small set of docs with many unique fields, then calls IW.deleteAll, many times? Such that the test would OOME on trunk today, and then pass with the fix? Would that test take too long to OOME (how slow is the leak?)? If so we could mark it as @Slow. I'll give that a shot. With trunk, my code ran for about 12 hours and was still limping along (at a rate of ~5 documents/second, versus ~250 documents/second in the first minute). But I was doing some heavy analysis and flushing to disk. Adding a doc with a bunch of doc value fields, calling deleteAll(), and repeating, I think the problem should reproduce pretty quickly. I think I still need to trigger a `FieldInfos.finish()` to force it to allocate the large array. Alternatively, we would get an overflow on `lowestUnassignedFieldNumber` (though it looks like FieldNumbers will happily just give out negative field numbers in that case). Anyway, I'll try to come up with a test that breaks quickly. ---------------------------------------------------------------- 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. 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
