msokolov commented on issue #11896: URL: https://github.com/apache/lucene/issues/11896#issuecomment-1301439989
I have an idea we can fix by reducing the result size we request to a smaller percentage of the total index size. Currently it's allowed to be as much as 50% of the total index size, and we know that there can be scenarios where smaller graphs are disconnected, leading to the search getting trapped in a disconnected component. I'd like to be able to demonstrate some improvement though, and althjough I could reproduce (thanks for the info, @rcmuir) and see a fix when reducing the requested k, when I tried to "run until failure" in IntelliJ I was never able to reproduce a failure (ran 36000 times). I wonder if IntelliJ is not randomizing enough. I'll try again with beast I guess -- 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
