uschindler commented on issue #15068: URL: https://github.com/apache/lucene/issues/15068#issuecomment-3211944928
Hi @ayinresh, this should be the case, but I am not 100% sure if it still might happen that the close is rejected by the JVM after the ReferenceManager has decided to close because maybe the other thread may still not have enetered a safepoint to make sure it is really done with accessing the already closed IndexInput? I am not 100% sure about this. I will try to create a reproducer what I figured out in the opened issue (we have a test for this that I can beef up to use a refcounted arena). Basically I'd like to remove the extra check in the RefCountedArena, plain simple. There is no need to throw an IllegalStateException pro-actively and just hand over the closing to the undelying `Arena#close()`. In addition are you really sure that no other thread that maybe still getting stored fields after all other threads have released the searcher? RefrenceManager only ensures no _new_ threads can aquire a searcher, but nothing prevents other threads from using an open searcher that might have been released! -- 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
