Description of problem: After upgrading from 6.4.4 to 6.4.11 (hornet 2.3.25.SP14-redhat-1), we are seeing the following: 22:11:17,162 WARN [org.hornetq.core.server] (Thread-24 (HornetQ-server-HornetQServerImpl::serverUUID=8fda6679-eca6-11e6-82b0-7b59b940a711-756430729)) HQ222015: Internal error! Delivery logic has identified a non delivery and still handled a consumer! It's on single node machine, and there is no failover, as discussed here: https://bugzilla.redhat.com/show_bug.cgi?id=1158920 We do have 40/50k load (http requests, where each triggers JMS transaction), per hour. We do a get a few of those, not lot's of them. Last night I noticed about 10 or 15 of those, after a 730k load test Version-Release number of selected component (if applicable): hornet 2.3.25.SP14-redhat-1 How reproducible: reproducing is hard, it's happening (randomly) during load-tests Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
not sure I can really provide more, as said, it happened/happens a few times, in larger load
From Miroslav Novak: I was tracking a message for which: "Internal error! Delivery logic has identified a non delivery and still handled a consumer!" is logged. It seems to be harmless as this message was delivered to consumer in the end. I can see that this message is successfully acked. The logic is quite complex but it seems that server tried to deliver message to number of consumers on queue jms.queue.BatchLoadedQueue. But those consumers had message selectors and refused this message as it did not match their selector. It seems that last consumer ,to which queue tried to deliver message, consumed this message and then this warning got logged. It seems be problem in the if(...) statement when this warning should be logged.
HORNETQ-1563
Hi, As per my investigation, this doesn't affect the messaging at all. It's just an incorrect calculation on the logic. I don't think we need to worry about it. Howard