gemmellr edited a comment on pull request #3568: URL: https://github.com/apache/activemq-artemis/pull/3568#issuecomment-834579719
> @gemmellr can you expand on "since things dont need to the producer link rather than the parent session or connection" > > It looks like the large message file is "gathered" in the receiver, then handed off to the broker. The issue is when the send is aborted or errors out, in both cases, there is still a need for close in general to do cleanup. Shoudl have said "dont need to _*close*_ the producer", i.e a client can close a session, without first closing a producer on it. Or a connection without first closing the sessions on it, or their producers or consumers. Prior to your latest change here, I dont think closing the producer or session called the close method containing the cleanup call. Any cases that needed cleaned up perhaps only happened due to the registered callback thing (which is registered with a completely different object 'in the server' as opposed to 'in the amqp bits'). Now, after your change, an explicit session closure does look to call that, and explicit connection closure. Its not clear to me if it would be doing that on unexpected connection drop, havent seen that code. It perhaps did with the callback registered and doesnt without, or I don't think Clebert would have added it originally. -- 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: [email protected]
