cdbartholomew commented on issue #5969: Acked messages unexpectedly redelivered when others are negatively acked URL: https://github.com/apache/pulsar/issues/5969#issuecomment-570642173 @jerrypeng @sijie @gmethvin I agree that redelivering batches is OK in most cases. Redelivery of unacked/nacked messages is the exception, not the rule, but when it occurs it should behave in a way that makes sense to the user of the API. It's not that hard for the client to filter out the already acked messages from the batch. It is already keeping track of this so it knows when it can ack the batch back to broker. It's just a matter of using this information to filter received messages. That approach doesn't cover all failure cases (ex if client restarts), but in the case where the application is NACKing a message, it would give reasonable behavior. Plus it will reduce the number of duplicate messages that get sent to the client when using batch messages. @zzzming and I are happy to work on a PR for this if everyone agrees this is the right approach.
---------------------------------------------------------------- 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] With regards, Apache Git Services
