AntonRoskvist commented on pull request #3858:
URL: https://github.com/apache/activemq-artemis/pull/3858#issuecomment-979776866


   No problem @gtully .
   
   Well, I'm thinking that if the queue is created by a joining consumer then 
the broker in question does not have any preexisting messages on that queue 
(because then the queue would already exist in that case) and since 
OFF_WITH_REDISTRIBUTION doesn't do any initial distribution the only case I 
could see where this could happen would be if you create the queue by sending a 
message to it, and then connect a consumer after... but if you cannot handle 
messages dispatched to another consumer for the same queue within a cluster 
then I don't know why you have multiple consumers to begin with? 
   
   The binding would also be created if a producer sends a message to a queue 
that up to that point did not exist on a broker... and probably if just 
administratively creating a queue, say through the console for instance.
   
   For my case the current behavior becomes an issue if within a cluster there 
are consumers connected to some brokers but not all and then messages are 
produced to a broker without a corresponding consumer. In this case no 
redistribution happens until at least one consumer disconnects and then 
reconnects. A somewhat similar scenario is outlined here that also should be 
covered by this change:
   https://issues.apache.org/jira/browse/ARTEMIS-3321


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to