Kyle Renfrow

Hey I just saw that redis message which brings up another thing... I see you're setting the correlation_id there -- but the correlation_id isn't being set in amqplibconnector... could that...

@gjedeer haha yeah I get it. I meant comment wise. Yeah I know ETA is set to null. But you were previously doing a array_merge to overwrite the null value...

correction_id is new in protocol 2... I think i remember reading replaced task id? Yes, confirmed here: http://docs.celeryproject.org/en/latest/internals/protocol.html Just search that page for correlation_id

Could that have any impact on a task being created within a task (that has a properly defined task_route) ending up getting dropped for Wrong Destination? I was stumped so...

Oh, flower is buggy as hell with celery 4 (task protocol 2) also because mochiweb is buggy because of it. hopefully these other projects mature quickly.

More info on correlation_id. http://stackoverflow.com/questions/20184755/practical-examples-of-how-correlation-id-is-used-in-messaging

What's the unit tests testing though? Like I said, I was able to get celery4 branch to create a task. I was wrong initially. BUT, that task spawns more tasks...

Are you able to do that? You should be able to test easily... es: ```` @app.task def mytask(): for i in range(1,10): othertask.apply_async() ```` $celery->posttask() creates the initial task successfully,...

NOTE: i am NOT using any default celery values for exchange/queue/key

@gjedeer sure thing... maybe i'll beat you too it. Want me to create a sep issue to write a unit test for ETA as well?