Hi all,

I am noticing a strange behavior in one of my topology. The configuration is as 
follows:

-          Using RabbitMQ for publishing/consuming messages

-          Single topology with 2 workers running 4 threads each (Using 
multilang protocol)

I log messages every time a message gets published to rabbit queue so I queued 
a message just once. But I noticed 4 processes trying to process the same 
message within a gap of 60 milli seconds.

I know only during some kind of failures Storm can replay messages based on 
certain configurations. I would expect to have some time gap in those scenario 
(like message timeouts interval or something) But is the above scenario 
possible at all that multiple threads process same message within 60 millisec? 
If yes, how?

Thanks,
Ganesh

Reply via email to