You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@storm.apache.org by 이승진 <sw...@navercorp.com> on 2015/09/08 07:37:15 UTC

Question regarding localOrShuffleGrouping

p{margin-top:0px;margin-bottom:0px;} I have a topology with 1 spout and 1 bolt, parallelism of each is set to 3, 3 workers for the topology.
 
After deploying, each worker has 1 spout and 1 bolt.
 
But although I use localOrShuffleGrouping, when I kill one worker process, entire topology stops for a while before rebalancing finishes.
 
In a worker log(of one worker which is not killed), there are many Netty connection errors and retries log, meanwhile process only tries to connect to worker which is dead.
 
Is it supposed to be like this or am I doing something wrong?