On 22 May 2017 at 16:09, Guozhang Wang <wangg...@gmail.com> wrote: > For > that issue I'd suspect that there is a network issue, or maybe the network > is just saturated already and the heartbeat request / response were not > exchanged in time between the consumer and the broker, or the sockets being > dropped because of socket limit. Under this cases not all consumers may be > affected, but since the associated issue is from "AbstractCoordinator" > class which is part of the consumer client, I'd still be surprised if it is > actually due to Streams itself with the same consumer config settings, but > not to consumers. >
Yes. This is the conclusion that even we are coming to after further investigation. But didn't want to post it here until we were sure. We are experimenting with increasing the default timeouts, particularly hearbeat.interval.ms and session.timeout.ms. So far, the things have been running fine. But we will let it run for a few more days before closing this issue.
