Hi Sasaki, Thanks for the KIP. I think the timeout controls the maximum allowed time that the consumer will block for the next record. Maybe the meaning would be clearer with the more concise name `--timeout`? That also fits with the old consumer which overrides the `consumer.timeout.ms` property.
By the way, it seems like the default value was intentionally set low for both the old and new consumers, but I'm not sure of the reason. We could leave the default as it is if we want to be safe, but increasing it seems ok to me. Perhaps we could start a little lower, though, say 10 seconds? In any case, we should make it clear to the user that the timeout was reached. It's surprising to see only the incomplete reported results following a timeout. Thanks, Jason On Wed, Apr 4, 2018 at 4:37 AM, Sasaki Toru <sasaki...@oss.nttdata.com> wrote: > Hello everyone, > > I would like to start a discussion for KIP 278. Cloud you please give > comments and advice ? > <https://cwiki.apache.org/confluence/display/KAFKA/KIP-278+- > +Add+version+option+to+Kafka%27s+commands> > > JIRA ticket and Pull Request are bellow: > <https://issues.apache.org/jira/browse/KAFKA-2061> > <https://github.com/apache/kafka/pull/639> > > > Many thanks, > > Sasaki > > -- > Sasaki Toru(sasaki...@oss.nttdata.com) NTT DATA CORPORATION > >
