You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@kafka.apache.org by chaitanya puchakayala <ch...@gmail.com> on 2017/05/04 05:24:36 UTC

What is the impact of setting the consumer config max.poll.interval.ms to high value

What will be the impact of setting the consumer config max.poll.interval.ms
to 10 or 15 mins?

How will impact nodes joining , nodes leaving, nodes crashing.?

Chaitanya
-- 
Chaitanya

Re: What is the impact of setting the consumer config max.poll.interval.ms to high value

Posted by "Matthias J. Sax" <ma...@confluent.io>.
If your consumer fails (ie. whole process dies) setting the value high
is not a problem (because the heartbeat thread dies, too, and the
failure will be detected quickly).

It's only a problem if you "main processing thread" dies (and everything
else is still up and running), or if you main processing thread has a
bug and ends up in an infinite loop not making any progress anymore. For
this case, the problem would only be detected late (ie, after 10/15
minutes).

Besides this late detention issue, there is no problem setting this
value high. For example, Streams API, sets this value to infinite (ie,
Integer.MAX_VALUE) -- even if I would not recommend this value for a
plain consumer application!

See details in the KIP:
https://cwiki.apache.org/confluence/display/KAFKA/KIP-62%3A+Allow+consumer+to+send+heartbeats+from+a+background+thread



-Matthias

On 5/3/17 10:24 PM, chaitanya puchakayala wrote:
> What will be the impact of setting the consumer config max.poll.interval.ms
> to 10 or 15 mins?
> 
> How will impact nodes joining , nodes leaving, nodes crashing.?
> 
> Chaitanya
>