Hi Saeed,

There could be several reasons why things appear slow and it is difficult to 
say without knowing the exact details of the setup and the results you are 
observing.
One thing to check is the number of threads you have assigned to the Kafka 
Stream application. By default just one thread is used. Perhaps you want more 
(depending on number of cores you have). An example way to change that in your 
app:

props.put(StreamsConfig.NUM_STREAM_THREADS_CONFIG, 4 /* four threads */);

Thanks
Eno

> On 9 Jun 2016, at 18:08, Saeed Ansari <saeed.ans...@gmail.com> wrote:
> 
> Hi Eno,
> Thank you for the response. Actually I did not know it automatically
> assigns partitions to consumers. Now I have one  Kafkastream reading from
> 12 partitions, like below:
> Controller is an actor that I am sending the message to and then it creates
> child actors to send messages out.
> 
> builder.stream("test").foreach((x, y) -> {
>    controller.tell(y, controller.noSender());
> });
> 
> 
> The msg/sec rate I get from receiving messages to sending them out is
> really slow!
> 
> Do you think it is about how consume messages?
> 
> Thank you,
> Saeed
> 
> 
> 
> On Wed, Jun 8, 2016 at 3:08 AM, Eno Thereska <eno.there...@gmail.com> wrote:
> 
>> Hi Saeed,
>> 
>> Kafka Streams takes care of assigning partitions to consumers
>> automatically for you. You don't have to write anything explicit to do
>> that. See WordCountDemo.java as an example. Was there another reason you
>> wanted control over partition assignment?
>> 
>> Thanks
>> Eno
>> 
>>> On 7 Jun 2016, at 20:02, Saeed Ansari <saeed.ans...@gmail.com> wrote:
>>> 
>>> Hi,
>>> I have started a small Kafka streaming application. I need to assign
>>> partitions to consumers in a consumer group. I am not sure where to start
>>> and how to structure consumer groups in KafkaStreams.
>>> 
>>> I found that there is a StreamPartitionAssignor that can be added to
>>> config, but still I do not know how to configure it and what parameters
>> to
>>> config.
>>> 
>>> Any help is really appreciated.
>>> 
>>> Thank you,
>>> Saeed
>> 
>> 

Reply via email to