Kafka streams word count application

问题

I'm playing around with the kafka streaming API (Kakfa version: 0.10.2.0) trying to make a simple wordcount example work: Wordcount App gist. I'm running both producer and console consumer:

./kafka-console-producer.sh -topic input-topic --broker-list localhost:9092

./kafka-console-consumer.sh --topic output-topic --bootstrap-server localhost:9092 --from-beginning

start the application and everything seems to be working fine but when I type in some strings within the console producer, the consumer receives nothing at all. If I change the app to do a simple toUppercase on the input the consumer receives the stream (modified to upper case) fine:

//The following code works fine: val uppercasedWithMapValues: KStream[String, String] = textLines.mapValues(_.toUpperCase()) uppercasedWithMapValues.to("output-topic")

Does anyone know why I'm receiving nothing on the word-count example? Should I specify any serializer on the consumer? In my last test the console consumer processed the messages that I sent through the console but didn't show them, see below the output:

➜  bin ./kafka-console-consumer.sh \
           --topic output-topic \
           --bootstrap-server localhost:9092 \
           --from-beginning                                                                                
[2017-08-02 07:48:20,187]WARN Error while fetching metadata with correlation id 2 : 
{output-topic=LEADER_NOT_AVAILABLE} (org.apache.kafka.clients.NetworkClient)
[2017-08-02 07:48:20,197] WARN The following subscribed topics are not assigned 
to any members in the group console-consumer-91651 : [output-topic]  
(org.apache.kafka.clients.consumer.internals.ConsumerCoordinator)

^CProcessed a total of 7 messages

回答1:

KStream works because it doesn't use caching. For KTable you have to wait a bit, or set cache.max.bytes.buffering to 0 (but not in a production code!)

来源：https://stackoverflow.com/questions/45452685/kafka-streams-word-count-application