how to properly use pyspark to send data to kafka

问题:

I'm trying to write a simple pyspark job, which would receive data from a kafka broker topic, did some transformation on that data, and put the transformed data on a different kafka broker topic.

I have the following code, which reads data from a kafka topic, but has no effect running sendkafka function:

from pyspark import SparkConf, SparkContext

from operator import add
import sys
from pyspark.streaming import StreamingContext
from pyspark.streaming.kafka import KafkaUtils
import json
from kafka import SimpleProducer, KafkaClient

def sendkafka(messages):
    kafka = KafkaClient("localhost:9092")
    producer = SimpleProducer(kafka)
    for message in messages:
        yield producer.send_messages('spark.out', message)

def main():
    sc = SparkContext(appName="PythonStreamingDirectKafkaWordCount")
    ssc = StreamingContext(sc, 5)

    brokers, topic = sys.argv[1:]
    kvs = KafkaUtils.createDirectStream(ssc, [topic], {"metadata.broker.list": brokers})
    parsed = kvs.map(lambda (key, value): json.loads(value))
    parsed.pprint()

    sentRDD = kvs.mapPartitions(sendkafka)
    sentRDD.count()

    ssc.start()
    ssc.awaitTermination()
if __name__ == "__main__":

   main()

What should I change, in order to make my sendkafka function to actually send data to the spark.out kafka topic?

回答1:

Here is the correct code, which reads from Kafka into Spark, and writes spark data back to a different kafka topic:

from pyspark import SparkConf, SparkContext
from operator import add
import sys
from pyspark.streaming import StreamingContext
from pyspark.streaming.kafka import KafkaUtils
import json
from kafka import SimpleProducer, KafkaClient
from kafka import KafkaProducer

producer = KafkaProducer(bootstrap_servers='localhost:9092')

def handler(message):
    records = message.collect()
    for record in records:
        producer.send('spark.out', str(record))
        producer.flush()

def main():
    sc = SparkContext(appName="PythonStreamingDirectKafkaWordCount")
    ssc = StreamingContext(sc, 10)

    brokers, topic = sys.argv[1:]
    kvs = KafkaUtils.createDirectStream(ssc, [topic], {"metadata.broker.list": brokers})
    kvs.foreachRDD(handler)

    ssc.start()
    ssc.awaitTermination()
if __name__ == "__main__":

   main()

The way to run this is:

spark-submit --jars spark-streaming-kafka-assembly_2.10-1.6.1.jar s.py localhost:9092 test