1.建立kafka的topic:

./kafka-topics.sh --create --zookeeper xxxx:2181,xxxx:2181 --replication-factor 3 --partitions 3 --topic test

2.pyspark上傳具有spark客戶端的節點

vim ttt.py

# encoding:utf-8

from pyspark import SparkContext
from pyspark import SparkConf
from pyspark.streaming import StreamingContext
from pyspark.streaming.kafka import KafkaUtils


def start():
    conf = SparkConf().set("spark.python.profile", "true").set("spark.io.compression.codec", "snappy")#.setMaster('local[*]')
    conf.setAppName('spark-test')
    sc = SparkContext(conf=conf)
    ssc=StreamingContext(sc,6)

    brokers="xxx:6667,xxx:6667,xxx:666"
    topic='test'
    kafkaStreams = KafkaUtils.createDirectStream(ssc,[topic],kafkaParams={"metadata.broker.list": brokers})
    result=kafkaStreams.map(lambda x:(x[1],1)).reduceByKey(lambda x, y: x + y)
    kafkaStreams.transform(storeOffsetRanges).foreachRDD(printOffsetRanges)
    result.pprint()
    ssc.start()
    ssc.awaitTermination()

offsetRanges = []

def storeOffsetRanges(rdd):
    global offsetRanges
    offsetRanges = rdd.offsetRanges()
    return rdd

def printOffsetRanges(rdd):
    for o in offsetRanges:
        print "%s %s %s %s %s" % (o.topic, o.partition, o.fromOffset, o.untilOffset,o.untilOffset-o.fromOffset)

if __name__ == '__main__':
    start()

3.zip ttt.py ./ttt.py

4.提交程式：
  spark-submit --master yarn-cluster --driver-memory 1g --executor-memory 1g --num-executors 1 --executor-cores 1 --jars spark-streaming-kafka-0-8-assembly_2.11-2.3.1.jar  --py-files ttt.zip ttt.py

5.新開客戶端 
./kafka-console-producer.sh --broker-list xxx:6667,xxx:6667,xxx:6667 --topic test

end

    有問題加QQ群：877769335

    或者用QQ掃描二維碼加群：