1. 安装pykafka

pip install pykafka

    2. 生产者

from pykafka import KafkaClient
 
def get_kafka_producer(hosts, topics):
    client = KafkaClient(hosts=hosts)
    print(client.topics)
    topic = client.topics[topics]
    producer = topic.get_producer()
    return producer

测试

hosts = '192.168.20.203:9092,192.168.20.204:9092,192.168.20.205:9092'
topics = "test_kafka_topic"
producer = get_kafka_producer(hosts, topics)   
for i in range(10):
    msg = "test message " + str(i)
    # msg = bytes(msg, encoding='utf-8')    
    # producer.produce(msg)
    producer.produce(msg.encode())
producer.stop()

3. 消费者

def get_kafka_consumer(hosts, topics):
    client = KafkaClient(hosts=hosts)
    topic=client.topics[topics]
    consumer = topic.get_balanced_consumer(consumer_group='test_kafka_topic', auto_commit_enable=True,
zookeeper_connect='192.168.20.201:2181,192.168.20.202:2181,192.168.20.203:2181',
managed=True, consumer_timeout_ms=1000)
    # managed=True，即使用新式reblance分区方法，不需zk；managed=False则需通过zk来实现reblance      
    return consumer

测试

hosts = '192.168.20.203:9092,192.168.20.204:9092,192.168.20.205:9092'
topics = "test_kafka_topic"
consumer = get_kafka_consumer(hosts, topics)   
for msg in consumer:
    print(msg)
    if msg is not None:
        print(msg.offset)
        print(msg.value)