Python操作Kafka的简单示例

发表于 分类于 本文字数： 3.2k 阅读时长 ≈ 3 分钟

了解了Kafka的术语以后，尝试使用Python操作Kafka，下面是一个简单的demo，为啥不用Java呢…因为没学过…等我哪天研究研究👀

客户端

查找了一下，Kafka的Python客户端常用的基本就三种:Kafka-python、pykafka、confluent-kafka-python，也有人对这三种客户端做了Benchmark: Python Kafka Client Benchmarking，但是感觉Kafka-python相关的文章多一点，就先用这个吧，等熟悉下操作，看情况再说要不要更换其他的客户端。

安装Kafka-python:

1
pip install Kafka-python

创建Topic

仍然在单机服务上学习，先创建一个具有两个分区的Topic: 

1
2
3
$ bin/kafka-topics.sh --create --bootstrap-server 0.0.0.0:9092 --replication-factor 1 --partitions 2 --topic Hello
$ bin/kafka-topics.sh --list --bootstrap-server 0.0.0.0:9092
Hello

Producer

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
from kafka import KafkaProducer
from kafka.errors import KafkaError


def main():
    producer = KafkaProducer(bootstrap_servers=["top:9092"])

    for _ in range(100):
        future = producer.send("Hello", b'msg')
        try:
            record_metadata = future.get(timeout=10)
            print(record_metadata)
        except KafkaError as e:
            print(e)
    producer.flush()


if __name__ == '__main__':
    main()

Consumer

看过书后，我这里采用异步与同步组合的方式手动提交偏移量，避免提交时产生问题。 

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
from kafka import KafkaConsumer
from kafka.structs import TopicPartition
import time


class Consumer:
    def __init__(self):
        self.consumer = KafkaConsumer(
            group_id="my",
            auto_offset_reset='earliest',
            enable_auto_commit=False,
            bootstrap_servers=["top:9092"]
        )

    def consumer_data(self, topic, partition):
        my_partition = TopicPartition(topic=topic, partition=partition)
        self.consumer.assign([my_partition])

        print(f"consumer start position: {self.consumer.position(my_partition)}")

        try:
            while True:
                poll_num = self.consumer.poll(timeout_ms=1000, max_records=5)
                if poll_num == {}:
                    print("consumer poll is empty, will exit")
                    exit(1)
                for key, record in poll_num.items():
                    for message in record:
                        print(
                            f"{message.topic}:{message.partition}:{message.offset}: key={message.key} value={message.value}")

                try:
                    self.consumer.commit_async()
                    time.sleep(5)
                except Exception as e:
                    print(e)
        except Exception as e:
            print(e)
        finally:
            try:
                self.consumer.commit()
            finally:
                self.consumer.close()


def main():
    topic = "Hello"
    partition = 0
    my_consumer = Consumer()
    my_consumer.consumer_data(topic, partition)


if __name__ == '__main__':
    main()

结果

上面的代码中Consumer中通过group_id的方式可以让多个Consumer消费同一Topic的不同分区，以消费Hello主题的0分区为例，运行Producer代码向Hello中送入数据，再启动Consumer代码可以看到: 

1
2
3
4
5
6
7
8
9
> python kz_consumer.py
consumer start position: 0
Hello:0:0: key=None value=b'msg'
Hello:0:1: key=None value=b'msg'
Hello:0:2: key=None value=b'msg'
Hello:0:3: key=None value=b'msg'
Hello:0:4: key=None value=b'msg'
Hello:0:5: key=None value=b'msg'
...

同时通过终端可以查看到该消费组的LAG信息: 

1
2
3
4
5
6
7
8
9
10
11
$ bin/kafka-consumer-groups.sh --bootstrap-server 0.0.0.0:9092 --group my --describe
Consumer group 'my' has no active members.

GROUP           TOPIC           PARTITION  CURRENT-OFFSET  LOG-END-OFFSET  LAG             CONSUMER-ID     HOST            CLIENT-ID
my              Hello           0          10              53              43              -               -               -

$ bin/kafka-consumer-groups.sh --bootstrap-server 0.0.0.0:9092 --group my --describe
Consumer group 'my' has no active members.

GROUP           TOPIC           PARTITION  CURRENT-OFFSET  LOG-END-OFFSET  LAG             CONSUMER-ID     HOST            CLIENT-ID
my              Hello           0          53              53              0               -               -               -

到这一个简单的demo完事，要继续去看看官方文档具体的API怎么用…

参考

1、https://kafka-python.readthedocs.io/en/master/usage.html
2、Kafka权威指南