Kafka和Spring集成实践

目录 [−]

  1. 安装Zookeeper
  2. 安装Kafka
  3. 创建一个Spring项目
    1. 使用Producer API发送消息到Kafka
    2. 使用Kafka High Level API接收消息
    3. 使用spring-integration-kafka发送消息
    4. 使用spring-integration-kafka接收消息

本文以单机的环境演示如何将Kafka和Spring集成。
单机的环境最容易搭建, 并且只需在自己的PC上运行即可, 不需要很多的硬件环境,便于学习。 况且,本文的目的不是搭建ZooKeeper的集群环境, 而是重点介绍Kafka和Spring的应用。
具体的软件环境如下:

  • OS: CentOS 6.4
  • Zookepper: zookeeper-3.4.6
  • Kafka: kafka_2.9.1-0.8.2-beta
  • Java: JDK 1.7.0_45-b18
  • Spring:4.0.6

本例子在我的这个环境中运行正常, 全部代码可以到 github 下载。

本文所有的操作系统用户都是root。 实际产品中可能安全标准需要特定的用户如zookeeper, kafka等。

安装Zookeeper

首先下载解压zookeeper,选择合适的镜像站点以加快下载速度。
我们可以将zookeeper加到系统服务中, 增加一个/etc/init.d/zookeeper文件。

1
2
3
4
cd /opt
wget http://apache.fayea.com/apache-mirror/zookeeper/zookeeper-3.4.6/zookeeper-3.4.6.tar.gz
tar zxvf zookeeper-3.4.6.tar.gz
vi /etc/init.d/zookeeper

https://raw.githubusercontent.com/apache/zookeeper/trunk/src/packages/rpm/init.d/zookeeper文件的内容拷贝到这个文件,修改其中的运行zookeeper的用户以及zookeeper的文件夹位置。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
......
start() {
echo -n $"Starting $desc (zookeeper): "
daemon --user root /opt/zookeeper-3.4.6/zkServer.sh start
RETVAL=$?
echo
[ $RETVAL -eq 0 ] && touch /var/lock/subsys/zookeeper
return $RETVAL
}
stop() {
echo -n $"Stopping $desc (zookeeper): "
daemon --user root /opt/zookeeper-3.4.6/zkServer.sh stop
RETVAL=$?
sleep 5
echo
[ $RETVAL -eq 0 ] && rm -f /var/lock/subsys/zookeeper $PIDFILE
}
......
1
2
chmod 755 /etc/init.d/zookeeper
service zookeeper start

如果你不想加到服务,也可以直接运行zookeeper。

1
/opt/zookeeper-3.4.6/zkServer.sh start

安装Kafka

从合适的镜像站点下载最新的kafka并解压。

1
2
3
wget http://apache.01link.hk/kafka/0.8.2-beta/kafka_2.9.1-0.8.2-beta.tgz
tar zxvf kafka_2.9.1-0.8.2-beta.tgz
cd kafka_2.9.1-0.8.2-beta

启动Kafka:

1
bin/kafka-server-start.sh config/server.properties

创建一个test的topic:

1
bin/kafka-topics.sh --create --zookeeper localhost:2181 --replication-factor 1 --partitions 1 --topic test

可以利用kafka的命令启动一个生产者和消费者试验一下:

1
2
3
> bin/kafka-console-producer.sh --broker-list localhost:9092 --topic test
This is a message
This is another message
1
2
3
> bin/kafka-console-consumer.sh --zookeeper localhost:2181 --topic test --from-beginning
This is a message
This is another message

更多的介绍可以查看我翻译整理的 Kafka快速入门

创建一个Spring项目

以上的准备环境完成,让我们开始创建一个项目。
以前我写过一篇简单介绍: Spring 集成 Kafka.
spring-integration-kafka这个官方框架我就不介绍了。 我们主要使用它做集成。

首先我们先看一下使用Kafka自己的Producer/Consumer API发送/接收消息的例子。

使用Producer API发送消息到Kafka

OK,现在我们先看一个使用Kafka 自己的producer API发送消息的例子:

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
public class NativeProducer {
public static void main(String[] args) {
String topic= "test";
long events = 100;
Random rand = new Random();
Properties props = new Properties();
props.put("metadata.broker.list", "localhost:9092");
props.put("serializer.class", "kafka.serializer.StringEncoder");
props.put("request.required.acks", "1");
ProducerConfig config = new ProducerConfig(props);
Producer<String, String> producer = new Producer<String, String>(config);
for (long nEvents = 0; nEvents < events; nEvents++) {
String msg = "NativeMessage-" + rand.nextInt() ;
KeyedMessage<String, String> data = new KeyedMessage<String, String>(topic, nEvents + "", msg);
producer.send(data);
}
producer.close();
}
}

这个例子中首先初始化Producer对象,指定相应的broker和serializer, 然后发送100个字符串消息给Kafka。

运行mvn package编译代码,执行查看结果:

1
java -cp target/lib/*:target/spring-kafka-demo-0.2.0-SNAPSHOT.jar com.colobu.spring_kafka_demo.NativeProducer

上面的消费者控制台窗口会打印出收到的消息:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
......
NativeMessage--1645592376
NativeMessage-534168193
NativeMessage--1899432197
NativeMessage-1642480773
NativeMessage--911267171
NativeMessage-251458151
NativeMessage--55710397
NativeMessage-455515562
NativeMessage-1108982916
NativeMessage--1710296834
NativeMessage-2102648373
NativeMessage-499979365
NativeMessage--1200107003
NativeMessage-1184836299
NativeMessage--1161123005
NativeMessage-912582115
NativeMessage--1557863408
NativeMessage--1036456356
......

使用Kafka High Level API接收消息

用High level Consumer API接收消息,

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
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Properties;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import kafka.consumer.ConsumerConfig;
import kafka.consumer.ConsumerIterator;
import kafka.consumer.KafkaStream;
import kafka.javaapi.consumer.ConsumerConnector;
public class NativeConsumer {
private final ConsumerConnector consumer;
private final String topic;
private ExecutorService executor;
public NativeConsumer(String a_zookeeper, String a_groupId, String a_topic) {
consumer = kafka.consumer.Consumer.createJavaConsumerConnector(createConsumerConfig(a_zookeeper, a_groupId));
this.topic = a_topic;
}
public void shutdown() {
if (consumer != null)
consumer.shutdown();
if (executor != null)
executor.shutdown();
}
public void run(int a_numThreads) {
Map<String, Integer> topicCountMap = new HashMap<String, Integer>();
topicCountMap.put(topic, new Integer(a_numThreads));
Map<String, List<KafkaStream<byte[], byte[]>>> consumerMap = consumer.createMessageStreams(topicCountMap);
List<KafkaStream<byte[], byte[]>> streams = consumerMap.get(topic);
// now launch all the threads
//
executor = Executors.newFixedThreadPool(a_numThreads);
// now create an object to consume the messages
//
int threadNumber = 0;
for (final KafkaStream stream : streams) {
executor.submit(new ConsumerTest(stream, threadNumber));
threadNumber++;
}
}
private static ConsumerConfig createConsumerConfig(String a_zookeeper, String a_groupId) {
Properties props = new Properties();
props.put("zookeeper.connect", a_zookeeper);
props.put("group.id", a_groupId);
props.put("zookeeper.session.timeout.ms", "400");
props.put("zookeeper.sync.time.ms", "200");
props.put("auto.commit.interval.ms", "1000");
return new ConsumerConfig(props);
}
public static void main(String[] args) {
String zooKeeper = "localhost:2181";
String groupId = "mygroup";
String topic = "test";
int threads = 1;
NativeConsumer example = new NativeConsumer(zooKeeper, groupId, topic);
example.run(threads);
try {
Thread.sleep(10000);
} catch (InterruptedException ie) {
}
//example.shutdown();
}
}
class ConsumerTest implements Runnable {
private KafkaStream m_stream;
private int m_threadNumber;
public ConsumerTest(KafkaStream a_stream, int a_threadNumber) {
m_threadNumber = a_threadNumber;
m_stream = a_stream;
}
public void run() {
ConsumerIterator<byte[], byte[]> it = m_stream.iterator();
while (it.hasNext())
System.out.println("Thread " + m_threadNumber + ": " + new String(it.next().message()));
System.out.println("Shutting down Thread: " + m_threadNumber);
}
}

在生产者控制台输入几条消息,可以看到运行这个例子的控制台可以将这些消息打印出来。

教程的代码中还包括一个使用Simple Consumer API接收消息的例子。 因为spring-integration-kafka不支持这种API,这里也不列出对比代码了。

使用spring-integration-kafka发送消息

Outbound Channel Adapter用来发送消息到Kafka。 消息从Spring Integration Channel中读取。 你可以在Spring application context指定这个channel。
一旦配置好这个Channel,就可以利用这个Channel往Kafka发消息。 明显地,Spring Integration特定的消息发送给这个Adaptor,然后发送前在内部被转为Kafka消息。当前的版本要求你必须指定消息key和topic作为头部数据 (header),消息作为有载荷(payload)。
例如

1
2
3
4
5
6
final MessageChannel channel = ctx.getBean("inputToKafka", MessageChannel.class);
channel.send(
MessageBuilder.withPayload(payload) //设置有效载荷
.setHeader("messageKey", "key") //指定key
.setHeader("topic", "test").build()); /指定topic/

实际代码如下:

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
import java.util.Random;
import org.springframework.context.support.ClassPathXmlApplicationContext;
import org.springframework.integration.support.MessageBuilder;
import org.springframework.messaging.MessageChannel;
public class Producer {
private static final String CONFIG = "/context.xml";
private static Random rand = new Random();
public static void main(String[] args) {
final ClassPathXmlApplicationContext ctx = new ClassPathXmlApplicationContext(CONFIG, Producer.class);
ctx.start();
final MessageChannel channel = ctx.getBean("inputToKafka", MessageChannel.class);
for (int i = 0; i < 100; i++) {
channel.send(MessageBuilder.withPayload("Message-" + rand.nextInt()).setHeader("messageKey", String.valueOf(i)).setHeader("topic", "test").build());
}
try {
Thread.sleep(100000);
} catch (InterruptedException e) {
e.printStackTrace();
}
ctx.close();
}
}

Spring 配置文件:

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
<?xml version="1.0" encoding="UTF-8"?>
<beans xmlns="http://www.springframework.org/schema/beans"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:int="http://www.springframework.org/schema/integration"
xmlns:int-kafka="http://www.springframework.org/schema/integration/kafka"
xmlns:task="http://www.springframework.org/schema/task"
xsi:schemaLocation="http://www.springframework.org/schema/integration/kafka http://www.springframework.org/schema/integration/kafka/spring-integration-kafka.xsd
http://www.springframework.org/schema/integration http://www.springframework.org/schema/integration/spring-integration.xsd
http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd
http://www.springframework.org/schema/task http://www.springframework.org/schema/task/spring-task.xsd">
<int:channel id="inputToKafka">
<int:queue/>
</int:channel>
<int-kafka:outbound-channel-adapter id="kafkaOutboundChannelAdapter"
kafka-producer-context-ref="kafkaProducerContext"
auto-startup="false"
channel="inputToKafka"
order="3"
>
<int:poller fixed-delay="1000" time-unit="MILLISECONDS" receive-timeout="0" task-executor="taskExecutor"/>
</int-kafka:outbound-channel-adapter>
<task:executor id="taskExecutor" pool-size="5" keep-alive="120" queue-capacity="500"/>
<bean id="producerProperties"
class="org.springframework.beans.factory.config.PropertiesFactoryBean">
<property name="properties">
<props>
<prop key="topic.metadata.refresh.interval.ms">3600000</prop>
<prop key="message.send.max.retries">5</prop>
<prop key="serializer.class">kafka.serializer.StringEncoder</prop>
<prop key="request.required.acks">1</prop>
</props>
</property>
</bean>
<int-kafka:producer-context id="kafkaProducerContext"
producer-properties="producerProperties">
<int-kafka:producer-configurations>
<int-kafka:producer-configuration broker-list="localhost:9092"
topic="test"
compression-codec="default"/>
</int-kafka:producer-configurations>
</int-kafka:producer-context>
</beans>

int:channel是配置Spring Integration Channel, 此channel基于queue。
int-kafka:outbound-channel-adapter是outbound-channel-adapter对象, 内部使用一个线程池处理消息。关键是kafka-producer-context-ref
int-kafka:producer-context配置producer列表,要处理的topic,这些Producer最终要转换成Kafka的Producer。

producer的配置参数如下:

1
2
3
4
5
6
7
8
9
10
11
12
broker-list List of comma separated brokers that this producer connects to
topic Topic name or Java regex pattern of topic name
compression-codec Compression method to be used. Default is no compression. Supported compression codec are gzip and snappy.
Anything else would result in no compression
value-encoder Serializer to be used for encoding messages.
key-encoder Serializer to be used for encoding the partition key
key-class-type Type of the key class. This will be ignored if no key-encoder is provided
value-class-type Type of the value class. This will be ignored if no value-encoder is provided.
partitioner Custom implementation of a Kafka Partitioner interface.
async True/False - default is false. Setting this to true would make the Kafka producer to use
an async producer
batch-num-messages Number of messages to batch at the producer. If async is false, then this has no effect.

value-encoder 和key-encoder可以是其它实现了Kafka Encoder接口的Bean。同样partitioner也是实现了Kafka的Partitioner接口的Bean。
一个Encoder的例子:

1
2
3
<bean id="kafkaEncoder" class="org.springframework.integration.kafka.serializer.avro.AvroSpecificDatumBackedKafkaEncoder">
<constructor-arg value="com.company.AvroGeneratedSpecificRecord" />
</bean>

Spring Integration Kafka 也提供了个基于Avro的Encoder。 Avro也是Apache的一个项目, 在大数据处理时也是一个常用的序列化框架。
不指定Encoder将使用Kafka缺省的Encoder (kafka.serializer.DefaultEncoder, byte[] --> same byte[])。

producerProperties可以用来设置配置属性进行调优。配置属性列表请参考 http://kafka.apache.org/documentation.html#producerconfigs

使用spring-integration-kafka接收消息

同样的原理实现一个消费者:

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
56
57
58
59
60
61
package com.colobu.spring_kafka_demo;
import java.util.Collection;
import java.util.HashMap;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.Random;
import java.util.Set;
import java.util.concurrent.ConcurrentHashMap;
import org.slf4j.LoggerFactory;
import org.springframework.context.support.ClassPathXmlApplicationContext;
import org.springframework.integration.channel.QueueChannel;
import org.springframework.messaging.Message;
import ch.qos.logback.classic.Level;
public class Consumer {
private static final String CONFIG = "/consumer_context.xml";
private static Random rand = new Random();
@SuppressWarnings({ "unchecked", "unchecked", "rawtypes" })
public static void main(String[] args) {
ch.qos.logback.classic.Logger rootLogger = (ch.qos.logback.classic.Logger)LoggerFactory.getLogger(ch.qos.logback.classic.Logger.ROOT_LOGGER_NAME);
rootLogger.setLevel(Level.toLevel("info"));
final ClassPathXmlApplicationContext ctx = new ClassPathXmlApplicationContext(CONFIG, Consumer.class);
ctx.start();
final QueueChannel channel = ctx.getBean("inputFromKafka", QueueChannel.class);
Message msg;
while((msg = channel.receive()) != null) {
HashMap map = (HashMap)msg.getPayload();
Set<Map.Entry> set = map.entrySet();
for (Map.Entry entry : set) {
String topic = (String)entry.getKey();
System.out.println("Topic:" + topic);
ConcurrentHashMap<Integer,List<byte[]>> messages = (ConcurrentHashMap<Integer,List<byte[]>>)entry.getValue();
Collection<List<byte[]>> values = messages.values();
for (Iterator<List<byte[]>> iterator = values.iterator(); iterator.hasNext();) {
List<byte[]> list = iterator.next();
for (byte[] object : list) {
String message = new String(object);
System.out.println("\tMessage: " + message);
}
}
}
}
try {
Thread.sleep(100000);
} catch (InterruptedException e) {
e.printStackTrace();
}
ctx.close();
}
}

Spring的配置文件如下:

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
<?xml version="1.0" encoding="UTF-8"?>
<beans xmlns="http://www.springframework.org/schema/beans"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:int="http://www.springframework.org/schema/integration"
xmlns:int-kafka="http://www.springframework.org/schema/integration/kafka"
xmlns:task="http://www.springframework.org/schema/task"
xsi:schemaLocation="http://www.springframework.org/schema/integration/kafka http://www.springframework.org/schema/integration/kafka/spring-integration-kafka.xsd
http://www.springframework.org/schema/integration http://www.springframework.org/schema/integration/spring-integration.xsd
http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd
http://www.springframework.org/schema/task http://www.springframework.org/schema/task/spring-task.xsd">
<int:channel id="inputFromKafka">
<int:queue/>
</int:channel>
<int-kafka:inbound-channel-adapter
id="kafkaInboundChannelAdapter" kafka-consumer-context-ref="consumerContext"
auto-startup="false" channel="inputFromKafka">
<int:poller fixed-delay="10" time-unit="MILLISECONDS"
max-messages-per-poll="5" />
</int-kafka:inbound-channel-adapter>
<bean id="consumerProperties"
class="org.springframework.beans.factory.config.PropertiesFactoryBean">
<property name="properties">
<props>
<prop key="auto.offset.reset">smallest</prop>
<prop key="socket.receive.buffer.bytes">10485760</prop> <!-- 10M -->
<prop key="fetch.message.max.bytes">5242880</prop>
<prop key="auto.commit.interval.ms">1000</prop>
</props>
</property>
</bean>
<int-kafka:consumer-context id="consumerContext"
consumer-timeout="4000" zookeeper-connect="zookeeperConnect" consumer-properties="consumerProperties">
<int-kafka:consumer-configurations>
<int-kafka:consumer-configuration
group-id="mygroup" max-messages="5000">
<int-kafka:topic id="test" streams="4" />
</int-kafka:consumer-configuration>
<!-- <int-kafka:consumer-configuration group-id="default3" value-decoder="kafkaSpecificDecoder"
key-decoder="kafkaReflectionDecoder" max-messages="10"> <int-kafka:topic-filter
pattern="regextopic.*" streams="4" exclude="false" /> </int-kafka:consumer-configuration> -->
</int-kafka:consumer-configurations>
</int-kafka:consumer-context>
<int-kafka:zookeeper-connect id="zookeeperConnect"
zk-connect="localhost:2181" zk-connection-timeout="6000"
zk-session-timeout="400" zk-sync-time="200" />
</beans>

这个配置和Producer类似, 同样声明一个channel, 定义inbound-channel-adapter, 它引用Bean kafka-consumer-context,
kafka-consumer-context定义了消费者的列表。 consumer-configuration还提供了topic-filter,使用正则表达式建立白名单或者黑名单(exclude属性)。

消费者上下文还需要zookeeper-connect

由于spring-integration-kafka只实现了high level Consumer API,这也就意味着你不可能回滚重新查看以前的消息, 因为high level API不提供offset管理。

注意Channel中得到的有效负载的类型是:
Map<String, Map<Integer, List<Object>>>,
这个Map的key是topic, 值还是另外的Map对象。
这个值Map的key值是分区号,value值是消息列表。 在本例中由于消息是字符串, 转换成了byte[]数组。

这种复杂的结构是由于Kafka的设计造成的。 Kafka保证对于同一个topic的同一个分区的消息是严格有序的。所有这种数据结构可以提供有序的消息。