大家好，又见面了，我是你们的朋友全栈君。

Flume 对接 Kafka

Flume日志采集组件；Flume对接kafka主要是为了通过kafka的topic功能，动态的增加或者减少接收的节点，并且Flume要对接多个节点是需要多个channel和sink的会导致内存不够的情况。

那么可以实现的场景就是Flume采集日志文件，通过kafka给多给业务线使用。

1)配置 flume(flume-kafka.conf)

# define
a1.sources = r1
a1.sinks = k1
a1.channels = c1

# source
a1.sources.r1.type = netcat
a1.sources.r1.bind = localhost
a1.sources.r1.port = 44444

# sink
a1.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
a1.sinks.k1.kafka.bootstrap.servers = hadoop113:9092,hadoop114:9092,hadoop115:9092
a1.sinks.k1.kafka.topic = first
a1.sinks.k1.kafka.flumeBatchSize = 20
a1.sinks.k1.kafka.producer.acks = 1
a1.sinks.k1.kafka.producer.linger.ms = 1

# channel
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100

# bind
a1.sources.r1.channels = c1
a1.sinks.k1.channel = c1

2. 启动消费者

kafka-console-consumer.sh --zookeeper hadoop113:2181 --topic first

3. 进入 flume 根目录下,启动 flume

bin/flume-ng agent -c conf/ -n a1 -f jobs/flume-kafka.conf

4)启动nc发送数据

[bd@hadoop113 ~]$ nc localhost 44444
hello
OK
word
OK

结果如下
[bd@hadoop113 ~]$ kafka-console-consumer.sh --zookeeper hadoop113:2181 --topic first
hello
word

Kafka数据分类

依据Kafka Sink的配置

在消息头中携带了topic字段的话，该消息就会被发送到topic字段对应的topic去。

那么在flume接收到消息之后，可以通过拦截器为topic加上header，即可将其进行分类。

Flume拦截器如下：

public class JudgeTestStringInterceptor implements Interceptor { 

// 声明一个存放事件的List
private List<Event> allEvents;
public void initialize () { 

// 初始化
allEvents = new ArrayList<Event>();
}
/** * 单个事件拦截 * @param event * @return */
public Event intercept (Event event) { 

// 1、获取事件中的头信息
Map<String, String> headers = event.getHeaders();
// 2、获取事件中的body信息
String body = new String(event.getBody());
// 3、根据body中是否有“test”来决定添加怎样的头信息
// 有的话添加<topic, first>没有则添加<topic, second>
if (body.contains("test")) { 

headers.put("topic", "first");
} else { 

headers.put("topic", "second");
}
return event;
// 如果返回null则认为该事件无用，将会被过滤
}
/** * 批量事件拦截 * @param list * @return */
public List<Event> intercept (List<Event> list) { 

// 1、清空集合
allEvents.clear();
// 2、遍历event
for (Event event : list) { 

// 3、给每个事件添加头信息
allEvents.add(intercept(event));
}
return allEvents;
}
public void close () { 

}
// 定义一个Builder对象
public static class Builder implements Interceptor.Builder { 

public Interceptor build () { 

return new JudgeTestStringInterceptor();
}
public void configure (Context context) { 

}
}
}

配置文件type-kafka.conf如下：

# define
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# source
a1.sources.r1.type = netcat
a1.sources.r1.bind = localhost
a1.sources.r1.port = 44444
# interceptor
a1.sources.r1.interceptors = i1
a1.sources.r1.interceptors.i1.type = com.starnet.interceptor.JudgeTestStringInterceptor$Builder
# sink
a1.sinks.k1.type = org.apache.flume.sink.kafka.KafkaSink
a1.sinks.k1.kafka.bootstrap.servers = hadoop113:9092,hadoop114:9092,hadoop115:9092
a1.sinks.k1.kafka.topic = first
a1.sinks.k1.kafka.flumeBatchSize = 20
a1.sinks.k1.kafka.producer.acks = 1
a1.sinks.k1.kafka.producer.linger.ms = 1
# channel
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100
# bind
a1.sources.r1.channels = c1
a1.sinks.k1.channel = c1

启动flume，两个消费者以及nc之后结果如下：

[bd@hadoop113 ~]$ nc localhost 44444
test
OK
hello
OK
word
OK
[bd@hadoop113 ~]$ kafka-console-consumer.sh --zookeeper hadoop113:2181 --topic first
test
[bd@hadoop113 ~]$ kafka-console-consumer.sh --zookeeper hadoop113:2181 --topic second
hello
word