flume(2)

接续上一篇：https://www.cnblogs.com/metianzing/p/9511852.html

这里也是主要记录配置文件。

以上一篇案例五为基础，考虑到日志服务器和采集日志的服务器往往不是一台，本篇采用多agent的形式。

agent1:　　source:TAILDIR

　　　　　 sink:avro

agent2:　　source:avro

　　　　　 sink:kafka

1>agent1配置文件

cd /usr/local/flume/conf
vim agent1_abtd.conf

编写　agent1_abtd.conf，内容如下：

# Name the components on this agent
agent1.sources = s1
agent1.sinks = k1
agent1.channels = ch1

# Describe/configure the source
agent1.sources.s1.type = TAILDIR
agent1.sources.s1.positionFile = /opt/classiclaw/nginx/logs/abtd_magent/taildir_position.json
agent1.sources.s1.filegroups = f1
agent1.sources.s1.filegroups.f1 = /opt/classiclaw/nginx/logs/access.log.*
agent1.sources.s1.headers.f1.headerKey1 = value1
agent1.sources.s1.fileHeader = true

# Describe the sink
agent1.sinks.k1.type = avro
agent1.sinks.k1.hostname = ubuntu
agent1.sinks.k1.port = 44444

# Set channel
agent1.channels.ch1.type = file
agent1.channels.ch1.checkpointDir = /opt/classiclaw/nginx/logs/flume_data/checkpoint
agent1.channels.ch1.dataDirs = /opt/classiclaw/nginx/logs/flume_data/data

# bind
agent1.sources.s1.channels = ch1
agent1.sinks.k1.channel = ch1

2>　　agent2配置文件

cd /usr/local/flume/conf
vim agent2_abtd.conf

编写agent2_abtd.conf　　，内容如下：

# Name the components on this agent
agent2.sources = s2
agent2.sinks = k2
agent2.channels = ch2

# Describe/configure the source
agent2.sources.s2.type = avro
agent2.sources.s2.bind = 0.0.0.0
agent2.sources.s2.port = 44444

# Describe the sink
agent2.sinks.k2.type = org.apache.flume.sink.kafka.KafkaSink
agent2.sinks.k2.brokerList = ubuntu:9092
agent2.sinks.k2.topic = abtd_magent
agent2.sinks.k2.kafka.flumeBatchSize = 20
agent2.sinks.k2.kafka.producer.acks = 1
agent2.sinks.k2.kafka.producer.linger.ms = 1
agent2.sinks.k2.kafka.producer.compression.type = snappy

# set channel
agent2.channels.ch2.type = memory
agent2.channels.ch2.capacity = 1000000
agent2.channels.ch2.transactionCapacity = 1000000

# Bind the source and sink to the channel
agent2.sources.s2.channels = ch2
agent2.sinks.k2.channel = ch2

3>创建　topic:abtd_magent

前提是已经启动zookeeper和Kafka。

cd /usr/local/kafka
./bin/kafka-topics.sh --zookeeper ubuntu:2181 --create --topic abtd_magent --replication-factor 1 --partitions 3 

4>启动agent2

cd /usr/local/flume
./bin/flume-ng agent --name agent2 --conf conf/ --conf-file conf/agent2_abtd.conf -Dflume.root.logger=INFO,console

5>启动agent1

cd /usr/local/flume
./bin/flume-ng agent --name agent1 --conf conf/ --conf-file conf/agent1_abtd.conf -Dflume.root.logger=INFO,console

6>agent1 :添加文件到flume source目录　　

cd /opt/classiclaw/nginx/logs
echo -e "this is a test file! 
http://www.aboutyun.com20170820">access.log.1

7>查看kafka consumer

agent1: