Flume-ng分布式部署和配置

Flume-ng分布式部署总结

1.日志总接收端的配置（负责接收各个节点发送过来的日志数据）,修改flume的安装目录下/conf/flume-conf.properties配置文件

（默认没有，cp flume-conf.properties.template flume-conf.properties）

agent.sources = avrosrc  
agent.channels = memoryChannel  
agent.sinks = hdfsSink  

# For each one of the sources, the type is defined  
agent.sources.avrosrc.type = avro  
agent.sources.avrosrc.bind = 192.168.35.100 
agent.sources.avrosrc.port = 44444  
 
# The channel can be defined as follows.  
agent.sources.avrosrc.channels = memoryChannel  
 
# Each channel's type is defined.  
agent.channels.memoryChannel.type = memory  
agent.channels.memoryChannel.keep-alive = 10  
agent.channels.memoryChannel.capacity = 100000  
agent.channels.memoryChannel.transactionCapacity =100000  
 
# Each sink's type must be defined  
agent.sinks.hdfsSink.type = hdfs  
agent.sinks.hdfsSink.channel = memoryChannel  

agent.sinks.hdfsSink.hdfs.path = /flume_logs/%Y%m%d       #存放文件的目录结构

#生成的文件的名称，“datacenter”各节点配置文件中定义的名称  

agent.sinks.hdfsSink.hdfs.filePrefix = %{datacenter}_     
agent.sinks.hdfsSink.hdfs.rollInterval = 0  
agent.sinks.hdfsSink.hdfs.rollSize = 4000000              
agent.sinks.hdfsSink.hdfs.rollCount = 0  
agent.sinks.hdfsSink.hdfs.writeFormat = Text  
agent.sinks.hdfsSink.hdfs.fileType = DataStream  
agent.sinks.hdfsSink.hdfs.batchSize = 10  

2.配置收集日志各节点的Flume的配置文件，flume的安装目录下/conf/flume-conf.properties
agent.sources = nodesource
agent.channels = nodeMemoryChannel
agent.sinks = nodeSink
 
agent.sources.nodesource.type = exec  
agent.sources.nodesource.command = tail -F /root/logs/log1.log  #监听的日志文件
agent.sources.nodesource.channels = nodeMemoryChannel  
 
agent.sources.nodesource.interceptors = host_int timestamp_int inter1  
agent.sources.nodesource.interceptors.host_int.type = host  
agent.sources.nodesource.interceptors.host_int.hostHeader = hostname  
 
agent.sources.nodesource.interceptors.timestamp_int.type = org.apache.flume.interceptor.TimestampInterceptor$Builder  
 
#agent.sources.nodesource.interceptors = inter1  

agent.sources.nodesource.interceptors.inter1.type = static  

#数据接收端通过“datacenter”获取value值定义文件名称

agent.sources.nodesource.interceptors.inter1.key = datacenter  

agent.sources.nodesource.interceptors.inter1.value = log102         #在数据接收端生成的文件名称
 
agent.channels.nodeMemoryChannel.type = memory  
agent.channels.nodeMemoryChannel.keep-alive = 10  
agent.channels.nodeMemoryChannel.capacity = 100000  
agent.channels.nodeMemoryChannel.transactionCapacity =100000  
 
agent.sinks.nodeSink.type = avro  
agent.sinks.nodeSink.hostname = 192.168.35.100 
agent.sinks.nodeSink.port = 44444  
agent.sinks.nodeSink.channel = nodeMemoryChannel  

3.启动（切换到flume的安装路径下启动）
#将日志信息打印到控制台
bin/flume-ng agent -n agent -c conf -f conf/flume-conf.properties -Dflume.root.logger=INFO,console
#将产生的日志信息保存到日志文件中,并后台运行
bin/flume-ng agent -n agent -c conf -f conf/flume-conf.properties >logs/flume_log.log &