1、目标场景
2、flume agent配置文件
# 01 define agent name, source/sink/channel namea1.sources = r1a1.sinks = k1a1.channels = c1# 02 source,http,jsonhandlera1.sources.r1.type = httpa1.sources.r1.bind = mastera1.sources.r1.port = 6666a1.sources.r1.handler = org.apache.flume.source.http.JSONHandler# 03 timestamp and host interceptors work before sourcea1.sources.r1.interceptors = i1 i2 ??????# 两个interceptor串联,依次作用于eventa1.sources.r1.interceptors.i1.type = timestamp a1.sources.r1.interceptors.i1.preserveExisting = false ??a1.sources.r1.interceptors.i2.type = host # flume event的头部将添加 “hostname”:实际主机名a1.sources.r1.interceptors.i2.hostHeader = hostname ?# 指定key,value将填充为flume agent所在节点的主机名a1.sources.r1.interceptors.i2.useIP = false ?????????# IP和主机名,二选一即可# 04 hdfs sinka1.sinks.k1.type = hdfs ?a1.sinks.k1.hdfs.path = hdfs://master:9000/flume/%Y-%m-%d/ ??# hdfs sink将根据event header中的时间戳进行替换# 和hostHeader的值保持一致,hdfs sink将提取event中key为hostnmae的值,基于该值创建文件名前缀a1.sinks.k1.hdfs.filePrefix = %{hostname} ??# hdfs sink将根据event header中的hostnmae对应的value进行替换a1.sinks.k1.hdfs.fileType = DataStreama1.sinks.k1.hdfs.writeFormat = Texta1.sinks.k1.hdfs.rollInterval = 0a1.sinks.k1.hdfs.rollCount = 10a1.sinks.k1.hdfs.rollSize = 1024000# channel,memorya1.channels.c1.type = memorya1.channels.c1.capacity = 1000a1.channels.c1.transactionCapacity = 100# bind source,sink to channel a1.sinks.k1.channel = c1a1.sources.r1.channels = c1
3、验证timestamp+host interceptor
验证思路:
1)先将interceptor作用后的event,通过logger sink打印到console,验证header是否正常添加
2)修改sink为hdfs, 观察目录和文件的名称是否能够按照预期创建(时间戳-目录,hostname-文件前缀)
验证过程:
1)发送header为空的http请求,logger sink打印event到终端,观察event header中是否被添加了timestamp以及hostname
2)ogger打印到console的event,header发生了变化
3)修改sink为hdfs, 观察HDFS的目录名(时间戳)和文件前缀(hostnme)
*目录名被正常替换(基于event header中的时间戳)
*文件前缀被正常替换(基于event header中的hostname:实际主机名)
* 文件内容被写入为event的body
06_Flume_interceptor_时间戳+Host
原文地址:http://www.cnblogs.com/shay-zhangjin/p/7965828.html