我正在测试flume http源代码,hdfs作为sink。下面是使用的conf文件。
httpagent.sources = httpsource
httpagent.sinks = hdfs-file-sink
httpagent.channels = ch3
httpagent.sources.httpsource.type = http
httpagent.sources.httpsource.bind = address
httpagent.sources.httpsource.handler =org.apache.flume.sink.solr.morphline.BlobHandler httpagent.sources.httpsource.channels = ch3
httpagent.sources.httpsource.port = port
httpagent.sinks.hdfs-file-sink.type = hdfs
httpagent.sinks.hdfs-file-sink.hdfs.path = hdfs://localhost:8020/flume/events
httpagent.sinks.hdfs-file-sink.hdfs.fileType=DataStream
httpagent.sinks.hdfs-file-sink.hdfs.filePrefix = events-
httpagent.sinks.hdfs-file-sink.hdfs.rollInterval = 30
httpagent.sinks.hdfs-file-sink.channel = ch3
httpagent.channels.ch3.type = memory
请求主体将保存在hdfs中。但是我想把http头也附加到post内容中。我该怎么做?
1条答案
按热度按时间mrphzbgm1#
您可能想看看事件序列化程序(https://flume.apache.org/flumeuserguide.html#event-序列化程序)
序列化程序可以按中所述进行设置https://flume.apache.org/flumeuserguide.html#hdfs-Flume
如果需要这些头文件,应该使用avro事件序列化程序或编写自定义的事件序列化程序。