Flume读取kafka的JSON数据到hdfs，提示OOM

1、安装Flume：解压flume安装包，配置环境变量，删除（重命名）jar包

2、flume配置文件，新建文件如下

#声明变量

at2.sources =st2

at2.channels = ct2

at2.sinks = kt2

# For each one of the sources, the type is defined

#定义数据来源

at2.sources.st2.type = org.apache.flume.source.kafka.KafkaSource

at2.sources.st2.kafka.bootstrap.servers = node0.***:9200,node1.***:9200,node2.***:9200

#kafka主题

at2.sources.st2.kafka.topics = flume-test

#kafka的groupid

at2.sources.st2.kafka.consumer.group.id = flume-to-file-test-group-id

#设置从头开始读取kafka

at2.sources.st2.kafka.consumer.auto.offset.reset=earliest

# The channel can be defined as follows.

at2.sources.st2.channels = ct2

at2.sources.st2.kafka.batchSize = 100

# Each sink's type must be defined

#设置数据去向（写入到文件夹）

at2.sinks.kt2