Flume NG 1.x 是Flume 0.9.x的重构版本，基本面目全非了，Master和zookeeper没有了，collector没有了，Web console没有了，只有
source （avro：很简单使用；exec：使用shell命令）sink （我用的hdfs）channl
这3个组件，俨然从一个分布式系统变成了传输工具。
下面是一个例子（参数经过优化），使用avro作为source，hdfs作为sink，memory作为channel
1. 配置hadoop客户端，使hadoop fs -ls /执行无误
2.修改配置文件flume.conf：
# Define a memory channel called ch1 on agent1agent1.channels.ch1.type = memoryagent1.channels.ch1.capacity = 100000agent1.channels.ch1.transactionCapacity = 100000agent1.channels.ch1.keep-alive = 30# Define an Avro source called avro-source1 on agent1 and tell it# to bind to 0.0.0.0:41414. Connect it to channel ch1.agent1.sources.avro-source1.channels = ch1agent1.sources.avro-source1.type = avroagent1.sources.avro-source1.bind = 127.0.0.1agent1.sources.avro-source1.port = 41414agent1.sources.avro-source1.threads = 5# Define a logger sink that simply logs all events it receives# and connect it to the other end of the same channel.agent1.sinks.log-sink1.channel = ch1agent1.sinks.log-sink1.type = hdfsagent1.sinks.log-sink1.hdfs.path = hdfs://162.105.80.1:9000/flume/agent1.sinks.log-sink1.hdfs.writeFormat = Textagent1.sinks.log-sink1.hdfs.fileType = DataStreamagent1.sinks.log-sink1.hdfs.rollInterval = 0agent1.sinks.log-sink1.hdfs.rollSize = 60554432agent1.sinks.log-sink1.hdfs.rollCount = 0agent1.sinks.log-sink1.hdfs.batchSize = 1000agent1.sinks.log-sink1.hdfs.txnEventMax = 1000agent1.sinks.log-sink1.hdfs.callTimeout = 60000agent1.sinks.log-sink1.hdfs.appendTimeout = 60000# Finally, now that we’ve defined all of our components, tell# agent1 which ones we want to activate.agent1.channels = ch1agent1.sources = avro-source1agent1.sinks = log-sink1
3.修改flume-env.sh配置：
# Note that the Flume conf directory is always included in the classpath.#FLUME_CLASSPATH=""FLUME_CLASSPATH=.jars=`ls /usr/lib/flume-ng/lib`for jar in $jarsdo FLUME_CLASSPATH="$FLUME_CLASSPATH:/usr/lib/flume-ng/lib/$jar"done
4.根据需要在log4j.properties中设置flume.log文件的路径
5.确保当前用户跟hadoop用户一致（假定hadoop没有用户认证机制），并且对flume.log有写权限
6.启动avro agent，，参数“ageng1”在上述配置文件定义。
flume-ng agent –conf ./conf/ -f conf/flume.conf -n agent1
7. client端上传文件:
flume-ng avro-client –conf conf -H 127.0.0.1 -p 41414 -F /data/xx.txt
心得：
感觉就是一个传输工具，使用配置很简单，但是参数还得调，要不报错。上传文件可以自动按照大小、行或者处理时间分割为多个文件。测试了单机上传700MB文件，上传速度和hadoop fs -put速度相同，记录没有丢失。
纠结：
输出文件只能在flume.conf定义吗？如何将文件按照规则汇总到HDFS（比如按照客户合并输出）？基于事务的传输保证每个事务内Event（日志行）有保证传到HDFS，但是如果传输整个文件中途中断，还是会有部分数据上传遗留到HDFS。
结论：FlumeNG1.1，尚不能满足复杂业务要求，尽管支持自定义source和sink等组件，系统使用过程中觉得软件不够坚固，非常简单的示例，都报错，让人心里没底，还好源码够简单，这样看来只适合做简单传输。
错误记录：
org.apache.flume.ChannelException: Space for commit to queue couldn’t be acquired Sinks are likely not keeping up with sources, or the buffer size is too tight
解决：设置agent1.channels.<channel_name>.keep-alive = 30
资料：
FlumeNG 架构
https://blogs.apache.org/flume/entry/flume_ng_architecture
Flume User Guide
美好的生命应该充满期待、惊喜和感激
