Flume的安装和测试故障转移

[b]1.实现功能 [/b]
配置Flume监控本地文件夹变化，将变化的文件上传到hdfs上。
[b]2.集群规划（3台机器都需要安装）[/b]
[img]http://dl2.iteye.com/upload/attachment/0118/1297/28d83bb4-6f97-36c7-8f54-8b948969bfe4.jpg[/img]
[img]http://dl2.iteye.com/upload/attachment/0118/1299/3b4fea49-a0a0-335f-b225-0f8e3038bc37.png[/img]
[b]3.软件准备 [/b]
下载软件包：http://flume.apache.org/download.html 选择当前最新版本：apache-flume-1.6.0-bin.tar.gz
并将其上传到虚拟机的/usr/local/flume目录下，如果没有创建目录；
运行命令：root@master1:/usr/local/flume# tar -zxvf apache-flume-1.6.0-bin.tar.gz解压；

[b]4.配置环境变量（3台机器环境变量配置一样）[/b]
编辑.bashrc文件，添加下面的内容：

export FLUME_HOME=/usr/local/flume/apache-flume-1.6.0-bin 
export FLUME_CONF_DIR=${FLUME_HOME}/conf 
export PATH=.:${JAVA_HOME}/bin:${SCALA_HOME}/bin:${HADOOP_HOME}/bin:${HADOOP_HOME}/sbin:${SPARK_HOME}/bin:${ZOOKEEPER_HOME}/bin:${HIVE_HOME}/bin:${FLUME_HOME}/bin:$PATH

运行root@master1:/usr/local/flume# source ~/.bashrc命令使之生效；
[b]5.各个节点配置 [/b]
master节点上配置/conf/flume-client.properties(source的日志收集)

#agent1 name  
agent1.channels = c1  
agent1.sources = r1  
agent1.sinks = k1 k2  

#set gruop  
agent1.sinkgroups = g1   

#set channel  
agent1.channels.c1.type = memory  
agent1.channels.c1.capacity = 1000  
agent1.channels.c1.transactionCapacity = 100  

agent1.sources.r1.channels = c1  
agent1.sources.r1.type = spooldir  
agent1.sources.r1.spoolDir =/usr/local/flume/tmp/TestDir  

agent1.sources.r1.interceptors = i1 i2  
agent1.sources.r1.interceptors.i1.type = static  
agent1.sources.r1.interceptors.i1.key = Type  
agent1.sources.r1.interceptors.i1.value = LOGIN  
agent1.sources.r1.interceptors.i2.type = timestamp  

# set sink1  
agent1.sinks.k1.channel = c1  
agent1.sinks.k1.type = avro  
agent1.sinks.k1.hostname = worker1  
agent1.sinks.k1.port = 52020  

# set sink2  
agent1.sinks.k2.channel = c1  
agent1.sinks.k2.type = avro  
agent1.sinks.k2.hostname = worker2  
agent1.sinks.k2.port = 52020  

#set sink group  
agent1.sinkgroups.g1.sinks = k1 k2  

#set failover  
agent1.sinkgroups.g1.processor.type = failover  
agent1.sinkgroups.g1.processor.priority.k1 = 10  
agent1.sinkgroups.g1.processor.priority.k2 = 1  
agent1.sinkgroups.g1.processor.maxpenalty = 10000  

worker1、worker2节点上配置/conf/flume-servre.properties
worker1上：

 
#set Agent name  
a1.sources = r1  
a1.channels = c1  
a1.sinks = k1  

#set channel  
a1.channels.c1.type = memory  
a1.channels.c1.capacity = 1000  
a1.channels.c1.transactionCapacity = 100  

# other node,nna to nns  
a1.sources.r1.type = avro  
a1.sources.r1.bind = worker1  
a1.sources.r1.port = 52020  
a1.sources.r1.interceptors = i1  
a1.sources.r1.interceptors.i1.type = static  
a1.sources.r1.interceptors.i1.key = Collector  
a1.sources.r1.interceptors.i1.value = worker1  
a1.sources.r1.channels = c1  

#set sink to hdfs  
a1.sinks.k1.type=hdfs  
a1.sinks.k1.hdfs.path=/library/flume  
a1.sinks.k1.hdfs.fileType=DataStream  
a1.sinks.k1.hdfs.writeFormat=TEXT  
a1.sinks.k1.hdfs.rollInterval=1  
a1.sinks.k1.channel=c1  
a1.sinks.k1.hdfs.filePrefix=%Y-%m-%d 

worker2上：

#set Agent name  
a1.sources = r1  
a1.channels = c1  
a1.sinks = k1  

#set channel  
a1.channels.c1.type = memory  
a1.channels.c1.capacity = 1000  
a1.channels.c1.transactionCapacity = 100  

# other node,nna to nns  
a1.sources.r1.type = avro  
a1.sources.r1.bind = worker2  
a1.sources.r1.port = 52020  
a1.sources.r1.interceptors = i1  
a1.sources.r1.interceptors.i1.type = static  
a1.sources.r1.interceptors.i1.key = Collector  
a1.sources.r1.interceptors.i1.value = worker2  
a1.sources.r1.channels = c1  
#set sink to hdfs  
a1.sinks.k1.type=hdfs  
a1.sinks.k1.hdfs.path=/library/flume  
a1.sinks.k1.hdfs.fileType=DataStream  
a1.sinks.k1.hdfs.writeFormat=TEXT  
a1.sinks.k1.hdfs.rollInterval=1  
a1.sinks.k1.channel=c1  
a1.sinks.k1.hdfs.filePrefix=%Y-%m-%d  

[b]6.启动Flume集群(必须先启动Hadoop集群) [/b]
a)先启动collector端，即worker1和worker2上的配置文件：

root@worker1:/usr/local/flume/apache-flume-1.6.0-bin/conf# flume-ng agent -n a1 -c conf -f flume-server.properties -Dflume.root.logger=DEBUG,console  
root@worker2:/usr/local/flume/apache-flume-1.6.0-bin/conf# flume-ng agent -n a1 -c conf -f flume-server.properties -Dflume.root.logger=DEBUG,console 

b)再启动agent端，即master1的配置文件：

root@master1:/usr/local/flume/apache-flume-1.6.0-bin/conf# flume-ng agent -n agent1 -c conf -f flume-client.properties -Dflume.root.logger=DEBUG,console

注意因为在master1上配置了agent1.sources.r1.spoolDir =/usr/local/flume/tmp/TestDir 所以在启动master1上的agent前必须先确保该目录存在，如果不存在会报错； 在worker1和worker2上我们虽然配置了a1.sinks.k1.hdfs.path=/library/flume ，但是可以不用事先创建，如果/usr/local/flume/tmp/TestDir目录下有文件变化时，flume会自动在hdfs上创建/library/flume目录。
集群启动后，查看各自的进程:Flume的进程名字叫做“Application”

[img]http://dl2.iteye.com/upload/attachment/0118/1301/ce145b8c-423e-3e38-aee2-36df6a6ca74a.png[/img]

[b]7.测试数据的传输 [/b]
在master1节点指定的路径下，生成新的文件（模拟web Server产生新的日志文件）：
首先，TestDir下没有文件（但是会有一个.flumespool的隐藏文件夹，以英文句号开头的文件是隐藏文件）：

root@master1:/usr/local/flume/tmp/TestDir# ll 
总用量 12 
drwxr-xr-x 3 root root 4096  6月 16 21:16 ./ 
drwxr-xr-x 3 root root 4096  6月 16 21:16 ../ 
drwxr-xr-x 2 root root 4096  6月 16 21:16 .flumespool/ 
root@master1:/usr/local/flume/tmp/TestDir# 

而HDFS指定的路径下也没有文件！

我们在/usr/local/flume/tmp目录下创建一个test.log文件，内容如下：

This is a test file ! 
This is a test file ! 
This is a test file ! 
Spark Hadoop JAVA Scala 
SPark Spark 
Hadoop 

之后，拷贝test.log文件到此目录下：

root@master1:/usr/local/flume/tmp# cp test.log ./TestDir/ 

可以从master1上的控制台可以看到下面的日志信息：
[img]http://dl2.iteye.com/upload/attachment/0118/1303/4db8e72a-8890-3584-a2cf-df848db65a6c.png[/img]
此时，看worker1上的控制台打印的日志如下：
[img]http://dl2.iteye.com/upload/attachment/0118/1305/749df0d8-1c5d-320e-abab-1a20da97ae53.png[/img]
最后，查看HDFS上的生成的文件的内容：
[img]http://dl2.iteye.com/upload/attachment/0118/1307/5e7cf526-844d-323d-afc8-56709e39a6b5.png[/img]

[b]测试同时上传两个文件到TestDir目录中。[/b]
创建test_1.log 内容如下：
Hadoop Hadoop Hadoop
Java Java Java
Hive Hive Hive
再创建test_2.log 内容如下：
Scala Scala Scala
Spark Spark Spark
然后将两个文件同时拷贝到/usr/local/flume/tmp/TestDir目录中：
root@master1:/usr/local/flume/tmp# cp test_* ./TestDir/
查看master1的控制台：
[img]http://dl2.iteye.com/upload/attachment/0118/1310/3dd411d0-7933-3ac6-99ac-4ff621fa2e14.png[/img]
查看worker1的控制台：
[img]http://dl2.iteye.com/upload/attachment/0118/1312/a3456ca2-e0ba-3381-8103-c033b0b47dda.png[/img]
我们看到，源端的2个文件合并到HDFS上的1个文件中了。
再看源端的文件名称：
[img]http://dl2.iteye.com/upload/attachment/0118/1314/132efc93-077f-3abb-95ff-ae6c3282479b.png[/img]
并且文件名后面都加了.completed后缀。
注意：如果/usr/local/flume/tmp/TestDir目录存在了一个test.log的文件，你再次拷贝test.log文件到TestDir目录下，则flume会报错，错误信息如下：

16/06/16 22:17:57 INFO avro.ReliableSpoolingFileEventReader: Last read took us just up to a file boundary. Rolling to the next file, if there is one.  
16/06/16 22:17:57 INFO avro.ReliableSpoolingFileEventReader: Preparing to move file /usr/local/flume/tmp/TestDir/test.log to /usr/local/flume/tmp/TestDir/test.log.COMPLETED  
16/06/16 22:17:57 ERROR source.SpoolDirectorySource: FATAL: Spool Directory source r1: { spoolDir: /usr/local/flume/tmp/TestDir }: Uncaught exception in SpoolDirectorySource thread. Restart or reconfigure Flume to continue processing.  
java.lang.IllegalStateException: File name has been re-used with different files. Spooling assumptions violated for /usr/local/flume/tmp/TestDir/test.log.COMPLETED  
    at org.apache.flume.client.avro.ReliableSpoolingFileEventReader.rollCurrentFile(ReliableSpoolingFileEventReader.java:378)  
    at org.apache.flume.client.avro.ReliableSpoolingFileEventReader.retireCurrentFile(ReliableSpoolingFileEventReader.java:330)  
    at org.apache.flume.client.avro.ReliableSpoolingFileEventReader.readEvents(ReliableSpoolingFileEventReader.java:259)  
    at org.apache.flume.source.SpoolDirectorySource$SpoolDirectoryRunnable.run(SpoolDirectorySource.java:228)  
    at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)  
    at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308)  
    at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)  
    at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294)  
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)  
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)  
    at java.lang.Thread.run(Thread.java:745)  

此时，必须删除TestDir目录中的test.log文件，再重启master1上的agent才行；
但是flume也会将你本次拷贝到TestDir目录中的文件上传到hdfs上面,测试如下：
root@master1:/usr/local/flume/tmp# vim test.log
编辑并在最后一行添加Hive
root@master1:/usr/local/flume/tmp# cp test.log ./TestDir/
查看WebUI:
[img]http://dl2.iteye.com/upload/attachment/0118/1316/7f08a2cc-06b5-3170-a76d-2f6676355116.png[/img]

root@master1:/usr/local/flume/tmp# hdfs dfs -cat /library/flume/2016-06-16.1466088454203  
16/06/16 22:47:58 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable  
This is a test file !  
This is a test file !  
This is a test file !  
Spark Hadoop JAVA Scala  
SPark Spark  
Hadoop  
Hive  

[b]8.测试Failover(故障转移) [/b]
我们来测试下Flume NG集群的高可用（故障转移）。场景如下：我们在Agent1节点上传文件，由于我们配置Collector1的权重比Collector2大，所以Collector1优先采集并上传到存储系统。然后我们kill掉Collector1，此时有Collector2负责日志的采集上传工作，之后，我们手动恢复Collector1节点的Flume服务，再次在Agent1上次文件，发现Collector1恢复优先级别的采集工作:
worker1上：

root@worker1:~# jps 
23970 Application 
23826 NodeManager 
24677 Jps 
23690 DataNode 
root@worker1:~# kill -9 23970 
root@worker1:~# jps 
24688 Jps 
23826 NodeManager 
23690 DataNode 

此时worker1控制台上也显示Collector1被杀死了。
再次拷贝1个日志文件到TestDir目录：（拷贝的文件不要和之前拷贝到TestDir中的文件重名）
master1上：

root@master1:/usr/local/flume/tmp# vim test_3.log  
root@master1:/usr/local/flume/tmp# cat test_3.log   
Test Failover  
Test Failover  
root@master1:/usr/local/flume/tmp# cp test_3.log ./TestDir/ 

master1控制台，显示监听到文件变化，并上传到hdfs，之后会显示sink1报错 ，并一直提示

Attempting to create Avro Rpc client.  
16/06/16 22:41:11 INFO avro.ReliableSpoolingFileEventReader: Last read took us just up to a file boundary. Rolling to the next file, if there is one.  
16/06/16 22:41:11 INFO avro.ReliableSpoolingFileEventReader: Preparing to move file /usr/local/flume/tmp/TestDir/test_3.log to /usr/local/flume/tmp/TestDir/test_3.log.COMPLETED  
16/06/16 22:41:19 WARN sink.FailoverSinkProcessor: Sink k1 failed and has been sent to failover list  
org.apache.flume.EventDeliveryException: Failed to send events  
    at org.apache.flume.sink.AbstractRpcSink.process(AbstractRpcSink.java:392)  
    at org.apache.flume.sink.FailoverSinkProcessor.process(FailoverSinkProcessor.java:182)  
    at org.apache.flume.SinkRunner$PollingRunner.run(SinkRunner.java:147)  
    at java.lang.Thread.run(Thread.java:745)  
Caused by: org.apache.flume.EventDeliveryException: NettyAvroRpcClient { host: worker1, port: 52020 }: Failed to send batch  
    at org.apache.flume.api.NettyAvroRpcClient.appendBatch(NettyAvroRpcClient.java:315)  
    at org.apache.flume.sink.AbstractRpcSink.process(AbstractRpcSink.java:376)  
    ... 3 more  
Caused by: org.apache.flume.EventDeliveryException: NettyAvroRpcClient { host: worker1, port: 52020 }: RPC request exception  
    at org.apache.flume.api.NettyAvroRpcClient.appendBatch(NettyAvroRpcClient.java:365)  
    at org.apache.flume.api.NettyAvroRpcClient.appendBatch(NettyAvroRpcClient.java:303)  
    ... 4 more  
Caused by: java.util.concurrent.ExecutionException: java.io.IOException: Error connecting to worker1/192.168.112.131:52020  
    at java.util.concurrent.FutureTask.report(FutureTask.java:122)  
    at java.util.concurrent.FutureTask.get(FutureTask.java:206)  
    at org.apache.flume.api.NettyAvroRpcClient.appendBatch(NettyAvroRpcClient.java:357)  
    ... 5 more  
Caused by: java.io.IOException: Error connecting to worker1/192.168.112.131:52020  
    at org.apache.avro.ipc.NettyTransceiver.getChannel(NettyTransceiver.java:261)  
    at org.apache.avro.ipc.NettyTransceiver.getRemoteName(NettyTransceiver.java:386)  
    at org.apache.avro.ipc.Requestor.writeHandshake(Requestor.java:202)  
    at org.apache.avro.ipc.Requestor.access$300(Requestor.java:52)  
    at org.apache.avro.ipc.Requestor$Request.getBytes(Requestor.java:478)  
    at org.apache.avro.ipc.Requestor.request(Requestor.java:147)  
    at org.apache.avro.ipc.Requestor.request(Requestor.java:129)  
    at org.apache.avro.ipc.specific.SpecificRequestor.invoke(SpecificRequestor.java:84)  
    at com.sun.proxy.$Proxy5.appendBatch(Unknown Source)  
    at org.apache.flume.api.NettyAvroRpcClient$2.call(NettyAvroRpcClient.java:348)  
    at org.apache.flume.api.NettyAvroRpcClient$2.call(NettyAvroRpcClient.java:344)  
    at java.util.concurrent.FutureTask.run(FutureTask.java:266)  
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)  
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)  
    ... 1 more  
Caused by: java.net.ConnectException: 拒绝连接  
    at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method)  
    at sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelImpl.java:717)  
    at org.jboss.netty.channel.socket.nio.NioClientSocketPipelineSink$Boss.connect(NioClientSocketPipelineSink.java:496)  
    at org.jboss.netty.channel.socket.nio.NioClientSocketPipelineSink$Boss.processSelectedKeys(NioClientSocketPipelineSink.java:452)  
    at org.jboss.netty.channel.socket.nio.NioClientSocketPipelineSink$Boss.run(NioClientSocketPipelineSink.java:365)  
    ... 3 more  
16/06/16 22:41:22 INFO sink.AbstractRpcSink: Rpc sink k1: Building RpcClient with hostname: worker1, port: 52020  
16/06/16 22:41:22 INFO sink.AvroSink: Attempting to create Avro Rpc client.  
16/06/16 22:41:22 WARN api.NettyAvroRpcClient: Using default maxIOWorkers  
16/06/16 22:41:26 INFO sink.AbstractRpcSink: Rpc sink k1: Building RpcClient with hostname: worker1, port: 52020  
16/06/16 22:41:26 INFO sink.AvroSink: Attempting to create Avro Rpc client.  
16/06/16 22:41:26 WARN api.NettyAvroRpcClient: Using default maxIOWorkers  
16/06/16 22:41:37 INFO sink.AbstractRpcSink: Rpc sink k1: Building RpcClient with hostname: worker1, port: 52020  

worker2的控制台日志输出：

 
16/06/16 22:41:22 INFO hdfs.HDFSDataStream: Serializer = TEXT, UseRawLocalFileSystem = false  
16/06/16 22:41:23 INFO hdfs.BucketWriter: Creating /library/flume/2016-06-16.1466088082911.tmp  
16/06/16 22:41:23 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable  
16/06/16 22:41:26 INFO hdfs.BucketWriter: Closing /library/flume/2016-06-16.1466088082911.tmp  
16/06/16 22:41:26 INFO hdfs.BucketWriter: Renaming /library/flume/2016-06-16.1466088082911.tmp to /library/flume/2016-06-16.1466088082911  
16/06/16 22:41:26 INFO hdfs.HDFSEventSink: Writer callback called.  

查看WebUI：
[img]http://dl2.iteye.com/upload/attachment/0118/1318/ae9b1be3-160c-3c26-bbfa-6a452d7d7978.png[/img]

查看内容：

root@master1:/usr/local/flume/tmp# hdfs dfs -cat /library/flume/2016-06-16.1466088082911  
16/06/16 22:51:44 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable  
Test Failover  
Test Failover  

重启collector1的服务，worker1上运行：

root@worker1:/usr/local/flume/apache-flume-1.6.0-bin/conf# flume-ng agent -n a1 -c conf -f flume-server.properties -Dflume.root.logger=DEBUG,console  

重复上面过程，agent（master）上生成新文件，之后由collector1（worker1）重新接管服务：
master1：

root@master1:/usr/local/flume/tmp# vim test_4.log  
root@master1:/usr/local/flume/tmp# cat test_4.log   
Test Failover is good !!!   
root@master1:/usr/local/flume/tmp# cp test_4.log TestDir/  

master1控制台：

16/06/16 22:57:00 INFO avro.ReliableSpoolingFileEventReader: Last read took us just up to a file boundary. Rolling to the next file, if there is one.  
16/06/16 22:57:00 INFO avro.ReliableSpoolingFileEventReader: Preparing to move file /usr/local/flume/tmp/TestDir/test_4.log to /usr/local/flume/tmp/TestDir/test_4.log.COMPLETED  

Worker1上日志：

16/06/16 22:57:03 INFO hdfs.HDFSDataStream: Serializer = TEXT, UseRawLocalFileSystem = false  
16/06/16 22:57:03 INFO hdfs.BucketWriter: Creating /library/flume/2016-06-16.1466089023374.tmp  
16/06/16 22:57:03 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable  
16/06/16 22:57:06 INFO hdfs.BucketWriter: Closing /library/flume/2016-06-16.1466089023374.tmp  
16/06/16 22:57:06 INFO hdfs.BucketWriter: Renaming /library/flume/2016-06-16.1466089023374.tmp to /library/flume/2016-06-16.1466089023374  
16/06/16 22:57:06 INFO hdfs.HDFSEventSink: Writer callback called.

测试成功，flume确实实现了故障转移！

来源新浪微博：[url]http://weibo.com.ilovepains/[/url]