Apache Flume 安装文档、日志收集

首页 > 代码库 > Apache Flume 安装文档、日志收集

Apache Flume 安装文档、日志收集

2024-09-17 08:57:43 220人阅读

简介：

官网 http://flume.apache.org

文档 https://flume.apache.org/FlumeUserGuide.html

hadoop 生态系统中，flume 的职责是收集数据，一般用作收集各种日志数据。

Source -> Channel -> Sink 这是一个基本的工作流程。

Source 定义了数据从哪里来，Channel 是一个数据暂存的位置 ( disk / mem )，Sink 定义将数据流向哪里！

一、flume 安装

shell > yum -y install jdk1.8.0_111  # java 环境必不可少

shell > java -version
java version "1.8.0_111"
Java(TM) SE Runtime Environment (build 1.8.0_111-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.111-b14, mixed mode)

shell > cd /usr/local/src
shell > wget http://www.apache.org/dyn/closer.lua/flume/1.7.0/apache-flume-1.7.0-bin.tar.gz

shell > tar zxf apache-flume-1.7.0-bin.tar.gz -C /usr/local

二、httpd ( 模拟一个日志源 )

shell > yum -y install httpd

shell > /etc/init.d/httpd start

三、flume 基本配置

1、修改 flume 环境变量配置文件

shell > cd /usr/local/apache-flume-1.7.0-bin
shell > cp conf/flume-env.sh.template conf/flume-env.sh

shell > vim conf/flume-env.sh
# 定义 JAVA_HOME
export JAVA_HOME=/usr/java/default

2、修改 flume 配置文件

shell > vim conf/flume-conf.properties
# 定义 sources、channels、sinks
flumeAgent.sources = r_log
flumeAgent.channels = c_memory
flumeAgent.sinks = s_logger

# Source
flumeAgent.sources.r_log.type = exec
flumeAgent.sources.r_log.channels = c_memory
flumeAgent.sources.r_log.command = tail -F /var/log/httpd/access_log

# Channel
flumeAgent.channels.c_memory.type = memory
flumeAgent.channels.c_memory.capacity = 1000
flumeAgent.channels.c_memory.transactionCapacity = 100

# Sink
flumeAgent.sinks.s_logger.type = logger
flumeAgent.sinks.s_logger.channel = c_memory

# 这是一个测试的配置文件，flume 监听 access_log 并输出到 logger 

# flumeAgent.channels.c_memory.capacity  Channel 存储事件的最大数
# flumeAgent.channels.c_memory.transactionCapacity  从 Source 到 Sink  每传输事物中的最大事件数

四、flume 启动

shell > sh bin/flume-ng agent -c conf -f conf/flume-conf.properties -n flumeAgent -Dflume.root.logger=INFO,console

# 启动一个 agent，-c 指定配置文件目录，-f 指定配置文件，-n 指定 agent 名称，-D 指定日志级别、输出位置
# 当 access_log 有新记录时，就会有事件输出到终端

五、flume 收集日志写入 hdfs

shell > cd /usr/local/apache-flume-1.7.0-bin

shell > vim conf/flume-conf.properties
# 定义 sources、channels、sinks
flumeAgent.sources = r_log
flumeAgent.channels = c_file
flumeAgent.sinks = s_hdfs

# Source
flumeAgent.sources.r_log.type = taildir
flumeAgent.sources.r_log.channels = c_file
flumeAgent.sources.r_log.filegroups = f1
flumeAgent.sources.r_log.filegroups.f1 = /var/log/httpd/access_log
flumeAgent.sources.r_log.positionFile = /data/flume/taildir_position_log.json
flumeAgent.sources.r_log.skipToEnd = true

# Channel
flumeAgent.channels.c_file.type = file
flumeAgent.channels.c_file.checkpointDir = /data/flume/check
flumeAgent.channels.c_file.useDualCheckpoints = true
flumeAgent.channels.c_file.backupCheckpointDir = /data/flume/backup
flumeAgent.channels.c_file.dataDirs = /data/flume/data
flumeAgent.channels.c_file.keep-alive = 10
flumeAgent.channels.c_file.write-timeout = 10

# Sink
flumeAgent.sinks.s_hdfs.type = hdfs
flumeAgent.sinks.s_hdfs.channel = c_file
flumeAgent.sinks.s_hdfs.hdfs.path = hdfs://master.hadoop/webroot/logs
flumeAgent.sinks.s_hdfs.hdfs.filePrefix = api
flumeAgent.sinks.s_hdfs.hdfs.rollInterval = 0
flumeAgent.sinks.s_hdfs.hdfs.rollSize = 0
flumeAgent.sinks.s_hdfs.hdfs.rollCount = 0
flumeAgent.sinks.s_hdfs.hdfs.batchSize = 500
flumeAgent.sinks.s_hdfs.hdfs.fileType = DataStream
flumeAgent.sinks.s_hdfs.hdfs.writeFormat = Text
flumeAgent.sinks.s_hdfs.hdfs.threadsPoolSize = 10
flumeAgent.sinks.s_hdfs.hdfs.round = true
flumeAgent.sinks.s_hdfs.hdfs.roundValue = 24
flumeAgent.sinks.s_hfds.hdfs.roundUnit = hour
flumeAgent.sinks.s_hdfs.hdfs.useLocalTimeStamp = true

# 注意：上述配置文件中涉及到的所有目录都会自动生成，不需要手动创建 ( 当然也可以手动创建 )

shell > sh bin/flume-ng agent -c conf -f conf/flume-conf.properties -n flumeAgent -Dflume.root.logger=INFO,console

# 启动 Agent

shell > tree /data/flume
/data/flume
├── backup
│   ├── backupComplete
│   ├── checkpoint
│   ├── checkpoint.meta
│   ├── inflightputs
│   └── inflighttakes
├── check
│   ├── checkpoint
│   ├── checkpoint.meta
│   ├── inflightputs
│   ├── inflighttakes
│   └── queueset
├── data
│   ├── log-1
│   ├── log-1.meta
│   ├── log-2
│   └── log-2.meta
└── taildir_position_log.json

# 这便是配置中的所需目录及文件

shell > su - hadoop

hadoop shell > hdfs dfs -ls /webroot/logs
-rw-r--r--   3 root   supergroup      73322 2017-04-21 19:02 /webroot/logs/api.1492772551751.tmp

hadoop shell > hdfs dfs -tail -f /webroot/logs/api.1492772551751.tmp
192.168.1.246 - - [21/Apr/2017:19:04:18 +0800] "GET /icons/apache_pb.gif HTTP/1.1" 304 - "http://192.168.1.25/" "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.221 Safari/537.36 SE 2.X MetaSr 1.0"
192.168.1.246 - - [21/Apr/2017:19:04:18 +0800] "GET /icons/poweredby.png HTTP/1.1" 304 - "http://192.168.1.25/" "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.221 Safari/537.36 SE 2.X MetaSr 1.0"

# 可以看到数据正被写入 hdfs ( 注意：是从日志文件头部读取 )

报错管理：

1、hdfs 权限拒绝

2017-04-21 18:24:59,939 (SinkRunner-PollingRunner-DefaultSinkProcessor) [WARN - org.apache.flume.sink.hdfs.HDFSEventSink.process(HDFSEventSink.java:443)] HDFS IO error
org.apache.hadoop.security.AccessControlException: Permission denied: user=root, access=WRITE, inode="/webroot/logs":hadoop:supergroup:drwxr-xr-x
    at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.check(FSPermissionChecker.java:310)
    at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.checkPermission(FSPermissionChecker.java:219)
    at org.apache.hadoop.hdfs.server.namenode.FSPermissionChecker.checkPermission(FSPermissionChecker.java:189)
    at org.apache.hadoop.hdfs.server.namenode.FSDirectory.checkPermission(FSDirectory.java:1654)
    at org.apache.hadoop.hdfs.server.namenode.FSDirectory.checkPermission(FSDirectory.java:1638)
    at org.apache.hadoop.hdfs.server.namenode.FSDirectory.checkAncestorAccess(FSDirectory.java:1597)
    at org.apache.hadoop.hdfs.server.namenode.FSDirWriteFileOp.resolvePathForStartFile(FSDirWriteFileOp.java:293)
    at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFileInt(FSNamesystem.java:2211)
    at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.startFile(FSNamesystem.java:2154)
    at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.create(NameNodeRpcServer.java:727)
    at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.create(ClientNamenodeProtocolServerSideTranslatorPB.java:409)

解决方法：

shell > su - hadoop

hadoop shell > vim /usr/local/hadoop-2.8.0/etc/hadoop/hdfs-site.xml

    <property>
        <name>dfs.permissions.enabled</name>
        <value>false</value>
    </property>

# 暂时关闭了权限验证，还没找到更好的方法 ( 开放 hdfs / 权限，跟关闭权限验证没啥区别 )

Apache Flume 安装文档、日志收集

声明：以上内容来自用户投稿及互联网公开渠道收集整理发布，本网站不拥有所有权，未作人工编辑处理，也不承担相关法律责任，若内容有误或涉及侵权可进行投诉：投诉/举报工作人员会在5个工作日内联系你，一经查实，本站将立刻删除涉嫌侵权内容。

联系
我们

首页 > 代码库 > Apache Flume 安装文档、日志收集

Apache Flume 安装文档、日志收集

看完仍有疑问？有类似问题直接问程序猿