目录
1 Flume 安装部署
1.1 安装地址
(1)Flume 官网地址:http://flume.apache.org/
(2)文档查看地址:http://flume.apache.org/FlumeUserGuide.html
(3)下载地址:http://archive.apache.org/dist/flume/
1.2 安装部署
(1)将 apache-flume-1.9.0-bin.tar.gz 上传到 linux 的/opt/software 目录下
(2)解压 apache-flume-1.9.0-bin.tar.gz 到/opt/module/目录下
[atguigu@hadoop102 software]$ tar -zxf /opt/software/apache-flume-1.9.0-bin.tar.gz -C /opt/module/
(3)修改 apache-flume-1.9.0-bin 的名称为 flume
[atguigu@hadoop102 module]$ mv /opt/module/apache-flume-1.9.0-bin/opt/module/flume
(4)将 lib 文件夹下的 guava-11.0.2.jar 删除以兼容 Hadoop 3.1.3
[atguigu@hadoop102 lib]$ rm /opt/module/flume/lib/guava-11.0.2.jar
2 Flume 入门案例
2.1 监控端口数据官方案例
1 )案例需求:
使用 Flume 监听一个端口,收集该端口数据,并打印到控制台。
2 )需求分析:监听数据端口案例分析
3 )实现步骤:
(1)安装 netcat 工具
[atguigu@hadoop102 software]$ sudo yum install -y nc
(2)判断 44444 端口是否被占用
[atguigu@hadoop102 flume-telnet]$ sudo netstat -nlp | grep 44444
(3)创建 Flume Agent 配置文件 flume-netcat-logger.conf
(4)在 flume 目录下创建 job 文件夹并进入 job 文件夹。
[atguigu@hadoop102 flume]$ mkdir job
[atguigu@hadoop102 flume]$ cd job/
(5)在 job 文件夹下创建 Flume Agent 配置文件 flume-netcat-logger.conf。
[atguigu@hadoop102 job]$ vim flume-netcat-logger.conf
(6)在 flume-netcat-logger.conf 文件中添加如下内容。
添加内容如下:
# Name the components on this agent
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# Describe/configure the source
a1.sources.r1.type = netcat
a1.sources.r1.bind = localhost
a1.sources.r1.port = 44444
# Describe the sink
a1.sinks.k1.type = logger
# Use a channel which buffers events in memory
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100
# Bind the source and sink to the channel
a1.sources.r1.channels = c1
a1.sinks.k1.channel = c1
注:配置文件来源于官方手册http://flume.apache.org/FlumeUserGuide.html
配置文件解析
(7)先开启 flume 监听端口
第一种写法:
[atguigu@hadoop102 flume]$ bin/flume-ng agent --conf conf/ --namea1 --conf-file job/flume-netcat-logger.conf -Dflume.root.logger=INFO,console
第二种写法:
[atguigu@hadoop102 flume]$ bin/flume-ng agent -c conf/ -n a1 -fjob/flume-netcat-logger.conf -Dflume.root.logger=INFO,console
参数说明:
--conf/-c
:表示配置文件存储在 conf/目录
--name/-n
:表示给 agent 起名为 a1
--conf-file/-f
:flume 本次启动读取的配置文件是在 job 文件夹下的 flume-telnet.conf
文件。
-Dflume.root.logger=INFO,console :-D 表示 flume 运行时动态修改 flume.root.logger
参数属性值,并将控制台日志打印级别设置为 INFO 级别。日志级别包括:log、info、warn、
error。
(8)使用 netcat 工具向本机的 44444 端口发送内容
[atguigu@hadoop102 ~]$ nc localhost 44444
hello
atguigu
(9)在 Flume 监听页面观察接收数据情况
2.2 实时监控单个追加文件
1 )案例需求:实时监控 Hive 日志,并上传到 HDFS 中
2 ) 需求分析:实时读取本地文件到HDFS案例
3 )实现步骤:
(1)Flume 要想将数据输出到 HDFS,依赖 Hadoop 相关 jar 包
检查/etc/profile.d/my_env.sh 文件,确认 Hadoop 和 Java 环境变量配置正确
JAVA_HOME=/opt/module/jdk1.8.0_212
HADOOP_HOME=/opt/module/ha/hadoop-3.1.3
PATH=$PATH:$JAVA_HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin
export PATH JAVA_HOME HADOOP_HOME
(2)创建 flume-file-hdfs.conf 文件
创建文件
[atguigu@hadoop102 job]$ vim flume-file-hdfs.conf
注:要想读取 Linux 系统中的文件,就得按照 Linux 命令的规则执行命令。由于 Hive
日志在 Linux 系统中所以读取文件的类型选择:exec 即 execute 执行的意思。表示执行
Linux 命令来读取文件。
添加如下内容
# Name the components on this agent
a2.sources = r2
a2.sinks = k2
a2.channels = c2
# Describe/configure the source
a2.sources.r2.type = exec
a2.sources.r2.command = tail -F /opt/module/hive/logs/hive.log
# Describe the sink
a2.sinks.k2.type = hdfs
a2.sinks.k2.hdfs.path = hdfs://hadoop102:9820/flume/%Y%m%d/%H
#上传文件的前缀
a2.sinks.k2.hdfs.filePrefix = logs-
#是否按照时间滚动文件夹
a2.sinks.k2.hdfs.round = true
#多少时间单位创建一个新的文件夹
a2.sinks.k2.hdfs.roundValue = 1
#重新定义时间单位
a2.sinks.k2.hdfs.roundUnit = hour
#是否使用本地时间戳
a2.sinks.k2.hdfs.useLocalTimeStamp = true
#积攒多少个 Event 才 flush 到 HDFS 一次
a2.sinks.k2.hdfs.batchSize = 100
#设置文件类型,可支持压缩
a2.sinks.k2.hdfs.fileType = DataStream
#多久生成一个新的文件
a2.sinks.k2.hdfs.rollInterval = 60
#设置每个文件的滚动大小
a2.sinks.k2.hdfs.rollSize = 134217700
#文件的滚动与 Event 数量无关
a2.sinks.k2.hdfs.rollCount = 0
# Use a channel which buffers events in memory
a2.channels.c2.type = memory
a2.channels.c2.capacity = 1000
a2.channels.c2.transactionCapacity = 100
# Bind the source and sink to the channel
a2.sources.r2.channels = c2
a2.sinks.k2.channel = c2
注意:对于所有与时间相关的转义序列,Event Header 中必须存在以 “timestamp”的
key(除非 hdfs.useLocalTimeStamp 设置为 true,此方法会使用 TimestampInterceptor 自
动添加 timestamp)。
a3.sinks.k3.hdfs.useLocalTimeStamp = true
实时读取本地文件到HDFS案例
# Name the components on this agent
a2.sources = r2
a2.sinks = k2
a2.channels = c2
# Describe/configure the source
a2.sources.r2.type = exec
a2.sources.r2.command = tail -F /opt/module/hive/logs/hive.log
a2.sources.r2.shell = /bin/bash -c
# Describe the sink
a2.sinks.k2.type = hdfs
a2.sinks.k2.hdfs.path = hdfs://hadoop102:9000/flume/%Y%m%d/%H
a2.sinks.k2.hdfs.filePrefix = logs-
a2.sinks.k2.hdfs.round = true
a2.sinks.k2.hdfs.roundValue = 1
a2.sinks.k2.hdfs.roundUnit = hour
a2.sinks.k2.hdfs.useLocalTimeStamp = true
a2.sinks.k2.hdfs.batchSize = 1000
a2.sinks.k2.hdfs.fileType = DataStream
a2.sinks.k2.hdfs.rollInterval = 60
a2.sinks.k2.hdfs.rollSize = 134217700
a2.sinks.k2.hdfs.rollCount = 0
# Use a channel which buffers events in memory
a2.channels.c2.type = memory
a2.channels.c2.capacity = 1000
a2.channels.c2.transactionCapacity = 100
# Bind the source and sink to the channel
a2.sources.r2.channels = c2
a2.sinks.k2.channel = c2
(3)运行 Flume
[atguigu@hadoop102 flume]$ bin/flume-ng agent --conf conf/ --namea2 --conf-file job/flume-file-hdfs.conf
(4)开启 Hadoop 和 Hive 并操作 Hive 产生日志
[atguigu@hadoop102 hadoop-2.7.2]$ sbin/start-dfs.sh
[atguigu@hadoop103 hadoop-2.7.2]$ sbin/start-yarn.sh
[atguigu@hadoop102 hive]$ bin/hive
hive (default)>
(5)在 HDFS 上查看文件。
2.3 实时监控目录下多个新文件
1 ) 案例 需求:使用 F Fe lume 监听整个目录的文件 ,并上传至 HDFS
2 ) 需求分析:实时读取目录文件到HDFS案例
3 ) 实现步骤:
(1)创建配置文件 flume-dir-hdfs.conf
创建一个文件
[atguigu@hadoop102 job]$ vim flume-dir-hdfs.conf
添加如下内容
a3.sources = r3
a3.sinks = k3
a3.channels = c3
# Describe/configure the source
a3.sources.r3.type = spooldir
a3.sources.r3.spoolDir = /opt/module/flume/upload
a3.sources.r3.fileSuffix = .COMPLETED
a3.sources.r3.fileHeader = true
#忽略所有以.tmp 结尾的文件,不上传
a3.sources.r3.ignorePattern = ([^ ]*\.tmp)
# Describe the sink
a3.sinks.k3.type = hdfs
a3.sinks.k3.hdfs.path =
hdfs://hadoop102:9820/flume/upload/%Y%m%d/%H
#上传文件的前缀
a3.sinks.k3.hdfs.filePrefix = upload-
#是否按照时间滚动文件夹
a3.sinks.k3.hdfs.round = true
#多少时间单位创建一个新的文件夹
a3.sinks.k3.hdfs.roundValue = 1
#重新定义时间单位
a3.sinks.k3.hdfs.roundUnit = hour
#是否使用本地时间戳
a3.sinks.k3.hdfs.useLocalTimeStamp = true
#积攒多少个 Event 才 flush 到 HDFS 一次
a3.sinks.k3.hdfs.batchSize = 100
#设置文件类型,可支持压缩
a3.sinks.k3.hdfs.fileType = DataStream
#多久生成一个新的文件
a3.sinks.k3.hdfs.rollInterval = 60
#设置每个文件的滚动大小大概是 128M
a3.sinks.k3.hdfs.rollSize = 134217700
#文件的滚动与 Event 数量无关
a3.sinks.k3.hdfs.rollCount = 0
# Use a channel which buffers events in memory
a3.channels.c3.type = memory
a3.channels.c3.capacity = 1000
a3.channels.c3.transactionCapacity = 100
# Bind the source and sink to the channel
a3.sources.r3.channels = c3
a3.sinks.k3.channel = c3
实时读取目录文件到HDFS案例
a3.sources = r3
a3.sinks = k3
a3.channels = c3
# Describe/configure the source
a3.sources.r3.type = spooldir
a3.sources.r3.spoolDir = /opt/module/flume/upload
a3.sources.r3.fileSuffix = .COMPLETED
a3.sources.r3.fileHeader = true
a3.sources.r3.ignorePattern = ([^ ]*\.tmp)
# Describe the sink
a3.sinks.k3.type = hdfs
a3.sinks.k3.hdfs.path =
hdfs://hadoop102:9000/flume/upload/%Y%m%d/%
H
a3.sinks.k3.hdfs.filePrefix = upload-
a3.sinks.k3.hdfs.round = true
a3.sinks.k3.hdfs.roundValue = 1
a3.sinks.k3.hdfs.roundUnit = hour
a3.sinks.k3.hdfs.useLocalTimeStamp = true
a3.sinks.k3.hdfs.batchSize = 100
a3.sinks.k3.hdfs.fileType = DataStream
a3.sinks.k3.hdfs.rollInterval = 60
a3.sinks.k3.hdfs.rollSize = 134217700
a3.sinks.k3.hdfs.rollCount = 0
# Use a channel which buffers events in memory
a3.channels.c3.type = memory
a3.channels.c3.capacity = 1000
a3.channels.c3.transactionCapacity = 100
# Bind the source and sink to the channel
a3.sources.r3.channels = c3
a3.sinks.k3.channel = c3
(2)启动监控文件夹命令
[atguigu@hadoop102 flume]$ bin/flume-ng agent --conf conf/ --name
a3 --conf-file job/flume-dir-hdfs.conf
说明:在使用 Spooling Directory Source 时,不要在监控目录中创建并持续修改文
件;上传完成的文件会以.COMPLETED 结尾;被监控文件夹每 500 毫秒扫描一次文件变动。
(3)向 upload 文件夹中添加文件
在/opt/module/flume 目录下创建 upload 文件夹
[atguigu@hadoop102 flume]$ mkdir upload
向 upload 文件夹中添加文件
[atguigu@hadoop102 upload]$ touch atguigu.txt
[atguigu@hadoop102 upload]$ touch atguigu.tmp
[atguigu@hadoop102 upload]$ touch atguigu.log
(4)查看 HDFS 上的数据
2.4 实时监控目录下的多个追加文件
Exec source 适用于监控一个实时追加的文件,不能实现断点续传;Spooldir Source
适合用于同步新文件,但不适合对实时追加日志的文件进行监听并同步;而 Taildir Source
适合用于监听多个实时追加的文件,并且能够实现断点续传。
1 ) 案例需求: 使用 Flume 监听整个目录的 实时追加文件,并上传至 HDFS
2 ) 需求分析: 实时读取目录文件到HDFS案例
3 3 ) 实现步骤:
(1)创建配置文件 flume-taildir-hdfs.conf
创建一个文件
[atguigu@hadoop102 job]$ vim flume-taildir-hdfs.conf
添加如下内容
a3.sources = r3
a3.sinks = k3
a3.channels = c3
# Describe/configure the source
a3.sources.r3.type = TAILDIR
a3.sources.r3.positionFile = /opt/module/flume/tail_dir.json
a3.sources.r3.filegroups = f1 f2
a3.sources.r3.filegroups.f1 = /opt/module/flume/files/.*file.*
a3.sources.r3.filegroups.f2 = /opt/module/flume/files2/.*log.*
# Describe the sink
a3.sinks.k3.type = hdfs
a3.sinks.k3.hdfs.path =
hdfs://hadoop102:9820/flume/upload2/%Y%m%d/%H
#上传文件的前缀
a3.sinks.k3.hdfs.filePrefix = upload-
#是否按照时间滚动文件夹
a3.sinks.k3.hdfs.round = true
#多少时间单位创建一个新的文件夹
a3.sinks.k3.hdfs.roundValue = 1
#重新定义时间单位
a3.sinks.k3.hdfs.roundUnit = hour
#是否使用本地时间戳
a3.sinks.k3.hdfs.useLocalTimeStamp = true
#积攒多少个 Event 才 flush 到 HDFS 一次
a3.sinks.k3.hdfs.batchSize = 100
#设置文件类型,可支持压缩
a3.sinks.k3.hdfs.fileType = DataStream
#多久生成一个新的文件
a3.sinks.k3.hdfs.rollInterval = 60
#设置每个文件的滚动大小大概是 128M
a3.sinks.k3.hdfs.rollSize = 134217700
#文件的滚动与 Event 数量无关
a3.sinks.k3.hdfs.rollCount = 0
# Use a channel which buffers events in memory
a3.channels.c3.type = memory
a3.channels.c3.capacity = 1000
a3.channels.c3.transactionCapacity = 100
# Bind the source and sink to the channel
a3.sources.r3.channels = c3
a3.sinks.k3.channel = c3
实时读取目录文件到HDFS案例
a3.sources = r3
a3.sinks = k3
a3.channels = c3
# Describe/configure the source
a2.sources.r2.type = TAILDIR
a2.sources.r2.positionFile = /opt/module/flume/tail_dir.json
a2.sources.r2.filegroups = f1
a2.sources.r2.filegroups.f1 = /opt/module/flume/files/file*
# Describe the sink
a3.sinks.k3.type = hdfs
a3.sinks.k3.hdfs.path =
hdfs://hadoop102:9000/flume/upload/%Y%m%d/%H
a3.sinks.k3.hdfs.filePrefix = upload-
a3.sinks.k3.hdfs.round = true
a3.sinks.k3.hdfs.roundValue = 1
a3.sinks.k3.hdfs.roundUnit = hour
a3.sinks.k3.hdfs.useLocalTimeStamp = true
a3.sinks.k3.hdfs.batchSize = 100
a3.sinks.k3.hdfs.fileType = DataStream
a3.sinks.k3.hdfs.rollInterval = 60
a3.sinks.k3.hdfs.rollSize = 134217700
a3.sinks.k3.hdfs.rollCount = 0
# Use a channel which buffers events in memory
a3.channels.c3.type = memory
a3.channels.c3.capacity = 1000
a3.channels.c3.transactionCapacity = 100
# Bind the source and sink to the channel
a3.sources.r3.channels = c3
a3.sinks.k3.channel = c3
(2)启动监控文件夹命令
[atguigu@hadoop102 flume]$ bin/flume-ng agent --conf conf/ --namea3 --conf-file job/flume-taildir-hdfs.conf
(3)向 files 文件夹中追加内容
在/opt/module/flume 目录下创建 files 文件夹
[atguigu@hadoop102 flume]$ mkdir files
向 upload 文件夹中添加文件
[atguigu@hadoop102 files]$ echo hello >> file1.txt
[atguigu@hadoop102 files]$ echo atguigu >> file2.txt
(4)查看 HDFS 上的数据
r Taildir 说明:
Taildir Source 维护了一个 json 格式的 position File,其会定期的往 position File
中更新每个文件读取到的最新的位置,因此能够实现断点续传。Position File的格式如下:
{"inode":2496272,"pos":12,"file":"/opt/module/flume/files/file1.txt"}
{"inode":2496275,"pos":12,"file":"/opt/module/flume/files/file2.txt"}
注:Linux 中储存文件元数据的区域就叫做 inode,每个 inode 都有一个号码,操作系统
用 inode 号码来识别不同的文件,Unix/Linux 系统内部不使用文件名,而使用 inode 号码来
识别文件。