1. Flume 介绍
1.1. 概述
- Flume是一个分布式、可靠、和高可用的海量日志采集、聚合和传输的系统。
- Flume可以采集文件，socket数据包、文件、文件夹、kafka等各种形式源数据，又可以将采集到的数据(下沉sink)输出到HDFS、hbase、hive、kafka等众多外部存储系统中
- 一般的采集需求，通过对flume的简单配置即可实现
- Flume针对特殊场景也具备良好的自定义扩展能力，
因此，flume可以适用于大部分的日常数据采集场景
1.2. 运行机制
- Flume分布式系统中最核心的角色是agent，flume采集系统就是由一个个agent所连接起来形成
- 每一个agent相当于一个数据传递员，内部有三个组件：
- Source：采集组件，用于跟数据源对接，以获取数据
- Sink：下沉组件，用于往下一级agent传递数据或者往最终存储系统传递数据
-
Channel：传输通道组件，用于从source将数据传递到sink
image.png
1.3. Flume 结构图
简单结构
单个 Agent 采集数据
image.png
复杂结构
多级 Agent 之间串联
image.png
2. Flume 实战案例
案例：使用网络telent命令向一台机器发送一些网络数据，然后通过flume采集网络端口数据
image.png
2.1. Flume 的安装部署
Step 1: 下载解压修改配置文件
下载地址：
http://archive.apache.org/dist/flume/1.8.0/apache-flume-1.8.0-bin.tar.gz
Flume的安装非常简单，只需要解压即可，当然，前提是已有hadoop环境
上传安装包到数据源所在节点上
这里我们采用在第三台机器来进行安装
cd /export/softwares/
tar -zxvf apache-flume-1.8.0-bin.tar.gz -C ../servers/
cd /export/servers/apache-flume-1.8.0-bin/conf
cp flume-env.sh.template flume-env.sh
vim flume-env.sh
export JAVA_HOME=/export/servers/jdk1.8.0_141
Step 2: 开发配置文件
根据数据采集的需求配置采集方案，描述在配置文件中(文件名可任意自定义)
配置我们的网络收集的配置文件
在flume的conf目录下新建一个配置文件（采集方案）
vim /export/servers/apache-flume-1.8.0-bin/conf/netcat-logger.conf
# 定义这个agent中各组件的名字
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# 描述和配置source组件：r1
a1.sources.r1.type = netcat
a1.sources.r1.bind = 192.168.174.
a1.sources.r1.port = 44444
# 描述和配置sink组件：k1
a1.sinks.k1.type = logger
# 描述和配置channel组件，此处使用是内存缓存的方式
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100
# 描述和配置source channel sink之间的连接关系
a1.sources.r1.channels = c1
a1.sinks.k1.channel = c1
Step 3: 启动配置文件
指定采集方案配置文件，在相应的节点上启动flume agent
先用一个最简单的例子来测试一下程序环境是否正常
启动agent去采集数据
bin/flume-ng agent -c conf -f conf/netcat-logger.conf -n a1 -Dflume.root.logger=INFO,console
-
-c conf指定flume自身的配置文件所在目录 -
-f conf/netcat-logger.con指定我们所描述的采集方案 -
-n a1指定我们这个agent的名字
Step 4: 安装 Telnet 准备测试
在node02机器上面安装telnet客户端，用于模拟数据的发送
yum -y install telnet
telnet node03 44444 # 使用telnet模拟数据发送
2.2. 采集案例
2.2.3. 采集目录到 HDFS
需求
某服务器的某特定目录下，会不断产生新的文件，每当有新文件出现，就需要把文件采集到HDFS中去
思路
根据需求，首先定义以下3大要素
- 数据源组件，即source ——监控文件目录 : spooldir
- 监视一个目录，只要目录中出现新文件，就会采集文件中的内容
- 采集完成的文件，会被agent自动添加一个后缀：COMPLETED
- 所监视的目录中不允许重复出现相同文件名的文件
- 下沉组件，即sink——HDFS文件系统 : hdfs sink
- 通道组件，即channel——可用file channel 也可以用内存channel
Step 1: Flume 配置文件
cd /export/servers/apache-flume-1.8.0-bin/conf
mkdir -p /export/servers/dirfile
vim spooldir.conf
# Name the components on this agent
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# Describe/configure the source
##注意：不能往监控目中重复丢同名文件
a1.sources.r1.type = spooldir
a1.sources.r1.spoolDir = /export/servers/dirfile
a1.sources.r1.fileHeader = true
# Describe the sink
a1.sinks.k1.type = hdfs
a1.sinks.k1.channel = c1
a1.sinks.k1.hdfs.path = hdfs://node01:8020/spooldir/files/%y-%m-%d/%H%M/
a1.sinks.k1.hdfs.filePrefix = events-
a1.sinks.k1.hdfs.round = true
a1.sinks.k1.hdfs.roundValue = 10
a1.sinks.k1.hdfs.roundUnit = minute
a1.sinks.k1.hdfs.rollInterval = 3
a1.sinks.k1.hdfs.rollSize = 20
a1.sinks.k1.hdfs.rollCount = 5
a1.sinks.k1.hdfs.batchSize = 1
a1.sinks.k1.hdfs.useLocalTimeStamp = true
#生成的文件类型，默认是Sequencefile，可用DataStream，则为普通文本
a1.sinks.k1.hdfs.fileType = DataStream
# 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
Channel参数解释
capacity：默认该通道中最大的可以存储的event数量
trasactionCapacity：每次最大可以从source中拿到或者送到sink中的event数量
keep-alive：event添加到通道中或者移出的允许时间
Step 2: 启动 Flume
bin/flume-ng agent -c ./conf -f ./conf/spooldir.conf -n a1 -Dflume.root.logger=INFO,console
Step 3: 上传文件到指定目录
将不同的文件放到下面目录里面去，注意文件不能重名
cd /export/servers/dirfile
2.2.4. 采集文件到 HDFS
需求
比如业务系统使用log4j生成的日志，日志内容不断增加，需要把追加到日志文件中的数据实时采集到hdfs
分析
根据需求，首先定义以下3大要素
- 采集源，即source——监控文件内容更新 : exec ‘tail -F file’
- 下沉目标，即sink——HDFS文件系统 : hdfs sink
- Source和sink之间的传递通道——channel，可用file channel 也可以用 内存channel
Step 1: 定义 Flume 配置文件
cd /export/servers/apache-flume-1.8.0-bin/conf
vim tail-file.conf
agent1.sources = source1
agent1.sinks = sink1
agent1.channels = channel1
# Describe/configure tail -F source1
agent1.sources.source1.type = exec
agent1.sources.source1.command = tail -F /export/servers/taillogs/access_log
agent1.sources.source1.channels = channel1
# Describe sink1
agent1.sinks.sink1.type = hdfs
#a1.sinks.k1.channel = c1
agent1.sinks.sink1.hdfs.path = hdfs://node01:8020/weblog/flume-collection/%y-%m-%d/%H-%M
agent1.sinks.sink1.hdfs.filePrefix = access_log
agent1.sinks.sink1.hdfs.maxOpenFiles = 5000
agent1.sinks.sink1.hdfs.batchSize= 100
agent1.sinks.sink1.hdfs.fileType = DataStream
agent1.sinks.sink1.hdfs.writeFormat =Text
agent1.sinks.sink1.hdfs.round = true
agent1.sinks.sink1.hdfs.roundValue = 10
agent1.sinks.sink1.hdfs.roundUnit = minute
agent1.sinks.sink1.hdfs.useLocalTimeStamp = true
# Use a channel which buffers events in memory
agent1.channels.channel1.type = memory
agent1.channels.channel1.keep-alive = 120
agent1.channels.channel1.capacity = 500000
agent1.channels.channel1.transactionCapacity = 600
# Bind the source and sink to the channel
agent1.sources.source1.channels = channel1
agent1.sinks.sink1.channel = channel1
Step 2: 启动 Flume
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -c conf -f conf/tail-file.conf -n agent1 -Dflume.root.logger=INFO,console
Step 3: 开发 Shell 脚本定时追加文件内容
mkdir -p /export/servers/shells/
cd /export/servers/shells/
vim tail-file.sh
#!/bin/bash
while true
do
date >> /export/servers/taillogs/access_log;
sleep 0.5;
done
Step 4: 启动脚本
# 创建文件夹
mkdir -p /export/servers/taillogs
# 启动脚本
sh /export/servers/shells/tail-file.sh
2.2.5. Agent 级联
[图片上传中...(image.png-6a52c2-1566550110125-0)]
分析
第一个agent负责收集文件当中的数据，通过网络发送到第二个agent当中去
第二个agent负责接收第一个agent发送的数据，并将数据保存到hdfs上面去
Step 1: Node02 安装 Flume
将node03机器上面解压后的flume文件夹拷贝到node02机器上面去
cd /export/servers
scp -r apache-flume-1.8.0-bin/ node02:$PWD
Step 2: Node02 配置 Flume
在node02机器配置我们的flume
cd /export/servers/ apache-flume-1.8.0-bin/conf
vim tail-avro-avro-logger.conf
##################
# Name the components on this agent
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# Describe/configure the source
a1.sources.r1.type = exec
a1.sources.r1.command = tail -F /export/servers/taillogs/access_log
a1.sources.r1.channels = c1
# Describe the sink
##sink端的avro是一个数据发送者
a1.sinks = k1
a1.sinks.k1.type = avro
a1.sinks.k1.channel = c1
a1.sinks.k1.hostname = node03
a1.sinks.k1.port = 4141
a1.sinks.k1.batch-size = 10
# Use a channel which buffers events in memory
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100
# Bind the source and sink to the channel
a1.sources.r1.channels = c1
a1.sinks.k1.channel = c1
Step 3: 开发脚本向文件中写入数据
直接将node03下面的脚本和数据拷贝到node02即可，node03机器上执行以下命令
cd /export/servers
scp -r shells/ taillogs/ node02:$PWD
Step 4: Node03 Flume 配置文件
在node03机器上开发flume的配置文件
cd /export/servers/apache-flume-1.8.0-bin/conf
vim avro-hdfs.conf
# Name the components on this agent
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# Describe/configure the source
##source中的avro组件是一个接收者服务
a1.sources.r1.type = avro
a1.sources.r1.channels = c1
a1.sources.r1.bind = node03
a1.sources.r1.port = 4141
# Describe the sink
a1.sinks.k1.type = hdfs
a1.sinks.k1.hdfs.path = hdfs://node01:8020/av/%y-%m-%d/%H%M/
a1.sinks.k1.hdfs.filePrefix = events-
a1.sinks.k1.hdfs.round = true
a1.sinks.k1.hdfs.roundValue = 10
a1.sinks.k1.hdfs.roundUnit = minute
a1.sinks.k1.hdfs.rollInterval = 3
a1.sinks.k1.hdfs.rollSize = 20
a1.sinks.k1.hdfs.rollCount = 5
a1.sinks.k1.hdfs.batchSize = 1
a1.sinks.k1.hdfs.useLocalTimeStamp = true
#生成的文件类型，默认是Sequencefile，可用DataStream，则为普通文本
a1.sinks.k1.hdfs.fileType = DataStream
# 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
Step 5: 顺序启动
node03机器启动flume进程
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -c conf -f conf/avro-hdfs.conf -n a1 -Dflume.root.logger=INFO,console
node02机器启动flume进程
cd /export/servers/apache-flume-1.8.0-bin/
bin/flume-ng agent -c conf -f conf/tail-avro-avro-logger.conf -n a1 -Dflume.root.logger=INFO,console
node02机器启shell脚本生成文件
cd /export/servers/shells
sh tail-file.sh
3. flume的高可用方案-failover
在完成单点的Flume NG搭建后，下面我们搭建一个高可用的Flume NG集群，架构图如下所示：
3.1. 角色分配
Flume的Agent和Collector分布如下表所示：
| 名称 | HOST | 角色 |
|---|---|---|
| Agent1 | node01 | Web Server |
| Collector1 | node02 | AgentMstr1 |
| Collector2 | node03 | AgentMstr2 |
图中所示，Agent1数据分别流入到Collector1和Collector2，Flume NG本身提供了Failover机制，可以自动切换和恢复。在上图中，有3个产生日志服务器分布在不同的机房，要把所有的日志都收集到一个集群中存储。下 面我们开发配置Flume NG集群
3.2. Node01 安装和配置
将node03机器上面的flume安装包以及文件生产的两个目录拷贝到node01机器上面去
node03机器执行以下命令
cd /export/servers
scp -r apache-flume-1.8.0-bin/ node01:$PWD
scp -r shells/ taillogs/ node01:$PWD
node01机器配置agent的配置文件
cd /export/servers/apache-flume-1.8.0-bin/conf
vim agent.conf
#agent1 name
agent1.channels = c1
agent1.sources = r1
agent1.sinks = k1 k2
#
##set gruop
agent1.sinkgroups = g1
#
agent1.sources.r1.channels = c1
agent1.sources.r1.type = exec
agent1.sources.r1.command = tail -F /export/servers/taillogs/access_log
#
##set channel
agent1.channels.c1.type = memory
agent1.channels.c1.capacity = 1000
agent1.channels.c1.transactionCapacity = 100
#
## set sink1
agent1.sinks.k1.channel = c1
agent1.sinks.k1.type = avro
agent1.sinks.k1.hostname = node02
agent1.sinks.k1.port = 52020
#
## set sink2
agent1.sinks.k2.channel = c1
agent1.sinks.k2.type = avro
agent1.sinks.k2.hostname = node03
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
3.3. Node02 与 Node03 配置 FlumeCollection
node02机器修改配置文件
cd /export/servers/apache-flume-1.8.0-bin/conf
vim collector.conf
#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 = node02
a1.sources.r1.port = 52020
a1.sources.r1.channels = c1
#
##set sink to hdfs
a1.sinks.k1.type=hdfs
a1.sinks.k1.hdfs.path= hdfs://node01:8020/flume/failover/
a1.sinks.k1.hdfs.fileType=DataStream
a1.sinks.k1.hdfs.writeFormat=TEXT
a1.sinks.k1.hdfs.rollInterval=10
a1.sinks.k1.channel=c1
a1.sinks.k1.hdfs.filePrefix=%Y-%m-%d
#
node03机器修改配置文件
cd /export/servers/apache-flume-1.8.0-bin/conf
vim collector.conf
#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 = node03
a1.sources.r1.port = 52020
a1.sources.r1.channels = c1
#
##set sink to hdfs
a1.sinks.k1.type=hdfs
a1.sinks.k1.hdfs.path= hdfs://node01:8020/flume/failover/
a1.sinks.k1.hdfs.fileType=DataStream
a1.sinks.k1.hdfs.writeFormat=TEXT
a1.sinks.k1.hdfs.rollInterval=10
a1.sinks.k1.channel=c1
a1.sinks.k1.hdfs.filePrefix=%Y-%m-%d
3.4. 顺序启动
node03机器上面启动flume
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -n a1 -c conf -f conf/collector.conf -Dflume.root.logger=DEBUG,console
node02机器上面启动flume
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -n a1 -c conf -f conf/collector.conf -Dflume.root.logger=DEBUG,console
node01机器上面启动flume
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -n agent1 -c conf -f conf/agent.conf -Dflume.root.logger=DEBUG,console
node01机器启动文件产生脚本
cd /export/servers/shells
sh tail-file.sh
3.5. Failover 测试
下面我们来测试下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。具体截图如下所 示：
Collector1优先上传
[图片上传失败...(image-84445e-1566550242649)]
HDFS集群中上传的log内容预览
[图片上传失败...(image-8166de-1566550242649)]
Collector1宕机，Collector2获取优先上传权限
[图片上传失败...(image-84cead-1566550242649)]
重启Collector1服务，Collector1重新获得优先上传的权限
4. flume 的负载均衡
负载均衡是用于解决一台机器(一个进程)无法解决所有请求而产生的一种算法。Load balancing Sink Processor 能够实现 load balance 功能，如下图Agent1 是一个路由节点，负责将 Channel 暂存的 Event 均衡到对应的多个 Sink组件上，而每个 Sink 组件分别连接到一个独立的 Agent 上，示例配置，如下所示：
在此处我们通过三台机器来进行模拟flume的负载均衡
三台机器规划如下：
node01：采集数据，发送到node02和node03机器上去
node02：接收node01的部分数据
node03：接收node01的部分数据
第一步：开发node01服务器的flume配置
node01服务器配置：
cd /export/servers/apache-flume-1.8.0-bin/conf
vim load_banlancer_client.conf
#agent name
a1.channels = c1
a1.sources = r1
a1.sinks = k1 k2
#set gruop
a1.sinkgroups = g1
#set channel
a1.channels.c1.type = memory
a1.channels.c1.capacity = 1000
a1.channels.c1.transactionCapacity = 100
a1.sources.r1.channels = c1
a1.sources.r1.type = exec
a1.sources.r1.command = tail -F /export/servers/taillogs/access_log
# set sink1
a1.sinks.k1.channel = c1
a1.sinks.k1.type = avro
a1.sinks.k1.hostname = node02
a1.sinks.k1.port = 52020
# set sink2
a1.sinks.k2.channel = c1
a1.sinks.k2.type = avro
a1.sinks.k2.hostname = node03
a1.sinks.k2.port = 52020
#set sink group
a1.sinkgroups.g1.sinks = k1 k2
#set failover
a1.sinkgroups.g1.processor.type = load_balance
a1.sinkgroups.g1.processor.backoff = true
a1.sinkgroups.g1.processor.selector = round_robin
a1.sinkgroups.g1.processor.selector.maxTimeOut=10000
第二步：开发node02服务器的flume配置
cd /export/servers/apache-flume-1.8.0-bin/conf
vim load_banlancer_server.conf
# Name the components on this agent
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# Describe/configure the source
a1.sources.r1.type = avro
a1.sources.r1.channels = c1
a1.sources.r1.bind = node02
a1.sources.r1.port = 52020
# 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
第三步：开发node03服务器flume配置
node03服务器配置
cd /export/servers/apache-flume-1.8.0-bin/conf
vim load_banlancer_server.conf
# Name the components on this agent
a1.sources = r1
a1.sinks = k1
a1.channels = c1
# Describe/configure the source
a1.sources.r1.type = avro
a1.sources.r1.channels = c1
a1.sources.r1.bind = node03
a1.sources.r1.port = 52020
# 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
第四步：准备启动flume服务
启动node03的flume服务
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -n a1 -c conf -f conf/load_banlancer_server.conf -Dflume.root.logger=DEBUG,console
启动node02的flume服务
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -n a1 -c conf -f conf/load_banlancer_server.conf -Dflume.root.logger=DEBUG,console
启动node01的flume服务
cd /export/servers/apache-flume-1.8.0-bin
bin/flume-ng agent -n a1 -c conf -f conf/load_banlancer_client.conf -Dflume.root.logger=DEBUG,console
第五步：node01服务器运行脚本产生数据
cd /export/servers/shells
sh tail-file.sh