Nginx+Flume+Hadoop日志分析,Ngram+AutoComplete

配置Nginx

  • yum install nginx (在host99和host101)
  • service nginx start开启服务
  • ps -ef |grep nginx看一下进程
ps -ef |grep nginx
root     28230     1  0 14:54 ?        00:00:00 nginx: master process /usr/sbin/nginx -c /etc/nginx/nginx.conf
nginx    28231 28230  0 14:54 ?        00:00:00 nginx: worker process                   
nginx    28232 28230  0 14:54 ?        00:00:00 nginx: worker process                   
nginx    28234 28230  0 14:54 ?        00:00:00 nginx: worker process                   
nginx    28235 28230  0 14:54 ?        00:00:00 nginx: worker process 
....
View Code
  • 在本机浏览器也可以通过ip + port(默认80端口)访问

HTML

  • 很简单的网页配置,在/root/html下放了一个index.html文件
  • 先test一下是否正常,在browser里面输入ip地址就可以看到了。
  • 下一步配置一个输入框,模拟搜索数据。
  • index.html是特别简单的代码,(我是前端渣- -
  • <!DOCTYPE html>
    <html>
        <body>
    
            <form action="/result.html" method="GET">
                Please Input:<br>
                <input type="text" name="Input" value="Mickey">
                <br>
                <br>
                <input type="submit" value="Submit">
            </form> 
    
        </body>
    </html>
    View Code
  • 跳转到的页面result.html就随便写一句话啦

log日志

  • log日志放在/var/log/nginx目录的access.log中。
  • tail -f access.log

    在输入框输入数据,提交之后会看到log刷新,比如我输入“test”提交后log会刷新一条:

  • 10.109.255.90 - - [11/May/2017:14:24:16 +0800] "GET /action_page.php?Input=test HTTP/1.1" 404 169 "http://10.3.242.101/" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/602.4.8 (KHTML, like Gecko) Version/10.0.3 Safari/602.4.8"

配置Flume

  • 安装:(在host101)
  • wget http://mirrors.tuna.tsinghua.edu.cn/apache/flume/1.7.0/apache-flume-1.7.0-bin.tar.gz
    tar -xvf apache-flume-1.7.0-bin.tar.gz
  • 测试:
    • 在conf文件夹下新建一个test.conf文件如下:
    • # 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
      View Code
    • 然后,运行
    • bin/flume-ng agent -n a1 -c conf -f conf/test.conf
    • 然后在logs文件夹下运行如下,刷新log
    • tail -f flume.log
    • 然后telnet localhost 44444。随便输入一些数据。log刷新显示出
    • 11 May 2017 15:26:55,372 INFO  [SinkRunner-PollingRunner-DefaultSinkProcessor] (org.apache.flume.sink.LoggerSink.process:95)  - Event: { headers:{} body: 64 61 6B 6A 64 63 0D                            dakjdc. }
    • 就没问题啦

Nginx(WebServer)端 

  • Nginx端配置source-side flume agent来收集访问日志流数据
  • WebAccLog.sources = NginxAccess
    WebAccLog.sinks = AvroSink
    WebAccLog.channels = MemChannel
    
    WebAccLog.sources.NginxAccess.type = exec
    
    WebAccLog.sources.NginxAccess.command = tail -f /var/log/nginx/access.log
    WebAccLog.sources.NginxAccess.batchSize = 10
    
    WebAccLog.sources.NginxAccess.interceptors = itime
    WebAccLog.sources.NginxAccess.interceptors.itime.type = timestamp
    
    WebAccLog.sinks.AvroSink.type = avro
    WebAccLog.sinks.AvroSink.hostname = 10.3.242.99
    WebAccLog.sinks.AvroSink.port = 4545
    
    WebAccLog.channels.MemChannel.type = memory
    WebAccLog.sinks.AvroSink.channel = MemChannel
    WebAccLog.sources.NginxAccess.channels = MemChannel
  • 注意到我们在这里配置对source配置了一下interceptor拦截器。flume中source采集的日志首先会传入ChannelProcessor,在其内首先会通过interceptors进行过滤加工,然后通过ChannelSelector选择channel。
  • 这里配置了一个时间戳拦截器,后面会在指定hdfs.path的时候用到(根据时间戳来指定存放路径)。
  • 关于拦截器:flume内部实现了很多拦截器,同时还是先虑InterceptorChain用来链式处理event。
    • HostInterceptor:在所有拦截的events的header中上加上本机的host name或IP
    • TimestampInterceptor:在所有拦截的events的header中上加上它处理该时间的时间(in millis)。
  • 这里之前写的时候source忘记加s了,然后查一下log才发现的问题,说实话这个conf文件好容易眼花,要改的东西太多- -。

Hadoop/Spark cluster端

  • Hadoop/Spark cluster端配置receiver- side flume agent来接收数据.
  • 首先使用logger sink来测试下是否一切正常
    • 启动cluster端的flume,然后在web server端运行
    • bin/flume-ng avro-client -c ./conf -H 10.3.242.99 -p 4545 -F /var/log/nginx/access.log
    • 然后cluster端的flume会在log里面打出一堆数据,说明可以了。
  • 接下来,把两边连通
  • 下面是一个HDFS sink的例子,其他需求可以通过修改sink部分实现。
  • # clusterLogAgent
    
    # Naming the components of the current agent.
    clusterLogAgent.sources = AvroSource
    clusterLogAgent.sinks = HDFS
    clusterLogAgent.channels = MemChannel
    
    # Source configuration
    clusterLogAgent.sources.AvroSource.type = avro
    # hostname or IP address to listen on
    clusterLogAgent.sources.AvroSource.bind = 0.0.0.0
    clusterLogAgent.sources.AvroSource.port = 4545
    
    # sink configuration(write to HDFS)
    clusterLogAgent.sinks.HDFS.type = hdfs
    clusterLogAgent.sinks.HDFS.hdfs.path = /logFlume/nginx/accesslog
    # File format: currently SequenceFile, DataStream or CompressedStream
    clusterLogAgent.sinks.HDFS.hdfs.fileType = DataStream
    # Number of events written to file before it rolled (0 = never roll based on number of events)
    clusterLogAgent.sinks.HDFS.hdfs.rollCount = 0
    
    clusterLogAgent.channels.MemChannel.type = memory
    
    clusterLogAgent.sources.AvroSource.channels = MemChannel
    clusterLogAgent.sinks.HDFS.channel = MemChannel

启动

  • 首先开启cluster上的flume服务(这里要注意开启顺序
  • ./bin/flume-ng agent -n clusterLogAgent -c conf -f conf/flume-hdfsSink.conf
  • 开启Nginx端的flume服务
  • bin/flume-ng agent -n WebAccLog -c conf -f conf/flume-avro.conf
  • check HDFS来验证流访问log events是否成功写入
  • [root@host99 /home/hhh/apache-flume-1.7.0-bin/conf]$hadoop fs -ls /logFlume/nginx/accesslog
    17/05/11 17:25:47 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
    Found 3 items
    -rw-r--r--   2 root supergroup       1146 2017-05-11 17:15 /logFlume/nginx/accesslog/FlumeData.1494494113448
    -rw-r--r--   2 root supergroup       1190 2017-05-11 17:15 /logFlume/nginx/accesslog/FlumeData.1494494113449
    -rw-r--r--   2 root supergroup        952 2017-05-11 17:16 /logFlume/nginx/accesslog/FlumeData.1494494113450

  • [root@host101 /home/hhh/apache-flume-1.7.0-bin/conf]$hadoop fs -cat /logFlume/nginx/accesslog/FlumeData.1494494113448
    17/05/11 17:19:59 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
    10.109.255.90 - - [11/May/2017:16:56:35 +0800] "GET /result.html?Input=ttt HTTP/1.1" 200 23 "-" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/602.4.8 (KHTML, like Gecko) Version/10.0.3 Safari/602.4.8"
    10.109.255.90 - - [11/May/2017:16:56:45 +0800] "GET /result.html?Input=ttt HTTP/1.1" 200 23 "-" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/602.4.8 (KHTML, like Gecko) Version/10.0.3 Safari/602.4.8"
    10.109.255.90 - - [11/May/2017:16:57:04 +0800] "GET /result.html?Input=hhhhhh HTTP/1.1" 200 23 "http://10.3.242.101/" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/602.4.8 (KHTML, like Gecko) Version/10.0.3 Safari/602.4.8"
    10.109.255.90 - - [11/May/2017:16:57:07 +0800] "GET /result.html?Input=hhhhhh HTTP/1.1" 200 23 "http://10.3.242.101/" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/602.4.8 (KHTML, like Gecko) Version/10.0.3 Safari/602.4.8"

Flume conf调优

可优化的配置:

  • Avro Sink:
    • batch-size: 一次发送多少数据,默认是10
    • compression-type:压缩类型,默认不压缩。
    • compression-level: 压缩等级
    • maxIoWorkers:最大的I/O worker线程个数,默认2*机器最大可用处理器数目
  • Avro Source:

    • threads:最大worker线程数

  • HDFS sink:

    • HDFS sink支持创建text和sequence文件;支持压缩;

    • 文件可以被rolled。rolled基于elapse time或数据size或events数量

    • 可以根据events的属性(比如时间戳或机器)来划分数据(buckets/partitions data)
    • HDFS路径可以包含格式化转移序列(formatiting escape sequences),HDFS可以据其产生一个目录/文件名来存储events。如下:
    • AliasDescription
      %{host} Substitute value of event header named “host”. Arbitrary header names are supported.
      %t Unix time in milliseconds
      %a locale’s short weekday name (Mon, Tue, ...)
      %A locale’s full weekday name (Monday, Tuesday, ...)
      %b locale’s short month name (Jan, Feb, ...)
      %B locale’s long month name (January, February, ...)
      %c locale’s date and time (Thu Mar 3 23:05:25 2005)
      %d day of month (01)
      %e day of month without padding (1)
      %D date; same as %m/%d/%y
      %H hour (00..23)
      %I hour (01..12)
      %j day of year (001..366)
      %k hour ( 0..23)
      %m month (01..12)
      %n month without padding (1..12)
      %M minute (00..59)
      %p locale’s equivalent of am or pm
      %s seconds since 1970-01-01 00:00:00 UTC
      %S second (00..59)
      %y last two digits of year (00..99)
      %Y year (2010)
      %z +hhmm numeric timezone (for example, -0400)
      %[localhost] Substitute the hostname of the host where the agent is running
      %[IP] Substitute the IP address of the host where the agent is running
      %[FQDN] Substitute the canonical hostname of the host where the agent is running
    • 一些可能的优化配置:
      • hdfs.rollSize:触发roll的文件大小,in bytes,默认是1024
      • hdfs.rollCount: 在roll之前写入文件的events数目,默认是10。设置为0的话就不会根据events数量roll了。
      • hdfs.rollInterval:roll当前文件所等待的秒数,默认是9=30。设置为0的话就不会根据时间间隔roll了。
      • hdfs.batchSize: 在flushed到HDFS之前写入到文件的events数目
      • hdfs.threadsPoolSize:每个HDFS sink中HDFS IO操作的线程数目,默认10

已完成的优化

  • hdfs文件目录:按日期分目录
  • # flume-hdfs.sink

    clusterLogAgent.sinks.HDFS.hdfs.path = /logFlume/nginx/%y.%m.%d/

     然后就变成啦酱紫:

  

  • 配置滚动文件的大小,避免产生一堆小文件
  • #clusterLogAgent.sinks.HDFS.hdfs.rollSize =64*1024*1024
    clusterLogAgent.sinks.HDFS.hdfs.rollSize = 67108864
    clusterLogAgent.sinks.HDFS.hdfs.rollCount = 0 
    clusterLogAgent.sinks.HDFS.hdfs.rollInterval = 0

TBD

  • 压缩
  • 线程、worker数目
  • ...
  • flume内部是可以做一些过滤的。基于当前场景的话,是要用到正则的,主要是考虑这样的话会不会很影响效率。因为flume内部也不适合做复杂的过滤。

Log Analyse

 搜索词分析-Ngram

  • 提取搜索词:正则。
  • 10.109.255.90 - - [12/May/2017:14:31:46 +0800] "GET /result.html?Input=Mickey HTTP/1.1" 304 0 "http://10.3.242.101/" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_3) AppleWebKit/602.4.8 (KHTML, like Gecko) Version/10.0.3 Safari/602.4.8"
    10.30.146.74 - - [12/May/2017:14:33:55 +0800] "GET / HTTP/1.1" 304 0 "-" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/55.0.2883.87 Safari/537.36"
    

    可以看到就是要提取 "?Input=Mickey"这一部分。用正则“?Input=[^s]*s”来匹配这段。

   还有一点就是看到有的log是没有input,状态是304的。这种请求是因为用了浏览器缓存。考虑到搜索结果不能是固定的,所以在conf里面加了一句add_header Cache-Control no-store;来禁用缓存

  • 在mapper的中建Pattern,而不是在map()函数中每次初始化
  • 用搜索词做Ngram,这里直接假设搜索词已经是'+'-split(在搜索框space分隔的数据会在log里变成+分隔的)。【TODO:后续可以实现切分
  • MapReduce实现:
    • Job1: SplitNgram --> 正则匹配搜索词,切分搜索词,统计ngram出现次数;
    • Job2: GetNgram --> 过滤count小于指定threshold的ngram,同时得到每个origin词联想到的topK。

 模拟搜索

  • nili小可爱真是可怜又坚强 什么也没有 零数据支持
  • 模拟的方式简单,就写一个程序(python or java or scala or whatever),就好啦
  • 但是要稍微有点意义的搜索词,目前想到的方法就是找一堆数据,脚本一边读一边向nginx请求。
  • 模拟http请求的demo如下:
    • import httplib
      
      coon = httplib.HTTPConnection('host101')
      word = 'test'
      conn.request('GET', 'result.html?input=' + word)
  • 以及搜索数据目前能找到的方式就是,我在kaggle上看到的一个quora的比赛,可以拿到一些quora的问题,这在某种程度上也是搜索数据了吧。hhh自我满足ing...[所以其实后续想再扩展的话可以考虑爬quora呀23333]
  • 加上搜索数据的整个模拟demo如下:【注:连接的效率等问题暂未考虑..
    • def get_search_word(file_name):
          # id, qid1, qid2, question1, question2, is_duplicate
          with open(file_name) as fi:
              for line in fi:
                  splited = line.split(",")
                  if len(splited) < 6:
                      continue
                  conn = httplib.HTTPConnection('10.3.242.101')
                  conn.request('GET', 'result.html?input=' + splited[3])
                  # time.sleep(5)
                  conn = httplib.HTTPConnection('10.3.242.101')
                  conn.request('GET', 'result.html?input=' + splited[4])
                  time.sleep(5)
      
满地都是六便士,她却抬头看见了月亮。
原文地址:https://www.cnblogs.com/wttttt/p/6829807.html