日志分析平台ELK之日志收集器filebeat
前面我们了解了elk集群中的logstash的用法,使用logstash处理日志挺好的,但是有一个缺陷,就是太慢了;当然logstash慢的原因是它依赖jruby虚拟机,jruby虚拟机就是用java语言开发的ruby虚拟机,本身java程序运行在jvm上就已经很慢了,而logstash还要运行在用java语言开发的ruby虚拟机上,就相当于虚拟机上跑一个虚拟机,可想而知;如果我们只需要收集和处理日志,在agent端如果运行logstash,显得格外的消耗资源;为了解决这种问题,elastic开发了一款更加轻量级的日志收集器beats;而filebeat只是其中的一种,它是基于收集本地日志文件中的内容,然后输出到某个地方;中间不会对日志做过多的处理;有点类似rsyslog,只做日志转发;
前面我们了解了elk集群中的logstash的用法,使用logstash处理日志挺好的,但是有一个缺陷,就是太慢了;当然logstash慢的原因是它依赖jruby虚拟机,jruby虚拟机就是用java语言开发的ruby虚拟机,本身java程序运行在jvm上就已经很慢了,而logstash还要运行在用java语言开发的ruby虚拟机上,就相当于虚拟机上跑一个虚拟机,可想而知;如果我们只需要收集和处理日志,在agent端如果运行logstash,显得格外的消耗资源;为了解决这种问题,elastic开发了一款更加轻量级的日志收集器beats;而filebeat只是其中的一种,它是基于收集本地日志文件中的内容,然后输出到某个地方;中间不会对日志做过多的处理;有点类似rsyslog,只做日志转发;如果我们需要对日志做处理,我们可以把filebeat的输出源配置成logstash,让logstash运行在一个独立的服务器上,专门做日志处理;
filebeat收集日志过程
提示:以上是filebeat收集日志,然后把日志转发给logstash进行分析,然后logstash把filebeat发送过来的日志,做切词,分析,处理以后,然后在把日志发送给elasticsearch存储;
提示:如果后端的filebeat一旦增多,logstash的压力会非常大,为了解决这样的问题,我们可在中间加redis是做临时缓存;然后logstash就到redis里读日志;然后再把读到的日志存储到elasticsearch中;当然filebeat也是可以直接将日志数据发送给elasticsearch进行存储;
filebeat安装
下载同elasticsearch版本的filebeat rpm包
[root@node03 ~]# wget https://artifacts.elastic.co/downloads/beats/filebeat/filebeat-6.8.12-x86_64.rpm --2020-10-04 14:03:03-- https://artifacts.elastic.co/downloads/beats/filebeat/filebeat-6.8.12-x86_64.rpm Resolving artifacts.elastic.co (artifacts.elastic.co)... 151.101.230.222, 2a04:4e42:36::734 Connecting to artifacts.elastic.co (artifacts.elastic.co)|151.101.230.222|:443... connected. HTTP request sent, awaiting response... 200 OK Length: 11904164 (11M) [application/octet-stream] Saving to: ‘filebeat-6.8.12-x86_64.rpm’ 100%[================================================================================>] 11,904,164 9.76KB/s in 16m 35s 2020-10-04 14:19:41 (11.7 KB/s) - ‘filebeat-6.8.12-x86_64.rpm’ saved [11904164/11904164] [root@node03 ~]# ll total 184540 -rw-r--r-- 1 root root 11904164 Aug 18 19:35 filebeat-6.8.12-x86_64.rpm -rw-r--r-- 1 root root 177059640 Aug 18 19:41 logstash-6.8.12.rpm [root@node03 ~]#
安装filebeat-6.8.12.rpm包
[root@node03 ~]# yum install ./filebeat-6.8.12-x86_64.rpm -y Loaded plugins: fastestmirror Examining ./filebeat-6.8.12-x86_64.rpm: filebeat-6.8.12-1.x86_64 Marking ./filebeat-6.8.12-x86_64.rpm to be installed Resolving Dependencies --> Running transaction check ---> Package filebeat.x86_64 0:6.8.12-1 will be installed --> Finished Dependency Resolution Dependencies Resolved ========================================================================================================================== Package Arch Version Repository Size ========================================================================================================================== Installing: filebeat x86_64 6.8.12-1 /filebeat-6.8.12-x86_64 38 M Transaction Summary ========================================================================================================================== Install 1 Package Total size: 38 M Installed size: 38 M Downloading packages: Running transaction check Running transaction test Transaction test succeeded Running transaction Installing : filebeat-6.8.12-1.x86_64 1/1 Verifying : filebeat-6.8.12-1.x86_64 1/1 Installed: filebeat.x86_64 0:6.8.12-1 Complete! [root@node03 ~]#
示例:配置filebeat收集httpd的日志,然后将收集的日志输出到logstash
提示:以上配置表示开启filebeat插件收集/var/log/httpd/access_log中的日志;
提示:以上配置表示把filebeat收集的日志发送给node03:5044;
配置node03的logstash输入数据监听5044端口
提示:以上配置表示启动logstash中的beats插件作为数据输入,并监听5044端口;然后logstash将处理后端日志数据输出到标准输出;
启动filebeat和logstash
提示:可以看到logstash启动时,它监听5044端口;
用其他主机模拟互联网用户访问node03的httpd提供的页面
[root@node01 ~]# curl -H "X-Forwarded-For:$[$RANDOM%223+1].$[RANDOM%255].$[RANDOM%255].$[RANDOM%255]" http://node03/test$[$RANDOM%20+1].html page 18 [root@node01 ~]#
在node03的logstash的标准输出上,看看是否收集到httpd的访问日志?
/usr/share/logstash/vendor/bundle/jruby/2.5.0/gems/awesome_print-1.7.0/lib/awesome_print/formatters/base_formatter.rb:31: warning: constant ::Fixnum is deprecated { "host" => { "os" => { "platform" => "centos", "version" => "7 (Core)", "family" => "redhat", "name" => "CentOS Linux", "codename" => "Core" }, "containerized" => false, "architecture" => "x86_64", "name" => "node03.test.org", "id" => "002f3e572e3e4886ac9e98db8584b467" }, "prospector" => { "type" => "log" }, "auth" => "-", "clientip" => "25.99.168.124", "agent" => "\"curl/7.29.0\"", "tags" => [ [0] "beats_input_codec_plain_applied" ], "@timestamp" => 2020-10-04T06:49:34.000Z, "@version" => "1", "bytes" => "8", "offset" => 0, "verb" => "GET", "referrer" => "\"-\"", "source" => "/var/log/httpd/access_log", "log" => { "file" => { "path" => "/var/log/httpd/access_log" } }, "clientipInfo" => { "continent_code" => "EU", "longitude" => -0.1224, "country_code2" => "GB", "ip" => "25.99.168.124", "country_name" => "United Kingdom", "country_code3" => "GB", "location" => { "lat" => 51.4964, "lon" => -0.1224 }, "timezone" => "Europe/London", "latitude" => 51.4964 }, "beat" => { "hostname" => "node03.test.org", "version" => "6.8.12", "name" => "node03.test.org" }, "request" => "/test18.html", "input" => { "type" => "log" }, "ident" => "-", "response" => "200", "httpversion" => "1.1" }
提示:在node03的标准输出上能够看到我们刚才访问httpd的访问日志;
示例:配置filebeat将日志输出到elasticsearch
重启filebeat
验证:访问httpd看看elasticsearch中是否有保存httpd的访问日志?
在elasticsearch中查看是否有新的index生成?
提示:可以看到es上有一个新的index生成;
查看es上存储的日志内容
提示:从上面的返回的日志,存放在es中的日志并没有做拆分,说明filebeat只是把httpd的日志当作message字段的值处理,并没有去把ip地址信息做拆分;所以要想实现把日志内容拆分成不同字段,我们可以借助logstash,当然也可以在httpd上直接将日志格式记录为json格式,然后再由filebeat将日志信息传给es存储;
示例:配置filebeat将收集的日志信息输出到redis
提示:以上配置是配置filebeat将收集到的日志输出到redis;这里需要注意一点,这个配置文件是yml格式的文件,需要注意下面的缩进关系要对其;其次filebeat它不支持多路输出,例如,配置filebeat 输出到logstash,又配置filebeat输出到redis,这个它不支持,它支持单路输出;
重新启动filebeat
用其他主机模拟访问httpd
验证:去node04上的redis 3号库查看是否有key生成?是否有数据?
提示:可以看到此时redis的3号库有指定key生成,对应key里面也存了httpd的访问日志;
配置logstash到redis上读数据,并删除filebeat的冗余字段
[root@node03 ~]# cat /etc/logstash/conf.d/httpd-es.conf input { redis { host => ["node04"] port => 6379 password => "admin" key => "filebeat-node03-httpd-access_log" db => 3 data_type => "list" } } filter { grok { match => {"message" => "%{HTTPD_COMBINEDLOG}" } remove_field => "message" } date { match => ["timestamp","dd/MMM/YYYY:H:m:s Z"] remove_field => "timestamp" } geoip { source => "clientip" target => "geoip" database => "/etc/logstash/geoip/GeoLite2-City.mmdb" } mutate { rename => ["geoip", "clientipInfo" ] remove_field => ["@metadata","prospector","input","beat","host","id","containerized"] } } output { # elasticsearch { # hosts => ["http://node01:9200","http://node02:9200"] # index => "httpd.log" # codec => "rubydebug" # } stdout { codec => "rubydebug" } } [root@node03 ~]#
测试语法
启动logstash
查看输出到标准输出的日志信息是否还有filebeat生成的多余字段?
提示:现在从redis读出来的数据,然后经由logstash处理以后,filebeat生成的多余字段就删除了;后续我们就可以直接将这些日志数据放到es中存储;