本文共 9383 字,大约阅读时间需要 31 分钟。
grok是一个十分强大的Logstash filter插件,它可以通过正则解析任意文本,将非结构化日志数据弄成结构化和方便查询的结构。它是目前Logstash中解析非结构化日志数据最好的方式
**grok的语法规则:** ```%{语法:语义}```“语法”指的是匹配的模式。例如使用NUMBER模式可以匹配出数字,IP模式则会匹配出127.0.0.1这样的IP地址**举例:**原始数据:172.16.213.132 [07/Feb/2018:16:24:19 +0800] "GET /HTTP/1.1" 403 50391)过滤IP
input { stdin { }}filter{ grok{ match => {"message" => "%{IPV4:ip}"} }}output { stdout { }}
运行结果:
[root@:172.31.22.29 /etc/logstash/conf.d]#/usr/share/logstash/bin/logstash -f /etc/logstash/conf.d/l2.confSending Logstash logs to /var/log/logstash which is now configured via log4j2.properties172.16.213.132 [07/Feb/2018:16:24:19 +0800]"GET /HTTP/1.1" 403 5039 #手动输入此行信息{ "message" => "172.16.213.132 [07/Feb/2018:16:24:19 +0800]\"GET /HTTP/1.1\" 403 5039", "ip" => "172.16.213.132", "@version" => "1", "host" => "ip-172-31-22-29.ec2.internal", "@timestamp" => 2019-01-22T09:48:15.354Z}
2)过滤时间戳
filter{ grok{ match => {"message" => "%{IPV4:ip}\ \[%{HTTPDATE:timestamp}\]"} }}
运行结果:
[root@:172.31.22.29 /etc/logstash/conf.d]#/usr/share/logstash/bin/logstash -f /etc/logstash/conf.d/l2.confSending Logstash logs to /var/log/logstash which is now configured via log4j2.properties172.16.213.132 [07/Feb/2018:16:24:19 +0800]"GET /HTTP/1.1" 403 5039 手动输入此行信息{ "@version" => "1", "timestamp" => "07/Feb/2018:16:24:19 +0800", "@timestamp" => 2019-01-22T10:16:14.205Z, "message" => "172.16.213.132 [07/Feb/2018:16:24:19 +0800]\"GET /HTTP/1.1\" 403 5039", "ip" => "172.16.213.132", "host" => "ip-172-31-22-29.ec2.internal"}
可以看到我们已经过滤成功了,在配置文件中grok其实是使用正则表达式来进行过滤的。我们做个小实验,比如我现在在例子中的数据ip后面添加两个“-”。如图所示:
172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] “GET /HTTP/1.1” 403 5039
那么此时在配置文件中我就需要这样子来写:
filter{ grok{ match => {"message" => "%{IPV4:ip}\ -\ -\ \[%{HTTPDATE:timestamp}\]"} }}
那么此时在match行我就要匹配两个“-”,否则grok就不能正确匹配数据,从而不能解析数据
运行结果:
[root@:172.31.22.29 /etc/logstash/conf.d]#/usr/share/logstash/bin/logstash -f /etc/logstash/conf.d/l2.confSending Logstash logs to /var/log/logstash which is now configured via log4j2.properties172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] "GET /HTTP/1.1" 403 5039 #手动输入此行内容,然后按下enter键。{ "@timestamp" => 2019-01-22T10:25:46.687Z, "ip" => "172.16.213.132", "message" => "172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] \"GET /HTTP/1.1\" 403 5039", "timestamp" => "07/Feb/2018:16:24:19 +0800", "@version" => "1", "host" => "ip-172-31-22-29.ec2.internal"}
这时候我们就得到了信息,我这里是匹配IP和时间,当然你也可以直接匹配时间即可:
filter{ grok{ match => {"message" => "\ -\ -\ \[%{HTTPDATE:timestamp}\]"} }}
这个时候我们更加能理解grok使用正则匹配数据了
注意:正则中,匹配空格和中括号要加上转义符3)过滤出报文头信息
首先来写匹配的正则模式filter{ grok{ match => {"message" => "\ %{QS:referrer}\ "} }}
运行结果:
[root@:172.31.22.29 /etc/logstash/conf.d]#/usr/share/logstash/bin/logstash -f /etc/logstash/conf.d/l2.confSending Logstash logs to /var/log/logstash which is now configured via log4j2.properties172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] "GET /HTTP/1.1" 403 5039{ "@timestamp" => 2019-01-22T10:47:37.127Z, "message" => "172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] \"GET /HTTP/1.1\" 403 5039", "@version" => "1", "host" => "ip-172-31-22-29.ec2.internal", "referrer" => "\"GET /HTTP/1.1\""}
在上面我们有个例子是讲解timestamp字段,表示取出日志中的时间。但是在显示的时候除了显示你指定的timestamp外,还有一行是@timestamp信息,这两个时间是不一样的,@timestamp表示系统当前时间。两个时间并不是一回事,在ELK的日志处理系统中,@timestamp字段会被elasticsearch用到,用来标注日志的生产时间,如此一来,日志生成时间就会发生混乱,要解决这个问题,需要用到另一个插件,即date插件,这个时间插件用来转换日志记录中的时间字符串,变成Logstash::Timestamp对象,然后转存到@timestamp字段里面
接下来我们在配置文件中配置一下:
filter{ grok{ match => {"message" => "\ -\ -\ \[%{HTTPDATE:timestamp}\]"} } date{ match => ["timestamp","dd/MMM/yyyy:HH:mm:ss Z"] }}
注意:时区偏移量需要用一个字母Z来转换。还有这里的“dd/MMM/yyyy”,你发现中间是三个大写的M,没错,这里确实是三个大写的M,我尝试只写两个M的话,转换失败
运行结果:
[root@:172.31.22.29 /etc/logstash/conf.d]#/usr/share/logstash/bin/logstash -f /etc/logstash/conf.d/l2.confSending Logstash logs to /var/log/logstash which is now configured via log4j2.properties172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] "GET /HTTP/1.1" 403 5039 #手动输入此行信息{ "host" => "ip-172-31-22-29.ec2.internal", "timestamp" => "07/Feb/2018:16:24:19 +0800", "@timestamp" => 2018-02-07T08:24:19.000Z, "message" => "172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] \"GET /HTTP/1.1\" 403 5039", "@version" => "1"}
会发现@timestamp时间转换成功,还有一点就是在时间少8个小时
上面有几个例子已经讲到了date的用法。date插件对于排序事件和回填旧数据尤其重要,它可以用来转换日志记录中的时间字段,变成Logstash::timestamp对象,然后转存到@timestamp字段里面。
为什么要使用这个插件呢?
如果你要解析你的时间,你要使用字符来代替,用于解析日期和时间文本的语法使用字母来指示时间(年、月、日、时、分等)的类型。以及重复的字母来表示该值的形式。在上面看到的"dd/MMM/yyy:HH:mm:ss Z",他就是使用这种形式,我们列出字符的含义:
那我们是依据什么写出“dd/MMM/yyy:HH:mm:ss Z”这样子的形式的呢?比如上面的试验数据是:172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] “GET /HTTP/1.1” 403 5039
现在我们想转换时间,那就要写出"dd/MMM/yyy:HH:mm:ss Z",你发现中间有三个M,你要是写出两个就不行了,因为我们查表发现两个大写的M表示两位数字的月份,可是我们要解析的文本中,月份则是使用简写的英文,所以只能去找三个M。还有最后为什么要加上个大写字母Z,因为要解析的文本中含有“+0800”时区偏移,因此我们要加上去,否则filter就不能正确解析文本数据,从而转换时间戳失败remove_field的用法也是很常见的,其作用是去重,在前面的例子中你也看到了,不管是我们要输出什么样子的信息,都是有两份数据,即message里面是一份,HTTPDATE或者IP里面也有一份,这样子就造成了重复,过滤的目的就是筛选出有用的信息,重复的不要,因此我们看看如何去重呢?
1)输出IP为例
filter{ grok{ match => {"message" => "%{IP:ip_address}"} remove_field => ["message"] } }
运行结果:
[root@:172.31.22.29 /etc/logstash/conf.d]#/usr/share/logstash/bin/logstash -f /etc/logstash/conf.d/l5.confSending Logstash logs to /var/log/logstash which is now configured via log4j2.properties172.16.213.132 - - [07/Feb/2018:16:24:19 +0800] "GET /HTTP/1.1" 403 5039 #手动输入此行内容并按enter键{ "ip_address" => "172.16.213.132", "host" => "ip-172-31-22-29.ec2.internal", "@version" => "1", "@timestamp" => 2019-01-22T12:16:58.918Z}
这时候你会发现没有之前显示的那个message的那一行信息了。因为我们使用remove_field把他移除了
mutate插件是logstash另一个非常重要的插件,它提供了丰富的基础类型数据处理能力,包括重命名、删除、替换、修改日志事件中的字段。我们这里举几个常用的mutate插件:字段类型转换功能covert、正则表达式替换字段功能gsub、分隔符分隔字符串为数值功能split、重命名字段功能rename、删除字段功能remove_field
1)字段类型转换convert
先修改配置文件filter{ grok{ match => {"message" => "%{IPV4:ip}"} remove_field => ["message"] } mutate{ convert => ["ip","string"] }}
或者这样子写也行,写法区别较小:
filter{ grok{ match => {"message" => "%{IPV4:ip}"} remove_field => ["message"] } mutate{ convert => { "ip" => "string" } }}
运行结果:
[root@:172.31.22.29 /etc/logstash/conf.d]#/usr/share/logstash/bin/logstash -f /etc/logstash/conf.d/l6.confSending Logstash logs to /var/log/logstash which is now configured via log4j2.properties172.16.213.132 - - [07/Feb/2018:16:24:9 +0800] "GET /HTTP/1.1" 403 5039{ "@timestamp" => 2019-01-23T04:13:55.261Z, "ip" => "172.16.213.132", "host" => "ip-172-31-22-29.ec2.internal", "@version" => "1"}
在这里的ip行中,效果可能不太明显,但是确实是已经转化成string模式了
2)正则表达式替换匹配字段
gsub可以通过正则表达式替换字段中匹配到的值,但是这本身只对字符串字段有效首先修改配置文件
filter{ grok{ match => {"message" => "%{QS:referrer}"} remove_field => ["message"] } mutate{ gsub => ["referrer","/","-"] }}
运行结果:
172.16.213.132 - - [07/Feb/2018:16:24:9 +0800] "GET /HTTP/1.1" 403 5039{ "host" => "ip-172-31-22-29.ec2.internal", "@timestamp" => 2019-01-23T05:51:30.786Z, "@version" => "1", "referrer" => "\"GET -HTTP-1.1\""}
很不错,确实对QS的部分的分隔符换做横杠了
3)分隔符分隔字符串为数组
split可以通过指定的分隔符分隔字段中的字符串为数组首先修改配置文件
filter{ mutate{ split => ["message","-"] add_field => ["A is lower case :","%{[message][0]}"] }}
这里的意思是对一个字段按照“-”进行分隔为数组
运行结果:
a-b-c-d-e-f-g #手动输入此行内容,并按下enter键。{ "A is lower case :" => "a", "message" => [ [0] "a", [1] "b", [2] "c", [3] "d", [4] "e", [5] "f", [6] "g" ], "host" => "ip-172-31-22-29.ec2.internal", "@version" => "1", "@timestamp" => 2019-01-23T06:07:18.062Z}
4)重命名字段
rename可以实现重命名某个字段的功能filter{ grok{ match => {"message" => "%{IPV4:ip}"} remove_field => ["message"] } mutate{ convert => { "ip" => "string" } rename => { "ip"=>"IP" } }}
rename字段使用大括号{}括起来,其实我们也可以使用中括号达到同样的目的
mutate{ convert => { "ip" => "string" } rename => ["ip","IP"] }
运行结果:
172.16.213.132 - - [07/Feb/2018:16:24:9 +0800] "GET /HTTP/1.1" 403 5039 #手动输入此内容{ "@version" => "1", "@timestamp" => 2019-01-23T06:20:21.423Z, "host" => "ip-172-31-22-29.ec2.internal", "IP" => "172.16.213.132"}
5)删除字段,这个不多说,我们上面已经有例子了
6)添加字段add_field 添加字段多用于split分隔中filter { mutate { split => ["message", "|"] add_field => { "timestamp" => "%{[message][0]}" } }}
添加字段后,该字段会与@timestamp一样同等格式显示出来
转载地址:http://lrtii.baihongyu.com/