在前面一篇文章中,我们介绍了如何在linux上搭建logstash+kafka的环境(logstash+kafka),在这篇中将介绍如何连通java日志文件和logstash。
一般在java开发过程中最常用到的两种日志工具:
下面我们分别介绍这两种日志工具如何来和logstash连通,实现日志的收集。
log4j相对来说比较简单,它可以有两种配置方式:properties文件和xml文件
log4j.rootLogger=DEBUG, logstash
##SocketAppender
log4j.appender.logstash=org.apache.log4j.net.SocketAppender
##logstash等待接收的端口
log4j.appender.logstash.Port=4560
##logstash服务所在机器的ip
log4j.appender.logstash.RemoteHost=logstash_hostname
log4j.appender.logstash.ReconnectionDelay=60000
log4j.appender.logstash.LocationInfo=true
##日志准入级别
log4j.appender.logstash.Threshold=DEBUG
新增一个appender:
"LOGSTASH" class="org.apache.log4j.net.SocketAppender">
##logstash服务器IP
<param name="RemoteHost" value="logstash_hostname" />
<param name="ReconnectionDelay" value="60000" />
<param name="LocationInfo" value="true" />
##日志准入等级
<param name="Threshold" value="DEBUG" />
加入需要监控的logger中
"com.springapp.mvc" additivity="false">
class="org.apache.log4j.Level" value="debug"/>
ref ref="LOGSTASH"/>
ref ref="LOGFILE"/>
对应的logstash的conf目录中新建log4j-es.conf文件,配置如下:
input {
log4j {
host => "168.10.10.69"
port => 8801
}
}
output {
stdout {
codec => rubydebug
}
elasticsearch{
hosts => ["168.10.10.69:9200"]
index => "log4j-%{+YYYY.MM.dd}"
document_type => "log4j_type"
}
}
log4j产生json格式的数据,格式如下:
{
"message" => "Sorry, something wrong!",
"@version" => "1",
"@timestamp" => "2015-07-02T13:24:45.727Z",
"type" => "log4j-json",
"host" => "127.0.0.1:52420",
"path" => "HelloExample",
"priority" => "ERROR",
"logger_name" => "HelloExample",
"thread" => "main",
"class" => "HelloExample",
"file" => "HelloExample.java:9",
"method" => "main",
"stack_trace" => "java.lang.ArithmeticException: / by zero\n\tat HelloExample.divide(HelloExample.java:13)\n\tat HelloExample.main(HelloExample.java:7)"
}
只需要在日志配置文件中加入对应的appender即可,是不是so easy~
logback相对比较复杂,我们需要用到git上的一个开源项目logstash-logback-encoder 来进行日志的输出,首先我们引入maven依赖:
<dependency>
<groupId>net.logstash.logbackgroupId>
<artifactId>logstash-logback-encoderartifactId>
<version>4.9version>
dependency>
还要保证我们的项目中有如下的依赖:
在logback.xml文件中添加一个appender:
<appender name="logStash" class="net.logstash.logback.appender.LogstashTcpSocketAppender"> <destination>172.10.10.69:8801destination> <encoder class="net.logstash.logback.encoder.LogstashEncoder" /> <keepAliveDuration>15 minuteskeepAliveDuration> appender>
加入到需要监控的logger中:
<logger name="com.maidou.controller" level="ERROR"> <appender-ref ref="logStash"/> logger>
logstash-logback-encoder 中还提供很多种协议,多种encoder,以及多目标配置等等功能,有需要的可以自己学习README。
在logstash配置文件logback-es.conf中可以有多种交互方式,这里我们选择tcp协议:
input {
tcp {
host => "168.10.10.69"
port => 8801
mode => "server"
tags => ["tags"]
codec => json_lines
}
}
output {
stdout {
codec => rubydebug
}
elasticsearch{
hosts => ["168.10.10.69:9200"]
index => "logback-%{+YYYY.MM.dd}"
document_type => "logback_type"
}
}
按照上面的配置,输入的日志如下:
{
"@timestamp" => "2016-09-13T07:05:06.480Z",
"@version" => 1,
"message" => "HelloController exception",
"logger_name" => "com.springapp.mvc.HelloController",
"thread_name" => "http-bio-8080-exec-1",
"level" => "ERROR",
"level_value" => 40000,
"stack_trace" => "java.lang.ArithmeticException: / by zero",
"HOSTNAME" => "gejunqingdeMacBook-Pro.local",
"host" => "192.168.1.192",
"port" => 57635,
"tags" => [
[0] "tags"
]
}
配置和log4j相似,只是需要引入额外的jar包,所以繁琐一点。
至此,我们已经实现了log-->logstash-->kafka的配置,再加上邮件功能,我们的日志监控发邮件的系统就完成啦。