Log4j2 Analysis-优快云博客

本文详细对比了Log4j2与Log4j在性能上的改进，包括使用LMAX Disruptor技术实现的AsyncLogger，提供更高吞吐量和更低延迟。文中还介绍了参数化消息、运行时日志级别更改、不同Appender配置，以及如何在Web应用中减少垃圾回收。

2019独角兽企业重金招聘Python工程师标准>>>

Log4j2 improvement compare with Log4j :
   AsyncLogger :
       Implemented by LMAX Disruptor technology (a lock-free inter-thread communication library, instead of queues, resulting in higher throughput and lower latency) .
       Better performance than AsyncAppender(Internally Using JDK ArrayBlockingQueue )
   Garbage free or Low garbage when running in web application .
   Parameterized Messages

Migration
   Add bridge jar provided by log4j2
           Slf4j                   log4j-slf4j-impl.jar             route log request to log4j2
           Log4j                   log4j-1.2-api.jar (missing in our nexus repository)
           Common Logging           log4j-jcl.jar

   MDC to ThreadContext.
       To track all the logs related to a specific request by a transaction ID .
       ThreadContext.put(key,value)
       layout: %X{key}

   Change Log Level at Runtime :
       Loggng servlet
           get All Loggers : LogManager.getCurrentLoggers() ---> LoggerContext.getLoggers()
           set LogLevel : log.setLevel ---> Configurator.setLevel(loggerName, logLevel);

Appender:
   RollingFileAppender:
       -Dnode=node1
       <RollingFile name="cache_log" fileName="log/aspen-data-service-cache-${sys:node}.log"
           filePattern="log/aspen-data-service-cache-${sys:node}.log.%i">
           <PatternLayout
               pattern="[%d{yyyy-MM-dd HH:mm:ss.SSS}][%X{user}][%X{requestId}]@[%X{method}][%t] %-5p [%logger{36}:%L] - %msg%n" />
           <Policies>
               <SizeBasedTriggeringPolicy size="50MB" />
           </Policies>
           <DefaultRolloverStrategy max="10" />
       </RollingFile>

   FileAppender, RollingFileAppender, RollingRandomAccessFileAppender : (When set bufferedIO = true , log will be written to buffer . Better performance . Downside : buffer data might be lost when system got shutdown )
   Console:
   AsyncAppender: Wrap other append and write log in seperate thread . (Lock free AsyncLogger is better)
   JDBC,Kafka

Enable AsyncLogger    :
   Add disruptor-3.0.0.jar(missing in our nexus repository) or higher was required.

   All Loggers Asynchronous :
       Set system property -Dlog4j2.contextSelector=org.apache.logging.log4j.core.async.AsyncLoggerContextSelector.
           All normal loggers like <Root> and <Logger> will be asynchronous .

   Mixture mode:
       No need to set system property "log4j2.contextSelector" to any value .
       Use the <asyncRoot> or <asyncLogger> configuration elements to specify the loggers that need to be asynchronous.


Externalize Log Configuration :
   -Dlog4j.configurationFile=file:///filePath