Skip to content

Latest commit

 

History

History
145 lines (105 loc) · 7.78 KB

README.EN.md

File metadata and controls

145 lines (105 loc) · 7.78 KB
Raw

简体中文 | English

MyPerf4J

A high performance, non-intrusive Java performance monitoring and statistical tool designed for high-concurrency, low-latency applications.

GitHub (pre-)release Build Status Coverage Status GitHub issues GitHub closed issues GitHub

Features

  • High performance: Very low performance consumption, only 73 nano seconds per record, can be used in production environment for a long time.
  • No intrusion: Using JavaAgent mode, no intrusion to the application, no need to modify the application code.
  • Low memory: With memory multiplexing, only a small number of temporary objects are generated throughout the life cycle, and the GC of the application is not affected.
  • High precision: Using nanoseconds to calculate response time.
  • Real-time: Supports second level monitoring, minimum 1 second!

Usage scenarios

  • Quickly locate performance bottlenecks for Java applications in a development environment
  • Long-term monitoring of performance metrics for Java applications in a production environment

Multilingual document

What does it monitor?

MyPerf4J collects dozens of metrics per application. All these metrics are collected and visualized in real-time.

This is a list of what it currently monitors:

  • Method
    RPS, Count, Avg, Min, Max, StdDev, TP50, TP90, TP95, TP99, TP999, TP9999, TP99999, TP100 Markdown Markdown

  • JVM Thread
    TotalStarted, Runnable, Blocked, Waiting, TimedWaiting, Terminated, Active, Peak, Daemon, New Markdown

  • JVM Memory
    HeapInit, HeapUsed, HeapCommitted, HeapMax, NonHeapInit, NonHeapUsed, NonHeapCommitted, NonHeapMax Markdown

  • JVM GC
    CollectCount, CollectTime Markdown

  • JVM Class
    Total, Loaded, Unloaded Markdown

Want to know how to achieve the above effect? Please start the application according to the description of Quick Start, and then follow the instructions in here to install and configure.

Quick start

MyPerf4J adopts JavaAgent configuration mode, transparent access application, and the application code is completely no-intrusive.

Build

  • git clone git@github.com:ThinkpadNC5/MyPerf4J.git
  • mvn clean package
  • Rename MyPerf4J-ASM-${MyPerf4J-version}.jar to MyPerf4J-ASM.jar

If you are using JDK 7 or higher, you can try to download MyPerf4J-ASM.jar directly.

Configure

Add the following two parameters to the JVM startup parameters

-javaagent:/your/path/to/MyPerf4J-ASM.jar -DMyPerf4JPropFile=/your/path/to/myPerf4J.properties

Among them, the configuration of MyPerf4JPropFile is as follows:

#Application name
AppName=MyPerf4JTest

#Configure MetricsProcessors type 0:Output to stdout.log in a standard formatted structure 1:Output to disk in standard formatted structure  2:Output to disk in InfluxDB LineProtocol format
MetricsProcessorType=1

#Config metrics log file, option
MethodMetricsFile=/data/logs/MyPerf4J/method_metrics.log
#ClassMetricsFile=/data/logs/MyPerf4J/class_metrics.log
#GCMetricsFile=/data/logs/MyPerf4J/gc_metrics.log
#MemMetricsFile=/data/logs/MyPerf4J/memory_metrics.log
#ThreadMetricsFile=/data/logs/MyPerf4J/thread_metrics.log

#Configure the log file scrolling interval, which has three values: MINUTELY, HOURLY, and DAILY.
LogRollingTimeUnit=HOURLY

#Configure the number of backup Recorders. The default is 1, the minimum is 1, and the maximum is 8. When you need to count a large number of method performance data in a smaller MillTimeSlice, you can configure a larger number.
BackupRecordersCount=1
    
#configure RecordMode, accurate/rough
RecorderMode=accurate
    
#configure TimeSlice, time unit: ms, min:1s, max:600s
MillTimeSlice=60000

#config show method params type
ShowMethodParams=true
    
#configure packages, separated with ';'
IncludePackages=cn.perf4j;org.myperf4j
    
#configure packages, separated with ';'
ExcludePackages=org.spring;
    
#configure methods, separated with ';'
ExcludeMethods=equals;hash
    
#true/false
ExcludePrivateMethod=true
    
#General method execution time threshold in ms
ProfilingTimeThreshold=1000
    
#The number of times the method execution time threshold is exceeded in a time slice, valid only when RecorderMode=accurate
ProfilingOutThresholdCount=10

Run

  • The output is to /data/logs/MyPerf4J/method_metrics.log:

    MyPerf4J Method Metrics [2018-09-06 19:21:40, 2018-09-06 19:21:45]
Method[4]                           RPS  Avg(ms)  Min(ms)  Max(ms)   StdDev     Count     TP50     TP90     TP95     TP99    TP999   TP9999  TP99999    TP100
DemoServiceImpl.getId1(long)       1974     0.00        0        0     0.00      9870        0        0        0        0        0        0        0        0
DemoServiceImpl.getId2(long)       2995     0.50        0        2     0.01     14975        0        1        2        2        2        2        2        2
DemoServiceImplV2.getId1(long)      787     0.00        0        0     0.00      3938        0        0        0        0        0        0        0        0
DemoServiceImplV2.getId3(long)     1575     0.50        0        1     0.01      7876        1        1        1        1        1        1        1        1

Uninstall

Remove the following two parameters from the JVM startup parameters and restart to uninstall the tool.

-javaagent:/your/path/to/MyPerf4J-ASM.jar -DMyPerf4JPropFile=/your/path/to/myPerf4J.properties

Issues

If you encounter any issues or if you have a question, don't hesitate to create an issue or send email : )

Inspired by

More Information

For more information about the project, please see https://github.com/ThinkpadNC5/MyPerf4J/wiki/English-Doc.