Skip to content

tangxiangcsu/jvm-insight

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
.idea
.idea
 
 
src
src
 
 
target/classes
target/classes
 
 
README.md
README.md
 
 
ai-jvm.png
ai-jvm.png
 
 
pom.xml
pom.xml
 
 

Repository files navigation

🌈 什么是AI-JVM-Profiler？

一款对应用程序无侵害的JVM内存解析系统， 本地路径

📚 妙用AI-JVM-Profiler

多种解析器，挑选一下吧

🚀 CpuAndMemory解析器

CpuAndMenory负责解析程序运行过程中CPU、堆内堆外内存和GC的各种情况，收集指标如下：

1. 基本指标

Title metricValue Note Options
appId null 程序的唯一标识
name * 应用程序的名称
processUuid a0b98ea4-0995-451e-ba89-8ee40c563eef 进程标识
tag mytag 标签
heapMemoryMax 2109734912 堆内存的最大值
nonHeapMemoryMax -1 非堆内存的最大值
nonHeapMemoryTotalUsed 14656800 当前非堆内存的已使用量
nonHeapMemoryCommitted 18350080 非堆内存的已提交量
heapMemoryCommitted 132120576 堆内存的已提交量
heapMemoryTotalUsed 15865344 堆内存的已使用量
systemCpuLoad 0.00305371 系统CPU负载率
processCpuLoad 0.00052097 进程CPU负载率
processCpuTime 437500000 纳秒
epochMillis 1722683736110 数据采集的时间 毫秒，自1979.01.01 00：00：00
2. GC解析
title metricValue Note Options
name G1 Young Generation 垃圾回收器的名称 "G1 Young Generation"、"G1 Old Generation"、"PS MarkSweep"
collectionCount 2 垃圾回收次数 累加
collectionTime 1000 总回收时间（Mills） 累加

3. 内存解析，支持解析堆内内存（Eden、Old、Survivor）和堆外内存

title metricValue Note Options
name G1 Eden Space 内存池名称 Eden、Survivor..
type Heap memory 堆类型 堆、非堆
usageCommitted 35651584 向操作系统申请资源 -
usageMax -1 最大容量 -
usageUsed -1 已使用容量 -
peakUsageCommitted 24117248 峰值向操作系统申请资源 -
peakUsageMax 35651584 峰值最大已使用 -
peakUsageUsed 24117248 峰值已使用 -

4. JVM缓冲池

title metricValue Note Options
totalCapacity 0 缓冲区总容量
name mapped 缓冲区类型 dirct、map
count 0 缓冲区数量
memoryUsed 0 已使用内存

🚀 Method可持续时间分析解析器

title metricValue Note Options
metricName duration.max max,min,sum,count
className UserserviceImpl 所监控的类名
methodName getUser 监控的方法
metricValue 2814 运行时间
epochMillis 1722749944630 检测时间

🚀 StackTrace栈追踪解析器

栈追踪解析器使用JMX收集JVM运行过程中的线程信息，通过封装的threadInfo来收集栈使用信息，不限但包括main、Attach Listener、Signal Dispatcher

title metricValue Note Options
threadName main
stacktrace list 栈追踪链路显示 list
threadState RUNNABLE 线程状态 就绪、执行、阻塞、停止
startEpoch 1722749894625 开始收集时间
endEpoch 1722749899630 收集结束时间

🚀 IO解析器

IO解析器主要监控Linux系统中的/proc/self/io文件，收集当前程序运行过程中的IO指标，如下

title metricValue Note Options
rchar 读字符数 -
wchar 写字符数 -
read_bytes 读字节数 -
write_bytes 写字节数 -

🚀 JVM线程解析器

JVM线程解析器，监控程序执行过程中总线程数、峰值线程数、实时/活动线程数和新线程数。

title metricValue Note Options
totalStartedThreadCount 总线程数 -
newThreadCount 新线程数 -
liveThreadCount 活跃线程数 -
peakThreadCount 峰值线程数 -

怎么运行

使用mvn clean package将项目进行打包，比如我们手写了一个 demo方法位于com.jvminsight.jvmprofiler.demo.RunApplication 请使用一下命令来监控该程序在运行期间的各种指标

java -javaagent:target/jvm-insight-1.0-SNAPSHOT.jar=reporter=com.jvminsight.jvmprofiler.reporter.ConsoleOutputReporter,tag=mytag,metricInterval=5000,durationProfiling=com.jvminsight.jvmprofiler.demo.RunApplication.publicSleepMethod,argumentProfiling=com.jvminsight.jvmprofiler.demo.RunApplication.publicSleepMethod.1,sampleInterval=100 -cp target/jvm-insight-1.0-SNAPSHOT.jar com.jvminsight.jvmprofiler.demo.RunApplication

收集得到指标如下

{
    "heapMemoryMax": 2109734912,
    "nonHeapMemoryTotalUsed": 8130568,
    "bufferPools": [
        {
            "totalCapacity": 0,
            "name": "mapped",
            "count": 0,
            "memoryUsed": 0
        },
        {
            "totalCapacity": 0,
            "name": "direct",
            "count": 0,
            "memoryUsed": 0
        },
        {
            "totalCapacity": 0,
            "name": "mapped - \u0027non-volatile memory\u0027",
            "count": 0,
            "memoryUsed": 0
        }
    ],
    "heapMemoryTotalUsed": 19922944,
    "epochMillis": 1723376934076,
    "nonHeapMemoryCommitted": 13238272,
    "heapMemoryCommitted": 132120576,
    "memoryPools": [
        {
            "peakUsageMax": 5898240,
            "usageMax": 5898240,
            "peakUsageUsed": 1202816,
            "name": "CodeHeap \u0027non-nmethods\u0027",
            "peakUsageCommitted": 2555904,
            "usageUsed": 1186560,
            "type": "Non-heap memory",
            "usageCommitted": 2555904
        },
        {
            "peakUsageMax": -1,
            "usageMax": -1,
            "peakUsageUsed": 4699696,
            "name": "Metaspace",
            "peakUsageCommitted": 4915200,
            "usageUsed": 4699696,
            "type": "Non-heap memory",
            "usageCommitted": 4915200
        },
        {
            "peakUsageMax": 122880000,
            "usageMax": 122880000,
            "peakUsageUsed": 1429248,
            "name": "CodeHeap \u0027profiled nmethods\u0027",
            "peakUsageCommitted": 2555904,
            "usageUsed": 1429248,
            "type": "Non-heap memory",
            "usageCommitted": 2555904
        },
        {
            "peakUsageMax": 1073741824,
            "usageMax": 1073741824,
            "peakUsageUsed": 523608,
            "name": "Compressed Class Space",
            "peakUsageCommitted": 655360,
            "usageUsed": 523608,
            "type": "Non-heap memory",
            "usageCommitted": 655360
        },
        {
            "peakUsageMax": -1,
            "usageMax": -1,
            "peakUsageUsed": 11534336,
            "name": "G1 Eden Space",
            "peakUsageCommitted": 26214400,
            "usageUsed": 11534336,
            "type": "Heap memory",
            "usageCommitted": 26214400
        },
        {
            "peakUsageMax": 2109734912,
            "usageMax": 2109734912,
            "peakUsageUsed": 8388608,
            "name": "G1 Old Gen",
            "peakUsageCommitted": 105906176,
            "usageUsed": 8388608,
            "type": "Heap memory",
            "usageCommitted": 105906176
        },
        {
            "peakUsageMax": -1,
            "usageMax": -1,
            "peakUsageUsed": 0,
            "name": "G1 Survivor Space",
            "peakUsageCommitted": 0,
            "usageUsed": 0,
            "type": "Heap memory",
            "usageCommitted": 0
        },
        {
            "peakUsageMax": 122880000,
            "usageMax": 122880000,
            "peakUsageUsed": 291456,
            "name": "CodeHeap \u0027non-profiled nmethods\u0027",
            "peakUsageCommitted": 2555904,
            "usageUsed": 291456,
            "type": "Non-heap memory",
            "usageCommitted": 2555904
        }
    ],
    "processCpuLoad": 0.08335573935607224,
    "systemCpuLoad": 0.1235764904981842,
    "processCpuTime": 328125000,
    "processUuid": "abb532fb-68ef-4af4-9ef2-fe811fa34c6b",
    "nonHeapMemoryMax": -1,
    "tag": "mytag",
    "gc": [
        {
            "collectionTime": 0,
            "name": "G1 Young Generation",
            "collectionCount": 0
        },
        {
            "collectionTime": 0,
            "name": "G1 Old Generation",
            "collectionCount": 0
        }
    ]
}

##🔬 智能JVM调优建议 对于新手，可能当程序不停的出现各种GC、CPU|内存飙升的现象而手足无措， 我们引入智谱大模型通过**微调技术实现对JVM调优的建议以及你当前程序可能存在的问题。

📕我应该怎么引入智能化？

  1. 修改application.yml的ai.apikey为你自己的
  2. 调用aiManager.doSyncRequest() 方法

我们发送如上指令到大模型中进行分析，得到优化意见如下

[
  {
    "problems": [
      {
        "description": "CPU 使用率较低，可能存在未充分利用的情况",
        "key": "（1）"
      },
      {
        "description": "堆内存使用率偏低，可能分配过大",
        "key": "（2）"
      },
      {
        "description": "年轻代和年老代 GC 活动数据为零，可能没有进行垃圾回收",
        "key": "（3）"
      }
    ],
    "suggestions": [
      {
        "description": "调整堆内存大小以匹配实际需求",
        "key": "（1）"
      },
      {
        "description": "检查应用程序是否存在长时间运行的同步操作，导致 CPU 未充分利用",
        "key": "（2）"
      },
      {
        "description": "监控并分析 GC 日志，以确定 GC 策略是否需要调整",
        "key": "（3）"
      }
    ]
  }
]

🤗 email

如果你有任何问题，请尽管联系我 xingtang@csu.edu.cn

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages