• 垃圾回收G1日志解析


    G1垃圾回收器介绍

    首先要搞清楚G1垃圾回收的过程

    主要分为下面4个步骤:

    初始标记(Initial Marking)           由于常规标记对象是否直接关联到GC roots,过程简单,停顿时间短

    并发标记(Concurrent Marking) 由于可达性性分析,存在较长停顿

    最终标记(Final Marking)           由于把并发标记时漏掉的那部分垃圾内存标记起来,比较少,停顿时间短

    筛选回收(Live Data Couting and Evacuation)  对标记后的垃圾内存进行整理,按指定计划进行回收。

    在下面的案例中,红色部分明确标明了这4个阶段。

     G1垃圾收集器参数说明

    以我的spark参数为例:

    spark.executor.extraJavaOptions -Dtag=mobius.optimize -XX:+UseG1GC -XX:+PrintFlagsFinal -XX:+PrintReferenceGC -XX:+PrintGCDetails

    -XX:+PrintGCDateStamps -XX:+PrintGCTimeStamps-XX:+PrintAdaptiveSizePolicy -XX:+PrintGC -Xloggc:/tmp/spark.gc.log

    -XX:+UnlockDiagnosticVMOptions -XX:+G1SummarizeConcMark -XX:InitiatingHeapOccupancyPercent=35 -XX:ConcGCThreads=15

    -XX:+PrintGCApplicationConcurrentTime -XX:-UseGCOverheadLimit

    下面是其拆分后的注释:

    spark.executor.extraJavaOptions     传入java参数给spark各个excutor

    -Dtag=mobius.optimize       标记名

    -XX:+UseG1GC          指定垃圾收集器

    -XX:+PrintFlagsFinal        java对象引用分为四种:强引用FinalReference,软引用SoftReference,弱引用WeakReference,虚引用PhantomReference。一般都是强引用,比如new一个对象并赋值

    -XX:+PrintReferenceGC      打印垃圾对象的引用数量

    -XX:+PrintGCDetails         打印详细GC日志

    -XX:+PrintGCDateStamps      打印可读的日期时间戳 比如2018-11-12

    -XX:+PrintGCTimeStamps     打印可读的时间戳 比如 09:10:11

    -XX:+PrintAdaptiveSizePolicy          启用自适应大小(为了避免fullGC,允许改变eden空间和survivor空间大小)

    -XX:+PrintGC           打印GC日志

    -Xloggc:/tmp/spark.gc.log        指定GC日志路径

    -XX:+UnlockDiagnosticVMOptions

    -XX:+G1SummarizeConcMark          

    -XX:InitiatingHeapOccupancyPercent=35

    -XX:ParallelGCThreads=33        Parallel和G1收集器的标记阶段的线程数l量,默认33,对应GC日志 [ GC Workers: 33 ](并行:多个垃圾收集线程并行工作,此时用户线程处于等待状态,stop the world)

    -XX:ConcGCThreads=15            CMS和G1回收器的前三个标记阶段都是多线程的,指定GC并发线程数量,默认0(并发:用户线程与垃圾收集线程同时执行,不一定是并行,可能会交替执行)

    -XX:+PrintGCApplicationConcurrentTime          打印应用程序执行时间

    -XX:-UseGCOverheadLimit                取消垃圾回收时的内存限制(减号)

    -XX:MaxGCPauseMillis=200          垃圾回收停顿最长时间,默认是200ms

    日志分析案例

    下面一段日志是我在spark计算任务中产生的垃圾回收日志:

    2018-11-13T05:32:44.473+0800: 1.857: Application time: 0.2130324 seconds
    2018-11-13T05:32:44.571+0800: 1.954: Application time: 0.0970449 seconds
    2018-11-13T05:32:44.696+0800: 2.079: Application time: 0.1249397 seconds
    2018-11-13T05:32:44.874+0800: 2.258: Application time: 0.1776871 seconds
     2.258: [G1Ergonomics (Concurrent Cycles) request concurrent cycle initiation, reason: requested by GC cause, GC cause: Metadata GC Threshold]
     2.258: [G1Ergonomics (Concurrent Cycles) initiate concurrent cycle, reason: concurrent cycle initiation requested]
    2018-11-13T05:32:44.875+0800: 2.259: [GC pause (Metadata GC Threshold) (young) (initial-mark) 2.260: 
    [G1Ergonomics (CSet Construction) start choosing CSet, _pending_cards: 0, predicted base time: 10.00 ms, 
    remaining time: 190.00 ms, target pause time: 200.00 ms]
     2.260: [G1Ergonomics (CSet Construction) add young regions to CSet, eden: 60 regions, survivors: 0 regions, predicted young region time: 3605.71 ms]
     2.260: [G1Ergonomics (CSet Construction) finish choosing CSet, eden: 60 regions, survivors: 0 regions, old: 0 regions, 
    predicted pause time: 3615.71 ms, target pause time: 200.00 ms]
    2018-11-13T05:32:44.948+0800: 2.331: [SoftReference, 0 refs, 0.0002103 secs]2018-11-13T05:32:44.948+0800: 2.332: 
    [WeakReference, 970 refs, 0.0005817 secs]2018-11-13T05:32:44.949+0800: 2.332: [FinalReference, 4265 refs, 0.0266394 secs]
    2018-11-13T05:32:44.975+0800: 2.359: [PhantomReference, 0 refs, 3 refs, 0.0000585 secs]2018-11-13T05:32:44.976+0800: 2.359: 
    [JNI Weak Reference, 0.0001374 secs], 0.1040572 secs]
       [Parallel Time: 29.1 ms, GC Workers: 33]
          [GC Worker Start (ms): Min: 2260.1, Avg: 2265.0, Max: 2265.9, Diff: 5.8]
          [Ext Root Scanning (ms): Min: 0.2, Avg: 18.2, Max: 22.7, Diff: 22.5, Sum: 691.2]
          [Update RS (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.0]
             [Processed Buffers: Min: 0, Avg: 0.0, Max: 0, Diff: 0, Sum: 0]
          [Scan RS (ms): Min: 0.0, Avg: 0.0, Max: 0.1, Diff: 0.1, Sum: 0.3]
          [Code Root Scanning (ms): Min: 0.0, Avg: 0.3, Max: 8.6, Diff: 8.6, Sum: 10.0]
          [Object Copy (ms): Min: 0.0, Avg: 3.9, Max: 22.9, Diff: 22.9, Sum: 147.1]
          [Termination (ms): Min: 0.0, Avg: 1.3, Max: 2.3, Diff: 2.3, Sum: 50.3]
             [Termination Attempts: Min: 1, Avg: 3.0, Max: 17, Diff: 16, Sum: 115]
          [GC Worker Other (ms): Min: 0.0, Avg: 0.0, Max: 0.1, Diff: 0.1, Sum: 1.6]
          [GC Worker Total (ms): Min: 22.7, Avg: 23.7, Max: 28.6, Diff: 5.9, Sum: 900.6]
          [GC Worker End (ms): Min: 2288.6, Avg: 2288.7, Max: 2288.8, Diff: 0.2]
       [Code Root Fixup: 1.0 ms]
       [Code Root Purge: 0.1 ms]
       [Clear CT: 1.0 ms]
       [Other: 72.8 ms]
          [Choose CSet: 0.0 ms]
          [Ref Proc: 68.9 ms]
          [Ref Enq: 0.2 ms]
          [Redirty Cards: 0.4 ms]
          [Humongous Register: 0.1 ms]
          [Humongous Reclaim: 0.1 ms]
          [Free CSet: 0.5 ms]
       [Eden: 240.0M(512.0M)->0.0B(496.0M) Survivors: 0.0B->16.0M Heap: 254.0M(10.0G)->31.7M(10.0G)]
     [Times: user=0.44 sys=0.09, real=0.11 secs]
    2018-11-13T05:32:44.980+0800: 2.363: [GC concurrent-root-region-scan-start]
    2018-11-13T05:32:44.996+0800: 2.380: [GC concurrent-root-region-scan-end, 0.0167701 secs]
    2018-11-13T05:32:44.996+0800: 2.380: [GC concurrent-mark-start]
    2018-11-13T05:32:45.013+0800: 2.396: [GC concurrent-mark-end, 0.0160995 secs]
    2018-11-13T05:32:45.013+0800: 2.396: Application time: 0.0328728 seconds
    2018-11-13T05:32:45.013+0800: 2.397: [GC remark 2018-11-13T05:32:45.013+0800: 2.397: 
    [Finalize Marking, 0.0331140 secs] 2018-11-13T05:32:45.047+0800: 2.430: 
    [GC ref-proc2018-11-13T05:32:45.047+0800: 2.430: 
    [SoftReference, 0 refs, 0.0001959 secs]2018-11-13T05:32:45.047+0800: 2.430: 
    [WeakReference, 0 refs, 0.0001148 secs]2018-11-13T05:32:45.047+0800: 2.430: 
    [FinalReference, 0 refs, 0.0001158 secs]2018-11-13T05:32:45.047+0800: 2.430: 
    [PhantomReference, 0 refs, 0 refs, 0.0002233 secs]2018-11-13T05:32:45.047+0800: 2.431: 
    [JNI Weak Reference, 0.0000572 secs], 0.0008127 secs] 2018-11-13T05:32:45.047+0800: 2.431: [Unloading, 0.0078450 secs], 0.0442183 secs]
     [Times: user=0.36 sys=0.01, real=0.05 secs]
    2018-11-13T05:32:45.058+0800: 2.441: Application time: 0.0001231 seconds
    2018-11-13T05:32:45.058+0800: 2.442: [GC cleanup 35M->35M(10G), 0.0134109 secs]
     [Times: user=0.04 sys=0.00, real=0.01 secs]
    2018-11-13T05:32:45.144+0800: 2.528: Application time: 0.0721037 seconds
    2018-11-13T05:32:45.392+0800: 2.775: Application time: 0.2464978 seconds
    2018-11-13T05:32:45.805+0800: 3.188: Application time: 0.4129881 seconds
    2018-11-13T05:32:46.806+0800: 4.190: Application time: 1.0001058 seconds
    下面是每条日志含义解析:
    Application time: 应用程序时间
    。。。

  • 相关阅读:
    IDEA2019破解版安装
    Docker (一)安装与基本命令
    js导出PDF
    python入门笔记一安装
    微信内关闭当前页面
    微信一些网页
    内网穿透工具:钉钉HTTP内网穿透使用详解
    html生成二维码,qr
    IDEA无法正常启动(打不开&报错)
    微信测试号
  • 原文地址:https://www.cnblogs.com/geektcp/p/9951539.html
Copyright © 2020-2023  润新知