深入JVM内核3 常用JVM配置参数

作者: 香沙小熊 | 来源:发表于2018-12-20 16:47 被阅读40次

    深入JVM内核 目录

    Trace跟踪参数

    • -verbose:gc
    • -XX:+printGC
      可以打印GC的简要信息
    [GC 4790K->374K(15872K), 0.0001606 secs]
    [GC 4790K->374K(15872K), 0.0001474 secs]
    [GC 4790K->374K(15872K), 0.0001563 secs]
    [GC 4790K->374K(15872K), 0.0001682 secs]
    
    • -XX:+PrintGCDetails
      打印GC详细信息
    • -XX:+PrintGCTimeStamps
      打印CG发生的时间戳

    [GC[DefNew: 4416K->0K(4928K), 0.0001897 secs] 4790K->374K(15872K), 0.0002232 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
    
    • -XX:+PrintGCDetails的输出
    Heap
    def new generation   total 13824K, used 11223K [0x27e80000, 0x28d80000, 0x28d80000)
    eden space 12288K,  91% used [0x27e80000, 0x28975f20, 0x28a80000)
    from space 1536K,   0% used [0x28a80000, 0x28a80000, 0x28c00000)
    to   space 1536K,   0% used [0x28c00000, 0x28c00000, 0x28d80000)
    tenured generation   total 5120K, used 0K [0x28d80000, 0x29280000, 0x34680000)
    the space 5120K,   0% used [0x28d80000, 0x28d80000, 0x28d80200, 0x29280000)
    compacting perm gen  total 12288K, used 142K [0x34680000, 0x35280000, 0x38680000)
    the space 12288K,   1% used [0x34680000, 0x346a3a90, 0x346a3c00, 0x35280000)
    ro space 10240K,  44% used [0x38680000, 0x38af73f0, 0x38af7400, 0x39080000)
    rw space 12288K,  52% used [0x39080000, 0x396cdd28, 0x396cde00, 0x39c80000)
    
    • Xloggc:log/gc.log
      指定GC log的位置,以文件输出
      帮助开发人员分析问题
    • -XX:+PrintHeapAtGC
      每次一次GC后,都打印堆信息
    {Heap before GC invocations=0 (full 0):
     def new generation   total 3072K, used 2752K [0x33c80000, 0x33fd0000, 0x33fd0000)
    eden space 2752K, 100% used [0x33c80000, 0x33f30000, 0x33f30000)
    from space 320K,   0% used [0x33f30000, 0x33f30000, 0x33f80000)
    to   space 320K,   0% used [0x33f80000, 0x33f80000, 0x33fd0000)
    tenured generation   total 6848K, used 0K [0x33fd0000, 0x34680000, 0x34680000)
    the space 6848K,   0% used [0x33fd0000, 0x33fd0000, 0x33fd0200, 0x34680000)
    compacting perm gen  total 12288K, used 143K [0x34680000, 0x35280000, 0x38680000)
    the space 12288K,   1% used [0x34680000, 0x346a3c58, 0x346a3e00, 0x35280000)
    ro space 10240K,  44% used [0x38680000, 0x38af73f0, 0x38af7400, 0x39080000)
    rw space 12288K,  52% used [0x39080000, 0x396cdd28, 0x396cde00, 0x39c80000)
    
    [GC[DefNew: 2752K->320K(3072K), 0.0014296 secs] 2752K->377K(9920K), 0.0014604 secs] [Times: user=0.00 sys=0.00, real=0.00 secs]
    
    Heap after GC invocations=1 (full 0):
    def new generation   total 3072K, used 320K [0x33c80000, 0x33fd0000, 0x33fd0000)
    eden space 2752K,   0% used [0x33c80000, 0x33c80000, 0x33f30000)
    from space 320K, 100% used [0x33f80000, 0x33fd0000, 0x33fd0000)
    to   space 320K,   0% used [0x33f30000, 0x33f30000, 0x33f80000)
    tenured generation   total 6848K, used 57K [0x33fd0000, 0x34680000, 0x34680000)
    the space 6848K,   0% used [0x33fd0000, 0x33fde458, 0x33fde600, 0x34680000)
    compacting perm gen  total 12288K, used 143K [0x34680000, 0x35280000, 0x38680000)
    the space 12288K,   1% used [0x34680000, 0x346a3c58, 0x346a3e00, 0x35280000)
    ro space 10240K,  44% used [0x38680000, 0x38af73f0, 0x38af7400, 0x39080000)
    rw space 12288K,  52% used [0x39080000, 0x396cdd28, 0x396cde00, 0x39c80000)
    }
    
    • -XX:+TraceClassLoading
      监控类的加载
      [Loaded java.lang.Object from shared objects file]
      [Loaded java.io.Serializable from shared objects file]
      [Loaded java.lang.Comparable from shared objects file]
      [Loaded java.lang.CharSequence from shared objects file]
      [Loaded java.lang.String from shared objects file]
      [Loaded java.lang.reflect.GenericDeclaration from shared objects file]
      [Loaded java.lang.reflect.Type from shared objects file]
      跟踪调试的时候 那些类被加载出来

    • -XX:+PrintClassHistogram
      按下Ctrl+Break后,打印类的信息:

    num     #instances         #bytes  class name
    ----------------------------------------------
       1:        890617      470266000  [B
       2:        890643       21375432  java.util.HashMap$Node
       3:        890608       14249728  java.lang.Long
       4:            13        8389712  [Ljava.util.HashMap$Node;
       5:          2062         371680  [C
       6:           463          41904  java.lang.Class
    

    分别显示:序号、实例数量、总大小、类型

    堆的分配参数

    • Xmx –Xms
      指定最大堆和最小堆

    VM options: -Xmx20m -Xms5m 运行代码:

        public static void main(String[] args)  {
            System.out.print("Xmx=");
            System.out.println(Runtime.getRuntime().maxMemory()/1024.0/1024+"M");
    
            System.out.print("free mem=");
            System.out.println(Runtime.getRuntime().freeMemory()/1024.0/1024+"M");
    
            System.out.print("total mem=");
            System.out.println(Runtime.getRuntime().totalMemory()/1024.0/1024+"M");
        }
    

    结果:

    Xmx=18.0M
    free mem=4.660301208496094M
    total mem=5.5M
    
        public static void main(String[] args)  {
            byte[] b= new byte[4 * 1024 * 1024];
            System.out.print("Xmx=");
            System.out.println(Runtime.getRuntime().maxMemory()/1024.0/1024+"M");
    
            System.out.print("free mem=");
            System.out.println(Runtime.getRuntime().freeMemory()/1024.0/1024+"M");
    
            System.out.print("total mem=");
            System.out.println(Runtime.getRuntime().totalMemory()/1024.0/1024+"M");
        }
    

    结果:

    Xmx=18.0M
    free mem=5.074928283691406M
    total mem=10.0M
    

    总内存变多

        public static void main(String[] args)  {
            System.gc();
            System.out.print("Xmx=");
            System.out.println(Runtime.getRuntime().maxMemory()/1024.0/1024+"M");
    
            System.out.print("free mem=");
            System.out.println(Runtime.getRuntime().freeMemory()/1024.0/1024+"M");
    
            System.out.print("total mem=");
            System.out.println(Runtime.getRuntime().totalMemory()/1024.0/1024+"M");
        }
    

    结果:

    Xmx=18.0M
    free mem=5.9815826416015625M
    total mem=6.5M
    

    空闲内存变多

    • -Xmn
      设置新生代大小
    • -XX:NewRatio
      新生代(eden+2*s)和老年代(不包含永久区)的比值
      4 表示 新生代:老年代=1:4,即年轻代占堆的1/5
    • -XX:SurvivorRatio
      设置两个Survivor区和eden的比
      8表示 两个Survivor :eden=2:8,即一个Survivor占年轻代的1/10
        public static void main(String[] args) {
            byte[] b=null;
            for(int i=0;i<10;i++)
                b=new byte[1*1024*1024];
        }
    

    VM options: -Xmx20m -Xms20m -Xmn3m -XX:+PrintGCDetails
    输出:

    Heap
     PSYoungGen      total 2560K, used 1490K [0x00000007bfd00000, 0x00000007c0000000, 0x00000007c0000000)
      eden space 2048K, 72% used [0x00000007bfd00000,0x00000007bfe748a8,0x00000007bff00000)
      from space 512K, 0% used [0x00000007bff80000,0x00000007bff80000,0x00000007c0000000)
      to   space 512K, 0% used [0x00000007bff00000,0x00000007bff00000,0x00000007bff80000)
     ParOldGen       total 17408K, used 10240K [0x00000007bec00000, 0x00000007bfd00000, 0x00000007bfd00000)
      object space 17408K, 58% used [0x00000007bec00000,0x00000007bf6000a0,0x00000007bfd00000)
     Metaspace       used 3054K, capacity 4496K, committed 4864K, reserved 1056768K
      class space    used 334K, capacity 388K, committed 512K, reserved 1048576K
    
    1. 没有触发GC
    2. 全部分配在老年代

    VM options: -Xmx20m -Xms20m -Xmn15m -XX:+PrintGCDetails
    输出:

    Heap
     PSYoungGen      total 13824K, used 11966K [0x00000007bf100000, 0x00000007c0000000, 0x00000007c0000000)
      eden space 12288K, 97% used [0x00000007bf100000,0x00000007bfcaf8d0,0x00000007bfd00000)
      from space 1536K, 0% used [0x00000007bfe80000,0x00000007bfe80000,0x00000007c0000000)
      to   space 1536K, 0% used [0x00000007bfd00000,0x00000007bfd00000,0x00000007bfe80000)
     ParOldGen       total 5120K, used 0K [0x00000007bec00000, 0x00000007bf100000, 0x00000007bf100000)
      object space 5120K, 0% used [0x00000007bec00000,0x00000007bec00000,0x00000007bf100000)
     Metaspace       used 3051K, capacity 4496K, committed 4864K, reserved 1056768K
      class space    used 334K, capacity 388K, committed 512K, reserved 1048576K
    
    1. 没有触发GC
    2. 全部分配在eden
    3. 老年代没有使用

    VM options: -Xmx20m -Xms20m -Xmn15m -XX:+PrintGCDetails
    输出

    [GC (Allocation Failure) [PSYoungGen: 5797K->496K(6656K)] 6861K->2604K(19968K), 0.0011469 secs] [Times: user=0.01 sys=0.00, real=0.00 secs] 
    Heap
     PSYoungGen      total 6656K, used 1622K [0x00000007bf900000, 0x00000007c0000000, 0x00000007c0000000)
      eden space 6144K, 18% used [0x00000007bf900000,0x00000007bfa19ab8,0x00000007bff00000)
      from space 512K, 96% used [0x00000007bff80000,0x00000007bfffc010,0x00000007c0000000)
      to   space 512K, 0% used [0x00000007bff00000,0x00000007bff00000,0x00000007bff80000)
     ParOldGen       total 13312K, used 2108K [0x00000007bec00000, 0x00000007bf900000, 0x00000007bf900000)
      object space 13312K, 15% used [0x00000007bec00000,0x00000007bee0f030,0x00000007bf900000)
     Metaspace       used 2947K, capacity 4496K, committed 4864K, reserved 1056768K
      class space    used 325K, capacity 388K, committed 512K, reserved 1048576K
    
    1. 进行了1次新生代GC
    2. s0 s1 太小需要老年代担保

    VM options: -Xmx20m -Xms20m -Xmn6m -XX:SurvivorRatio=2 -XX:+PrintGCDetails

    [GC (Allocation Failure) [PSYoungGen: 2357K->1520K(4608K)] 2357K->1560K(18944K), 0.0010973 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
    [GC (Allocation Failure) [PSYoungGen: 3658K->1504K(4608K)] 3698K->1552K(18944K), 0.0009730 secs] [Times: user=0.01 sys=0.00, real=0.00 secs] 
    [GC (Allocation Failure) [PSYoungGen: 3621K->1504K(4608K)] 3669K->1576K(18944K), 0.0004732 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
    [GC (Allocation Failure) [PSYoungGen: 3604K->1520K(4608K)] 3676K->1608K(18944K), 0.0004172 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
    [GC (Allocation Failure) [PSYoungGen: 3623K->1520K(4608K)] 3711K->1616K(18944K), 0.0004710 secs] [Times: user=0.01 sys=0.00, real=0.01 secs] 
    Heap
     PSYoungGen      total 4608K, used 2619K [0x00000007bfa00000, 0x00000007c0000000, 0x00000007c0000000)
      eden space 3072K, 35% used [0x00000007bfa00000,0x00000007bfb12e48,0x00000007bfd00000)
      from space 1536K, 98% used [0x00000007bfd00000,0x00000007bfe7c020,0x00000007bfe80000)
      to   space 1536K, 0% used [0x00000007bfe80000,0x00000007bfe80000,0x00000007c0000000)
     ParOldGen       total 14336K, used 96K [0x00000007bec00000, 0x00000007bfa00000, 0x00000007bfa00000)
      object space 14336K, 0% used [0x00000007bec00000,0x00000007bec18000,0x00000007bfa00000)
     Metaspace       used 2929K, capacity 4496K, committed 4864K, reserved 1056768K
      class space    used 319K, capacity 388K, committed 512K, reserved 1048576K
    
    1. 进行了5次新生代GC
    2. s0 s1 增大
    • -XX:+HeapDumpOnOutOfMemoryError
      OOM时导出堆到文件
    • -XX:+HeapDumpPath
      导出OOM的路径
    • -XX:OnOutOfMemoryError
      在OOM时,执行一个脚本
      "-XX:OnOutOfMemoryError=D:/tools/jdk1.7_40/bin/printstack.bat %p“
      当程序OOM时,在D:/a.txt中将会生成线程的dump
      可以在OOM时,发送邮件,甚至是重启程序

    总结

    1. 根据实际事情调整新生代和幸存代的大小
    2. 官方推荐新生代占堆的3/8
    3. 幸存代占新生代的1/10
    4. 在OOM时,记得Dump出堆,确保可以排查线程

    永久区分配参数

    • -XX:PermSize -XX:MaxPermSize
      设置永久区的初始空间和最大空间
      他们表示,一个系统可以容纳多少个类型

    使用CGLIB等库的时候,可能会产生大量的类,这些类,有可能撑爆永久区导致OOM

    for(int i=0;i<100000;i++){
        CglibBean bean = new CglibBean("geym.jvm.ch3.perm.bean"+i,new HashMap());
    }
    

    不断产生新的类

    打开堆的Dump

    堆空间实际占用非常少
    但是永久区溢出 一样抛出OOM


    如果堆空间没有用完也抛出了OOM,有可能是永久区导致的

    栈大小分配

    • 通常只有几百K
    • 决定了函数调用的深度
    • 每个线程都有独立的栈空间
    • 局部变量、参数 分配在栈上
        private static int count=0;
        public static void recursion(long a,long b,long c){
            long e=1,f=2,g=3,h=4,i=5,k=6,q=7,x=8,y=9,z=10;
            count++;
            recursion(a,b,c);
        }
        public static void main(String args[]){
            try{
                recursion(0L,0L,0L);
            }catch(Throwable e){
                System.out.println("deep of calling = "+count);
                e.printStackTrace();
            }
        }
    

    VM options: -Xss160K

    deep of calling = 235
    java.lang.StackOverflowError
    ...
    

    VM options: -Xss280K

    deep of calling = 662
    java.lang.StackOverflowError
    
    特别感谢

    深入JVM内核—原理、诊断与优化

    相关文章

      网友评论

        本文标题:深入JVM内核3 常用JVM配置参数

        本文链接:https://www.haomeiwen.com/subject/ovashqtx.html