【JVM】JDK10的G1日志浅谈

• Read

最近学习《深入理解Java虚拟机》，书中使用的是JDK1.7，垃圾收集器使用CMS；而笔者环境为JDK10，垃圾收集默认使用G1。G1与CMS之间的差异挺大，日志格式也截然不同，而目前主流的是G1收集器，故重点学习G1。笔者最近入门学习JVM，以下理解有错的烦请指出，不甚感激～

测试一

1. 测试代码

1234567891011121314151617181920212223

package main.GCtest;public class { private static final int _1MB = 1024 * 1024; * VM参数： -Xms20M -Xmx20M -Xmn10M -Xlog:gc* * -Xmn10M 指分配10M给新生代，Java堆剩余的分配给老年代 * -Xlog:gc* 发生垃圾收集时打印内存回收日志，并在进程退出时输出当前的内存各区域分配情况 */ public static void () { byte[] allocation1, allocation2, allocation3, allocation4,allocation5,allocation6; allocation1 = new byte[2 * _1MB]; allocation2 = new byte[2 * _1MB]; allocation3 = new byte[2 * _1MB]; allocation4 = new byte[4 * _1MB]; } public static void main(String[] args){ testAllocation(); }}

2. 堆内存监控，每行代码断点几秒种，以便出现分层

3. GC日志

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162

#堆每块1M大小[0.013s][info][gc,heap] Heap region size: 1M#使用G1垃圾收集器[0.014s][info][gc ] Using G1#堆地址、大小[0.014s][info][gc,heap,coops] Heap address: 0x00000007bec00000, size: 20 MB, Compressed Oops mode: Zero based, Oop shift amount: 3#初始标记开始，引发GC的原因是Humongous Allocation[0.206s][info][gc,start ] GC(0) Pause Initial Mark (G1 Humongous Allocation)[0.206s][info][gc,task ] GC(0) Using 4 workers of 4 for evacuation[0.208s][info][gc,phases ] GC(0) Pre Evacuate Collection Set: 0.0ms[0.208s][info][gc,phases ] GC(0) Evacuate Collection Set: 2.2ms[0.208s][info][gc,phases ] GC(0) Post Evacuate Collection Set: 0.1ms[0.208s][info][gc,phases ] GC(0) Other: 0.1ms#Eden区回收3个region[0.208s][info][gc,heap ] GC(0) Eden regions: 3->0(3)#Survivor区增加2个region[0.208s][info][gc,heap ] GC(0) Survivor regions: 0->2(2)[0.208s][info][gc,heap ] GC(0) Old regions: 0->0#Humongous区仍然是9个region[0.208s][info][gc,heap ] GC(0) Humongous regions: 9->9#元空间不变[0.208s][info][gc,metaspace ] GC(0) Metaspace: 5242K->5242K(1056768K)#初始标记完成，11M减为10M[0.208s][info][gc ] GC(0) Pause Initial Mark (G1 Humongous Allocation) 11M->10M(20M) 2.464ms#初始标记耗时[0.208s][info][gc,cpu ] GC(0) User=0.00s Sys=0.01s Real=0.00s#并发周期[0.208s][info][gc ] GC(1) Concurrent Cycle#并发清除已标记的？[0.208s][info][gc,marking ] GC(1) Concurrent Clear Claimed Marks[0.208s][info][gc,marking ] GC(1) Concurrent Clear Claimed Marks 0.006ms[0.208s][info][gc,marking ] GC(1) Concurrent Scan Root Regions[0.209s][info][gc,marking ] GC(1) Concurrent Scan Root Regions 0.764ms[0.209s][info][gc,marking ] GC(1) Concurrent Mark (0.209s)#并发标记[0.209s][info][gc,marking ] GC(1) Concurrent Mark From Roots#使用一个进程标记[0.209s][info][gc,task ] GC(1) Using 1 workers of 1 for marking[0.209s][info][gc,marking ] GC(1) Concurrent Mark From Roots 0.055ms[0.209s][info][gc,marking ] GC(1) Concurrent Mark (0.209s, 0.209s) 0.073ms#重新标记[0.211s][info][gc,start ] GC(1) Pause Remark#还看不懂string and symbol table[0.211s][info][gc,stringtable] GC(1) Cleaned string and symbol table, strings: 3207 processed, 0 removed, symbols: 23766 processed, 0 removed[0.211s][info][gc ] GC(1) Pause Remark 15M->15M(20M) 0.727ms[0.211s][info][gc,cpu ] GC(1) User=0.00s Sys=0.00s Real=0.00s[0.211s][info][gc,marking ] GC(1) Concurrent Create Live Data[0.211s][info][gc,marking ] GC(1) Concurrent Create Live Data 0.034ms#清除垃圾[0.211s][info][gc,start ] GC(1) Pause Cleanup[0.212s][info][gc ] GC(1) Pause Cleanup 15M->15M(20M) 0.087ms[0.212s][info][gc,cpu ] GC(1) User=0.00s Sys=0.00s Real=0.00s[0.212s][info][gc,marking ] GC(1) Concurrent Cleanup for Next Mark[0.212s][info][gc,marking ] GC(1) Concurrent Cleanup for Next Mark 0.252ms[0.212s][info][gc ] GC(1) Concurrent Cycle 3.593ms[0.212s][info][gc,heap,exit ] Heap#目前堆内存使用情况[0.212s][info][gc,heap,exit ] garbage-first heap total 20480K, used 15547K [0x00000007bec00000, 0x00000007c0000000)#各个区使用的内存情况[0.212s][info][gc,heap,exit ] region size 1024K, 3 young (3072K), 2 survivors (2048K)[0.212s][info][gc,heap,exit ] Metaspace used 5279K, capacity 5352K, committed 5632K, reserved 1056768K[0.212s][info][gc,heap,exit ] class space used 445K, capacity 460K, committed 512K, reserved 1048576K

测试二

1. 测试代码

123456789101112131415161718192021222324252627282930313233343536373839404142

package main.GCtest;public class { 大专栏  【JVM】JDK10的G1日志浅谈 private static final int _1MB = 1024 * 1024; * VM参数：-verbose:gc -Xms20M -Xmx20M -Xmn10M -Xlog:gc* -XX:SurvivorRatio=8 * verbose:gc 显示GC的操作内容 * -Xmn10M 指分配10M给新生代，Java堆剩余的分配给老年代 * -Xlog:gc* 发生垃圾收集时打印内存回收日志，并在进程退出时输出当前的内存各区域分配情况 * -XX:SurvivorRatio=8 定义新生代中Eden区与Survivor区的空间比例是8：1 */ public static void () { byte[] allocation1, allocation2, allocation3, allocation4,allocation5,allocation6; allocation1 = new byte[_1MB / 3]; allocation2 = new byte[_1MB / 3]; allocation3 = new byte[_1MB / 3]; allocation1 = new byte[_1MB / 3]; allocation2 = new byte[_1MB / 3]; allocation3 = new byte[_1MB / 3]; allocation1 = new byte[_1MB / 3]; allocation2 = new byte[_1MB / 3]; allocation3 = new byte[_1MB / 3]; allocation1 = new byte[_1MB / 3]; allocation2 = new byte[_1MB / 3]; allocation3 = new byte[_1MB / 3]; allocation1 = new byte[_1MB / 3]; allocation2 = new byte[_1MB / 3]; allocation3 = new byte[_1MB / 3]; allocation1 = new byte[_1MB / 3]; allocation2 = new byte[_1MB / 3]; allocation3 = new byte[_1MB / 3]; allocation4 = new byte[4 * _1MB]; allocation5 = new byte[4 * _1MB];// 出现一次Minor GC } public static void main(String[] args){ testAllocation(); }}

2. 堆内存监控

3. GC日志

   12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061

   [0.016s][info][gc,heap] Heap region size: 1M[0.017s][info][gc ] Using G1[0.017s][info][gc,heap,coops] Heap address: 0x00000007bec00000, size: 20 MB, Compressed Oops mode: Zero based, Oop shift amount: 3#年轻代回收[0.242s][info][gc,start ] GC(0) Pause Young (G1 Evacuation Pause)[0.242s][info][gc,task ] GC(0) Using 4 workers of 4 for evacuation[0.245s][info][gc,phases ] GC(0) Pre Evacuate Collection Set: 0.0ms[0.245s][info][gc,phases ] GC(0) Evacuate Collection Set: 2.7ms[0.245s][info][gc,phases ] GC(0) Post Evacuate Collection Set: 0.1ms[0.245s][info][gc,phases ] GC(0) Other: 0.1ms#Eden回收9个region[0.245s][info][gc,heap ] GC(0) Eden regions: 9->0(8)[0.245s][info][gc,heap ] GC(0) Survivor regions: 0->2(2)[0.245s][info][gc,heap ] GC(0) Old regions: 0->1[0.245s][info][gc,heap ] GC(0) Humongous regions: 0->0[0.245s][info][gc,metaspace ] GC(0) Metaspace: 5296K->5296K(1056768K)#年轻代回收内存[0.245s][info][gc ] GC(0) Pause Young (G1 Evacuation Pause) 9M->2M(20M) 3.005ms[0.245s][info][gc,cpu ] GC(0) User=0.00s Sys=0.00s Real=0.01s#初始标记[0.248s][info][gc,start ] GC(1) Pause Initial Mark (G1 Humongous Allocation)[0.248s][info][gc,task ] GC(1) Using 4 workers of 4 for evacuation[0.249s][info][gc,phases ] GC(1) Pre Evacuate Collection Set: 0.0ms[0.249s][info][gc,phases ] GC(1) Evacuate Collection Set: 1.3ms[0.249s][info][gc,phases ] GC(1) Post Evacuate Collection Set: 0.1ms[0.249s][info][gc,phases ] GC(1) Other: 0.1ms[0.249s][info][gc,heap ] GC(1) Eden regions: 3->0(5)[0.249s][info][gc,heap ] GC(1) Survivor regions: 2->2(2)[0.249s][info][gc,heap ] GC(1) Old regions: 1->2[0.249s][info][gc,heap ] GC(1) Humongous regions: 5->5[0.249s][info][gc,metaspace ] GC(1) Metaspace: 5341K->5341K(1056768K)#初始标记完成[0.249s][info][gc ] GC(1) Pause Initial Mark (G1 Humongous Allocation) 9M->7M(20M) 1.488ms[0.249s][info][gc,cpu ] GC(1) User=0.01s Sys=0.00s Real=0.00s[0.249s][info][gc ] GC(2) Concurrent Cycle[0.249s][info][gc,marking ] GC(2) Concurrent Clear Claimed Marks[0.249s][info][gc,marking ] GC(2) Concurrent Clear Claimed Marks 0.006ms[0.249s][info][gc,marking ] GC(2) Concurrent Scan Root Regions[0.249s][info][gc,marking ] GC(2) Concurrent Scan Root Regions 0.046ms[0.249s][info][gc,marking ] GC(2) Concurrent Mark (0.249s)[0.249s][info][gc,marking ] GC(2) Concurrent Mark From Roots[0.249s][info][gc,task ] GC(2) Using 1 workers of 1 for marking[0.251s][info][gc,marking ] GC(2) Concurrent Mark From Roots 1.738ms[0.251s][info][gc,marking ] GC(2) Concurrent Mark (0.249s, 0.251s) 1.772ms[0.252s][info][gc,start ] GC(2) Pause Remark[0.254s][info][gc,stringtable] GC(2) Cleaned string and symbol table, strings: 3241 processed, 0 removed, symbols: 23981 processed, 0 removed[0.254s][info][gc ] GC(2) Pause Remark 12M->12M(20M) 2.252ms[0.254s][info][gc,cpu ] GC(2) User=0.01s Sys=0.00s Real=0.01s[0.254s][info][gc,marking ] GC(2) Concurrent Create Live Data[0.255s][info][gc,marking ] GC(2) Concurrent Create Live Data 0.333ms[0.258s][info][gc,start ] GC(2) Pause Cleanup[0.258s][info][gc ] GC(2) Pause Cleanup 12M->12M(20M) 0.103ms[0.258s][info][gc,cpu ] GC(2) User=0.00s Sys=0.00s Real=0.00s[0.258s][info][gc,marking ] GC(2) Concurrent Cleanup for Next Mark[0.258s][info][gc,marking ] GC(2) Concurrent Cleanup for Next Mark 0.240ms[0.258s][info][gc ] GC(2) Concurrent Cycle 8.979ms[0.261s][info][gc,heap,exit ] Heap[0.261s][info][gc,heap,exit ] garbage-first heap total 20480K, used 12827K [0x00000007bec00000, 0x00000007c0000000)[0.261s][info][gc,heap,exit ] region size 1024K, 3 young (3072K), 2 survivors (2048K)[0.261s][info][gc,heap,exit ] Metaspace used 5393K, capacity 5448K, committed 5632K, reserved 1056768K[0.261s][info][gc,heap,exit ] class space used 460K, capacity 492K, committed 512K, reserved 1048576K

疑问点

1. 每个Region大小为1M，代码中的所有对象都大于1M，应该存放Humongous，可为什么有部分存入Eden和Survivor？
2. 从堆内存监控可以看出，在没有生成局部标量前有2M的堆内存，这部分除了Metaspace和class space外，还存储什么内容？

来源：https://www.cnblogs.com/liuzhongrong/p/12347641.html