jvm GC日志解读

0.产生日志

0.1 IDE

运行下面代码会得到gc日志。

public class GCTest {
	private static final int _1MB = 1024 * 1024;
	
	/**
	 * VM参数：-Xms20M -Xmx20M -Xmn10M -XX:+PrintGCDetails -XX:SurvivorRatio=3  -Xloggc:d:/gclogdir/1.txt -XX:+PrintGCTimeStamps
	  */
	public static void testAllocation() {
	 	byte[][] allocation=new byte[10][];
	 	allocation[0] = new byte[1 * _1MB];
	 	allocation[1] = new byte[1 * _1MB];
	 	allocation[2] = new byte[1 * _1MB];
	 	allocation[3] = new byte[1 * _1MB];
	 	allocation[4] = new byte[1 * _1MB];
	 	allocation[5] = new byte[1 * _1MB];
	 	allocation[6] = new byte[1 * _1MB];
	 	allocation[7] = new byte[1 * _1MB];
	 	allocation[8] = new byte[1 * _1MB];
	 	allocation[8] = new byte[1 * _1MB];
	 	allocation[9] = new byte[1 * _1MB];
	}
	
	public static void main(String[] args) {
		testAllocation();
	}
	
}

/*
 * Java HotSpot(TM) 64-Bit Server VM (25.0-b70) for windows-amd64 JRE (1.8.0-b132), built on Mar  4 2014 03:46:18 by "java_re" with MS VC++ 10.0 (VS2010)
Memory: 4k page, physical 16568252k(8814280k free), swap 33134668k(24737308k free)
CommandLine flags: -XX:InitialHeapSize=20971520 -XX:MaxHeapSize=20971520 -XX:MaxNewSize=10485760 -XX:NewSize=10485760 -XX:+PrintGC -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:SurvivorRatio=3 -XX:+UseCompressedClassPointers -XX:+UseCompressedOops -XX:-UseLargePagesIndividualAllocation -XX:+UseParallelGC 
0.117: [GC (Allocation Failure) [PSYoungGen: 6085K->1672K(8192K)] 6085K->5776K(18432K), 0.0021273 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
0.119: [GC (Allocation Failure) [PSYoungGen: 6982K->1664K(8192K)] 11086K->9864K(18432K), 0.0021126 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
0.121: [Full GC (Ergonomics) [PSYoungGen: 1664K->0K(8192K)] [ParOldGen: 8200K->9739K(10240K)] 9864K->9739K(18432K), [Metaspace: 2508K->2508K(1056768K)], 0.0040641 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 
Heap
 PSYoungGen      total 8192K, used 1085K [0x00000000ff600000, 0x0000000100000000, 0x0000000100000000)
  eden space 6144K, 17% used [0x00000000ff600000,0x00000000ff70f748,0x00000000ffc00000)
  from space 2048K, 0% used [0x00000000ffe00000,0x00000000ffe00000,0x0000000100000000)
  to   space 2048K, 0% used [0x00000000ffc00000,0x00000000ffc00000,0x00000000ffe00000)
 ParOldGen       total 10240K, used 9739K [0x00000000fec00000, 0x00000000ff600000, 0x00000000ff600000)
  object space 10240K, 95% used [0x00000000fec00000,0x00000000ff582dc8,0x00000000ff600000)
 Metaspace       used 2515K, capacity 4486K, committed 4864K, reserved 1056768K
  class space    used 271K, capacity 386K, committed 512K, reserved 1048576K
*/

0.2 server

-Xloggc:../gclogdir/logc.txt  

指定gc日志的打印位置，注意必须指定到文件，不能为目录。

当应用重启时，会产生新的gc.log，旧的gc.log自动被重命名为形如 " gc.log.20161103103532"这样的格式。

1.Parallel Scavenge

这是一款年轻代GC器。

293.271: [GC [PSYoungGen: 300865K->6577K(310720K)] 392829K->108873K(417472K), 0.0176464 secs] [Times: user=0.06 sys=0.00, real=0.01 secs] 

每行是一次GC的信息，以本条为例，解读如下。
293.271是本次GC发生的时间，从jvm启动起开始计时，单位为秒。
GC表示这是一次Minor GC（新生代垃圾收集）。
[PSYoungGen: 300865K->6577K(310720K)] 。格式为[PSYoungGen: a->b(c)].
PSYoungGen，表示新生代使用的是多线程垃圾收集器Parallel Scavenge。a为GC前新生代已占用空间，b为GC后新生代已占用空间。新生代又细分为一个Eden区和两个Survivor区,Minor GC之后Eden区为空，b就是Survivor中已被占用的空间。括号里的c表示整个年轻代的大小。
392829K->108873K(417472K)，格式为x->y(z)。x表示GC前堆的已占用空间，y表示GC后堆已占用空间，z表示堆的总大小。
由新生代和Java堆占用大小可以算出年老代占用空间，此例中就是417472K-310720K=106752K。
0.0176464 secs表示本次GC所消耗的时间。
[Times: user=0.06 sys=0.00, real=0.01 secs] 提供cpu使用及时间消耗，user是用户态消耗的cpu时间，sys是系统态消耗的cpu时间,real是实际的消耗时间。

2.Parallel Old

这是Parallel Scavenge收集器的老年代版本。

0.107: [Full GC (Ergonomics) [PSYoungGen: 1664K->0K(8192K)] [ParOldGen: 8200K->9739K(10240K)] 9864K->9739K(18432K), [Metaspace: 2508K->2508K(1056768K)], 0.0038104 secs] [Times: user=0.00 sys=0.00, real=0.00 secs] 

3.CMS

2016-11-02T10:17:27.849+0800: 48234.161: [GC [1 CMS-initial-mark: 1781099K(2097152K)] 1962885K(4019584K), 0.0530810 secs] [Times: user=0.05 sys=0.00, real=0.05 secs] 
2016-11-02T10:17:27.903+0800: 48234.214: [CMS-concurrent-mark-start]
2016-11-02T10:17:30.294+0800: 48236.606: [CMS-concurrent-mark: 2.384/2.391 secs] [Times: user=2.55 sys=0.04, real=2.39 secs] 
2016-11-02T10:17:30.294+0800: 48236.606: [CMS-concurrent-preclean-start]
2016-11-02T10:17:30.303+0800: 48236.615: [CMS-concurrent-preclean: 0.009/0.009 secs] [Times: user=0.01 sys=0.00, real=0.01 secs] 
2016-11-02T10:17:30.303+0800: 48236.615: [CMS-concurrent-abortable-preclean-start]
 CMS: abort preclean due to time 2016-11-02T10:17:35.525+0800: 48241.836: [CMS-concurrent-abortable-preclean: 5.216/5.221 secs] [Times: user=5.41 sys=0.06, real=5.23 secs] 
2016-11-02T10:17:35.526+0800: 48241.837: [GC[YG occupancy: 289491 K (1922432 K)]48241.838: [Rescan (parallel) , 0.0558290 secs]48241.893: [weak refs processing, 0.0063180 secs]48241.900: [class unloading, 0.0323730 secs]48241.932: [scrub symbol & string tables, 0.0297790 secs] [1 CMS-remark: 1781099K(2097152K)] 2070590K(4019584K), 0.1310790 secs] [Times: user=0.26 sys=0.00, real=0.13 secs] 
2016-11-02T10:17:35.657+0800: 48241.969: [CMS-concurrent-sweep-start]
2016-11-02T10:17:36.176+0800: 48242.488: [CMS-concurrent-sweep: 0.503/0.519 secs] [Times: user=0.52 sys=0.01, real=0.52 secs] 
2016-11-02T10:17:36.176+0800: 48242.488: [CMS-concurrent-reset-start]
2016-11-02T10:17:36.183+0800: 48242.495: [CMS-concurrent-reset: 0.007/0.007 secs] [Times: user=0.01 sys=0.00, real=0.00 secs] 

一共经历 了4个阶段。需要关注的耗时为 初始标记与 重新标记，低于0.2秒就没什么问题。

[CMS-xxx   1781099K(2097152K)] 1962885K(4019584K)，形如[CMS-xxx   a(b)] c(d)

a: old区占用大小

b: old区总大小

c: heap占用大小

d: heap 总大小

[YG occupancy: 289491 K (1922432 K)] ，形如 [YG occupancy: a (b)]

a: young区占用大小

b: yong区总大小

3.1 CMS-initial-mark

此阶段暂停工作线程，用时0.05秒。

3.2 CMS-concurrent-mark

与工作线程并发运行。

3.3 CMS-remark 

此阶段暂停工作线程，用时0.13秒。

3.4 CMS-concurrent-sweep

与工作线程并发运行。

4. -XX:+PrintTenuringDistribution -XX:+PrintHeapAtGC

2016-11-03T14:41:12.877+0800: 14740.294: [GC 14740.295: [ParNew
Desired survivor size 178946048 bytes, new threshold 16 (max 20)
- age   1:   12374344 bytes,   12374344 total
- age   2:    2538032 bytes,   14912376 total
- age   3:    2081880 bytes,   16994256 total
- age   4:     290784 bytes,   17285040 total
- age   5:     787312 bytes,   18072352 total
- age   6:     450520 bytes,   18522872 total
- age   7:     335016 bytes,   18857888 total
- age   8:     661552 bytes,   19519440 total
- age   9:    1021048 bytes,   20540488 total
- age  10:     130816 bytes,   20671304 total
- age  11:     172536 bytes,   20843840 total
- age  12:     384968 bytes,   21228808 total
- age  13:     556328 bytes,   21785136 total
- age  14:      99112 bytes,   21884248 total
- age  15:   19091864 bytes,   40976112 total
: 3551236K->58224K(3844800K), 0.0259700 secs] 5186234K->1693221K(5941952K), 0.0262100 secs] [Times: user=0.08 sys=0.00, real=0.03 secs] 
Heap after GC invocations=73 (full 2):
 par new generation   total 3844800K, used 58224K [0x0000000668000000, 0x0000000768000000, 0x0000000768000000)
  eden space 3495296K,   0% used [0x0000000668000000, 0x0000000668000000, 0x000000073d560000)
  from space 349504K,  16% used [0x0000000752ab0000, 0x000000075638c030, 0x0000000768000000)
  to   space 349504K,   0% used [0x000000073d560000, 0x000000073d560000, 0x0000000752ab0000)
 concurrent mark-sweep generation total 2097152K, used 1634997K [0x0000000768000000, 0x00000007e8000000, 0x00000007e8000000)
 concurrent-mark-sweep perm gen total 274488K, used 197983K [0x00000007e8000000, 0x00000007f8c0e000, 0x0000000800000000)
}
{Heap before GC invocations=73 (full 2):
 par new generation   total 3844800K, used 3553520K [0x0000000668000000, 0x0000000768000000, 0x0000000768000000)
  eden space 3495296K, 100% used [0x0000000668000000, 0x000000073d560000, 0x000000073d560000)
  from space 349504K,  16% used [0x0000000752ab0000, 0x000000075638c030, 0x0000000768000000)
  to   space 349504K,   0% used [0x000000073d560000, 0x000000073d560000, 0x0000000752ab0000)
 concurrent mark-sweep generation total 2097152K, used 1634997K [0x0000000768000000, 0x00000007e8000000, 0x00000007e8000000)
 concurrent-mark-sweep perm gen total 274488K, used 197985K [0x00000007e8000000, 0x00000007f8c0e000, 0x0000000800000000)