现象

收到系统报警，查察一台呆板频繁FULLGC，且该处事超时。
这是一台4核8G的呆板, 利用jdk1.8.0_45-b14。

我们可以直接通过jstat等来调查。这次我先通过CPU开始。
top查察后该java历程的运行状况为

Tasks: 161 total,   3 running, 158 sleeping,   0 stopped,   0 zombie
Cpu(s): 32.1%us,  1.9%sy,  0.0%ni, 65.9%id,  0.0%wa,  0.0%hi,  0.1%si,  0.0%st
Mem:   8059416k total,  7733088k used,   326328k free,   147536k buffers
Swap:  2096440k total,        0k used,  2096440k free,  2012212k cached
  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
15178 x   20   0 7314m 4.2g  10m S 98.9 55.1   4984:05 java

RES 占用4.2g CPU占用98%

top -H -p 15178

后发明

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
15234 x   20   0 7314m 4.2g  10m R 70.2 55.1 781:50.32 java
15250 x   20   0 7314m 4.2g  10m S 52.9 55.1 455:12.27 java

一个PID为 15234 的轻量级历程，对应于java中的线程的当地线程号。
通过printf '%x\n' 15234计较出16进制的 3b82
通过jstack -l 15178 然后搜索该线程,发明其为
"Concurrent Mark-Sweep GC Thread" os_prio=0 tid=0x00007ff3e4064000 nid=0x3b82 runnable

定位了确实由GC导致的系统不行用。

jinfo -flags

查察该VM参数

Non-default VM flags: -XX:CICompilerCount=3 -XX:CMSFullGCsBeforeCompaction=0 -XX:CMSInitiatingOccupancyFraction=80 -XX:+DisableExplicitGC -XX:ErrorFile=null -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=null -XX:InitialHeapSize=2147483648 -XX:MaxHeapSize=2147483648 -XX:MaxNewSize=348913664 -XX:MaxTenuringThreshold=6 -XX:MinHeapDeltaBytes=196608 -XX:NewSize=348913664 -XX:OldPLABSize=16 -XX:OldSize=1798569984 -XX:+PrintCommandLineFlags -XX:+PrintGC -XX:+PrintGCDateStamps -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+UseCMSCompactAtFullCollection -XX:+UseCompressedClassPointers -XX:+UseCompressedOops -XX:+UseConcMarkSweepGC -XX:+UseParNewGC

jmap -heap 查察统计功效

Debugger attached successfully.
Server compiler detected.
JVM version is 25.45-b02
using parallel threads in the new generation.
using thread-local object allocation.
Concurrent Mark-Sweep GC
Heap Configuration:
   MinHeapFreeRatio         = 40
   MaxHeapFreeRatio         = 70
   MaxHeapSize              = 2147483648 (2048.0MB)
   NewSize                  = 348913664 (332.75MB)
   MaxNewSize               = 348913664 (332.75MB)
   OldSize                  = 1798569984 (1715.25MB)
   NewRatio                 = 2
   SurvivorRatio            = 8
   MetaspaceSize            = 21807104 (20.796875MB)
   CompressedClassSpaceSize = 1073741824 (1024.0MB)
   MaxMetaspaceSize         = 17592186044415 MB
   G1HeapRegionSize         = 0 (0.0MB)
Heap Usage:
New Generation (Eden + 1 Survivor Space):
   capacity = 314048512 (299.5MB)
   used     = 314048512 (299.5MB)
   free     = 0 (0.0MB)
   100.0% used
Eden Space:
   capacity = 279183360 (266.25MB)
   used     = 279183360 (266.25MB)
   free     = 0 (0.0MB)
   100.0% used
From Space:
   capacity = 34865152 (33.25MB)
   used     = 34865152 (33.25MB)
   free     = 0 (0.0MB)
   100.0% used
To Space:
   capacity = 34865152 (33.25MB)
   used     = 0 (0.0MB)
   free     = 34865152 (33.25MB)
   0.0% used
concurrent mark-sweep generation:
   capacity = 1798569984 (1715.25MB)
   used     = 2657780412087605496 (2.5346569176555684E12MB)
   free     = 15057529128475 MB
   1.4777186518907263E11% used
32778 interned Strings occupying 3879152 bytes.

利用jstat -gcutil 15178 1s调查一段时间GC状况

jstat -gccause 15178 1s
  S0     S1     E      O      M     CCS    YGC     YGCT    FGC    FGCT     GCT    LGCC                 GCC
100.00   0.00 100.00 100.00  97.78  95.59   1906   26.575 61433 217668.188 217694.763 Allocation Failure   Allocation Failure
  0.00   0.00  96.51 100.00  97.78  95.59   1906   26.575 61433 217672.991 217699.566 Allocation Failure   No GC
100.00   0.00 100.00 100.00  97.78  95.59   1906   26.575 61434 217672.991 217699.566 Allocation Failure   Allocation Failure
100.00   0.00 100.00 100.00  97.78  95.59   1906   26.575 61434 217672.991 217699.566 Allocation Failure   Allocation Failure
100.00   0.00 100.00 100.00  97.78  95.59   1906   26.575 61434 217672.991 217699.566 Allocation Failure   Allocation Failure

可以看到Old区满了，而且Eden区域的工具没有触发YGC直接提升到Old区中，可是Full GC没有释放出空间。这是由于在当暮年月的持续空间小于新生代所有工具巨细时，MinorGC前会查抄下平均每次提升到Old区的巨细是否大于Old区的剩余空间，假如大于可能当前的配置HandlePromotionFailure为false则直打仗发FullGc,不然会先举办MinorGC。
关于FullGC和MajorGC的区别，可以不要太纠结.