JVM 调优 ----- G1

2020-01-09T20:53:25.970+0800: 899.957: [Full GC (Allocation Failure)  2048M->355M(1192M), 0.8310098 secs]
   [Eden: 0.0B(816.0M)->0.0B(472.0M) Survivors: 0.0B->0.0B Heap: 2048.0M(2048.0M)->355.5M(1192.0M)], [Metaspace: 43882K->43741K(1089536K)]
 [Times: user=0.86 sys=0.10, real=0.83 secs] 
 
2020-01-09T20:57:40.986+0800: 1154.972: [Full GC (Allocation Failure)  2047M->79M(272M), 0.4907389 secs]
   [Eden: 0.0B(816.0M)->0.0B(104.0M) Survivors: 0.0B->0.0B Heap: 2047.9M(2048.0M)->79.8M(272.0M)], [Metaspace: 43843K->43831K(1089536K)]
 [Times: user=0.44 sys=0.10, real=0.49 secs]
 
2020-01-09T21:04:05.861+0800: 1539.847: [Full GC (Allocation Failure)  2047M->224M(752M), 0.6655056 secs]
   [Eden: 0.0B(816.0M)->0.0B(296.0M) Survivors: 0.0B->0.0B Heap: 2047.6M(2048.0M)->224.8M(752.0M)], [Metaspace: 43993K->43981K(1089536K)]
 [Times: user=0.65 sys=0.09, real=0.66 secs]
 
2020-01-09T21:20:02.111+0800: 2496.097: [Full GC (Allocation Failure)  2047M->242M(816M), 0.6904842 secs]
   [Eden: 0.0B(816.0M)->0.0B(320.0M) Survivors: 0.0B->0.0B Heap: 2047.5M(2048.0M)->242.8M(816.0M)], [Metaspace: 44214K->44202K(1089536K)]
 [Times: user=0.67 sys=0.11, real=0.69 secs] 

服务器GC频繁 youngGC  和少部分的FullGC

其中FullGC的效果非常明显，感觉年轻代小了，导致很多可以在年轻代回收的对象却到了老年代才回收

还有就是没有发生MixGC

每次FullGC前存在

2020-01-09T21:20:01.779+0800: 2495.765: [GC pause (G1 Evacuation Pause) (young) 
(to-space exhausted), 0.3133147 secs]
   [Parallel Time: 279.5 ms, GC Workers: 4]
      [GC Worker Start (ms): Min: 2495765.2, Avg: 2495765.3, Max: 2495765.3, Diff: 0.1]
      [Ext Root Scanning (ms): Min: 0.6, Avg: 0.7, Max: 0.9, Diff: 0.3, Sum: 3.0]
      [Update RS (ms): Min: 29.9, Avg: 30.1, Max: 30.3, Diff: 0.4, Sum: 120.3]
         [Processed Buffers: Min: 107, Avg: 127.8, Max: 152, Diff: 45, Sum: 511]
      [Scan RS (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.0]
      [Code Root Scanning (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.0]
      [Object Copy (ms): Min: 246.7, Avg: 248.0, Max: 248.6, Diff: 1.9, Sum: 992.0]
      [Termination (ms): Min: 0.0, Avg: 0.4, Max: 1.7, Diff: 1.7, Sum: 1.7]
         [Termination Attempts: Min: 1, Avg: 1.0, Max: 1, Diff: 0, Sum: 4]
      [GC Worker Other (ms): Min: 0.0, Avg: 0.0, Max: 0.0, Diff: 0.0, Sum: 0.1]
      [GC Worker Total (ms): Min: 279.2, Avg: 279.3, Max: 279.3, Diff: 0.1, Sum: 1117.1]
      [GC Worker End (ms): Min: 2496044.5, Avg: 2496044.6, Max: 2496044.6, Diff: 0.0]
   [Code Root Fixup: 0.0 ms]
   [Code Root Purge: 0.0 ms]
   [Clear CT: 0.1 ms]
   [Other: 33.7 ms]
      [Evacuation Failure: 30.7 ms]
      [Choose CSet: 0.0 ms]
      [Ref Proc: 0.3 ms]
      [Ref Enq: 0.0 ms]
      [Redirty Cards: 2.5 ms]
      [Humongous Register: 0.0 ms]
      [Humongous Reclaim: 0.0 ms]
      [Free CSet: 0.0 ms]
   [Eden: 88.0M(816.0M)->0.0B(816.0M) Survivors: 0.0B->0.0B Heap: 2047.5M(2048.0M)->2047.5M(2048.0M)]
 [Times: user=0.40 sys=0.08, real=0.32 secs] 
2020-01-09T21:20:02.093+0800: 2496.079: [GC pause (G1 Evacuation Pause) (young), 0.0176265 secs]
   [Parallel Time: 16.9 ms, GC Workers: 4]
      [GC Worker Start (ms): Min: 2496078.9, Avg: 2496084.2, Max: 2496087.0, Diff: 8.1]
      [Ext Root Scanning (ms): Min: 0.0, Avg: 0.7, Max: 2.9, Diff: 2.9, Sum: 2.9]
      [Update RS (ms): Min: 8.7, Avg: 10.8, Max: 13.7, Diff: 5.0, Sum: 43.1]
         [Processed Buffers: Min: 46, 

 

调试步骤  InitiatingHeapOccupancyPercent 调低了1/3  G1ReservePercent 扩大了2倍

 

由于本应用会产生大量的大对象，所以需要调低region的数量，让每个region大一点，以防对象不经过Young，直接到old区

 GC pause (G1 Humongous Allocation) (young) 直到不出现这个为准  (XX:G1HeapRegionSize)

 

上述优化之后，出现了mixGC   但是还是会出现  to-space exhausted

原因查看： Young区E:S1:S2=8:1:1，导致每个S只有81.6M，当S中存在88M的对象要复制到S时，S没有足够的空间，导致to-space exhausted，

 

目前的GC日志

4G的内存，堆给了2G

每次YGC，eden在做自行调整，但是老年代还是没有及时清除，堆看着都满了 下面图中已标红

 

增大一下GCPauseMillis时间，可以缩短GC的频率。减小一下年轻代大小

最后优化一下参数为

 

最终FullGC消失，只有YoungGC 和 MixGC