Java 11 - G1GC - 需要了解年轻代 gc 日志及其对应用程序线程的影响

问题描述

我正在运行一个启用了 G1GC 的 Java 应用程序，我正在打印如下所示的 gc 日志：

[4074.182s][info][gc] GC(1017) Pause Young (normal) (G1 Evacuation Pause) 6869M->6115M(16384M) 7430.860ms
[4082.328s][info][gc] GC(1018) Pause Young (normal) (G1 Evacuation Pause) 6875M->6121M(16384M) 7058.485ms
[4089.592s][info][gc] GC(1019) Pause Young (normal) (G1 Evacuation Pause) 6881M->6126M(16384M) 6172.693ms
[4105.799s][info][gc] GC(1021) Pause Young (normal) (G1 Evacuation Pause) 6892M->6137M(16384M) 6845.922ms
[4114.418s][info][gc] GC(1022) Pause Young (normal) (G1 Evacuation Pause) 6897M->6142M(16384M) 7563.613ms
[4122.805s][info][gc] GC(1023) Pause Young (normal) (G1 Evacuation Pause) 6902M->6148M(16384M) 7326.923ms
[4131.270s][info][gc] GC(1024) Pause Young (normal) (G1 Evacuation Pause) 6908M->6155M(16384M) 7402.222ms
[4140.574s][info][gc] GC(1025) Pause Young (normal) (G1 Evacuation Pause) 6915M->6161M(16384M) 8226.283ms
[4149.429s][info][gc] GC(1026) Pause Young (normal) (G1 Evacuation Pause) 6921M->6167M(16384M) 7779.282ms
[4157.078s][info][gc] GC(1027) Pause Young (normal) (G1 Evacuation Pause) 6927M->6172M(16384M) 6559.166ms

我对这个日志的理解是：

[4074.182s][info][gc] GC(1017) Pause Young (normal) (G1 Evacuation Pause) 6869M->6115M(16384M) 7430.860ms

As per my understanding this shows that 4074.182 seconds after application started,a young generation 
Stop the world gc was kicked in,and heap size before gc heap size was 6869M 
and after gc it reduced to 6115M and it took 7430.860ms (more than 7 seconds).

我的问题：

1. Is this young gc stopping all the application threads ?
2. If it is stopping all the application threads then looks like application is not 
   running at all during all these gc logs were printed because:
    after [4074.182s] young gc kicked in and it took aroung 7.5 seconds (7430.860ms)
    after ~8 seconds,[4082.328s] yound gc kicked in again and took around 7 SEOnds and so on.
   So it looks like application is not getting any free slot here to run ?  

解决方法

是的，您的应用程序已停止，在此期间所有线程都已暂停。有趣的是，GC 只能在大约 6-7 秒内回收 700MB，这绝对是可怕的。如果没有调试消息，就不可能说出为什么会发生这种情况，可能是机器上启动了线程，可能是区域之间的连接太多，因此需要花费大量时间扫描记住的集合，或者任何其他原因。

但最重要的是，在这些周期中，您的应用程序的性能非常糟糕；你说得对。