Young GC的时间超长

1.观察gc日志发现young gc日志很长

2020-01-18T11:01:46.486+0800: 10670069.343: [GC2020-01-18T11:01:54.876+0800: 10670077.733: [ParNew: 1119126K->770K(1258304K), 0.0078630 secs] 3377296K->2258940K(4054528K), 8.3983450 secs] [Times: user=0.05 sys=0.05, real=8.40 secs] 
Heap after GC invocations=445687 (full 9):
 par new generation   total 1258304K, used 770K [0x00000006fae00000, 0x0000000750350000, 0x0000000750350000)
  eden space 1118528K,   0% used [0x00000006fae00000, 0x00000006fae00000, 0x000000073f250000)
  from space 139776K,   0% used [0x000000073f250000, 0x000000073f310890, 0x0000000747ad0000)
  to   space 139776K,   0% used [0x0000000747ad0000, 0x0000000747ad0000, 0x0000000750350000)
 concurrent mark-sweep generation total 2796224K, used 2258170K [0x0000000750350000, 0x00000007fae00000, 0x00000007fae00000)
 concurrent-mark-sweep perm gen total 77656K, used 46556K [0x00000007fae00000, 0x00000007ff9d6000, 0x0000000800000000)

2.通过grafana查看资源使用，io负载高

3.参考文档分析原因

YoungGC 停顿超长时间
https://blog.csdn.net/taoy86/article/details/79746652
https://engineering.linkedin.com/blog/2016/02/eliminating-large-jvm-gc-pauses-caused-by-background-io-traffic

young gc时间过长可能是应为记录gc日志的操作被系统的io给阻塞了

4.解决方法

方案一：
将gc的日志记录到内存文件系统（/dev/shm）避免io阻塞
缺点:因为使用内存文件系统，重启时记录丢失

方案二:
单独为gc日志准备一个硬盘，或者SSD（高速硬盘）