应用线程的内存问题—容器中的业务频繁fullgc

1.现象

业务容器化后rt从200ms上涨至400多ms，翻了将近一倍

2.问题排查

rt上涨原因是应用一直fullgc

gc日志：

Java HotSpot(TM) 64-Bit Server VM (25.121-b13) for linux-amd64 JRE (1.8.0_121-b13), built on Dec 12 2016 16:36:53 by "java_re" with gcc 4.3.0 20080428 (Red Hat 4.3.0-8)
Memory: 4k page, physical 17825792k(24917156k free), swap 0k(0k free)
CommandLine flags: -XX:+CMSClassUnloadingEnabled -XX:CMSInitiatingOccupancyFraction=60 -XX:+CMSParallelRemarkEnabled -XX:ConcGCThreads=4 -XX:+DisableExplicitGC -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/home/www/njord/logs/java.hprof -XX:InitialHeapSize=10737418240 -XX:+ManagementServer -XX:MaxHeapSize=10737418240 -XX:MaxMetaspaceFreeRatio=100 -XX:MaxMetaspaceSize=2147483648 -XX:MaxNewSize=8589934592 -XX:MetaspaceSize=2147483648 -XX:MinMetaspaceFreeRatio=0 -XX:NewSize=8589934592 -XX:OldPLABSize=16 -XX:-OmitStackTraceInFastThrow -XX:ParallelGCThreads=8 -XX:+PrintGC -XX:+PrintGCDateStamps -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:SoftRefLRUPolicyMSPerMB=0 -XX:SurvivorRatio=8 -XX:+UseCMSCompactAtFullCollection -XX:+UseCompressedClassPointers -XX:+UseCompressedOops -XX:+UseConcMarkSweepGC -XX:+UseParNewGC
2020-06-30T15:11:30.700+0800: 5.942: [GC (Allocation Failure) 2020-06-30T15:11:30.700+0800: 5.942: [ParNew: 6710912K->64573K(7549760K), 0.0732810 secs] 6710912K->64573K(9646912K), 0.0734148 secs] [Times: user=0.23 sys=0.04, real=0.07 secs]
2020-06-30T15:11:35.976+0800: 11.218: [GC (Allocation Failure) 2020-06-30T15:11:35.976+0800: 11.218: [ParNew: 6775485K->95117K(7