使用 nohup 或 disown 如果你要让某个进程运行在后台。

es启动后JVM Heap占用越来越大

Elasticsearch | 作者 hide | 发布于2020年11月23日 | 阅读数:2970

配置:
es版本 6.6.2
java版本 1.8
服务器物理内存32G,j分配给es16G,就一个节点,磁盘一个T,并且装成了ZFS,并启用了lz4压缩
=====================================
现象:
启动es后一开始很正常,jvm heap就只占用一半左右,运行一周后就占用得越来越大,会占用14G。是内存没被回收导致的吗?
请大神指教
=====================================
[2020-11-23T09:09:57,038][WARN ][o.e.m.j.JvmGcMonitorService] [node_master] [gc][851707] overhead, spent [39s] collecting in the last [39.2s]
[2020-11-23T09:10:15,110][INFO ][o.e.m.j.JvmGcMonitorService] [node_master] [gc][851725] overhead, spent [275ms] collecting in the last [1s]
[2020-11-23T09:11:09,384][WARN ][o.e.m.j.JvmGcMonitorService] [node_master] [gc][old][851746][16337] duration [33.1s], collections [1]/[34.1s], total [33.1s]/[46.4m], memory [15.4gb]->[12.8gb]/[15.9gb], all_pools {[young] [139.8mb]->[66.6mb]/[532.5mb]}{[survivor] [62.7mb]->[44.6mb]/[66.5mb]}{[old] [15.2gb]->[12.7gb]/[15.3gb]}
[2020-11-23T09:11:09,384][WARN ][o.e.m.j.JvmGcMonitorService] [node_master] [gc][851746] overhead, spent [33.5s] collecting in the last [34.1s]
已邀请:

hide

赞同来自:

屏幕截图_2020-11-23_111007.png

 

要回复问题请先登录注册