使用 nohup 或 disown 如果你要让某个进程运行在后台。

gc时查询耗时变高

Elasticsearch | 作者 kwan | 发布于2018年04月16日 | 阅读数:5291

查询语句很简单,就是filter几个字段,平时耗时100ms以内,但是在集群GC时,查询会非常慢,达到9秒。有什么优化思路和建议呢
 
日志:
[2018-04-16T13:03:54,800][WARN ][o.e.m.j.JvmGcMonitorService] [search-02] [gc][young][3961579][208874] duration [1s], collections [1]/[1.2s], total [1s]/[2.3h], memory [7.7gb]->[6.6gb]/[30.8gb], all_pools {[young] [1.1gb]->[2.1mb]/[1.4gb]}{[survivor] [41.6mb]->[47mb]/[191.3mb]}{[old] [6.5gb]->[6.5gb]/[29.1gb]}
[2018-04-16T13:03:54,868][WARN ][o.e.m.j.JvmGcMonitorService] [search-02] [gc][3961579] overhead, spent [1s] collecting in the last [1.2s]
[2018-04-16T13:04:08,946][WARN ][o.e.m.j.JvmGcMonitorService] [search-02] [gc][young][3961591][208877] duration [2.3s], collections [1]/[3s], total [2.3s]/[2.3h], memory [8gb]->[6.6gb]/[30.8gb], all_pools {[young] [1.4gb]->[4.3mb]/[1.4gb]}{[survivor] [38.2mb]->[31.3mb]/[191.3mb]}{[old] [6.5gb]->[6.5gb]/[29.1gb]}
[2018-04-16T13:04:08,946][WARN ][o.e.m.j.JvmGcMonitorService] [search-02] [gc][3961591] overhead, spent [2.3s] collecting in the last [3s]
[2018-04-16T13:05:17,747][INFO ][o.e.m.j.JvmGcMonitorService] [search-02] [gc][young][3961647][208887] duration [880ms], collections [1]/[1.2s], total [880ms]/[2.3h], memory [7.2gb]->[6.6gb]/[30.8gb], all_pools {[young] [645.4mb]->[546.9kb]/[1.4gb]}{[survivor] [41.1mb]->[30mb]/[191.3mb]}{[old] [6.5gb]->[6.6gb]/[29.1gb]}
[2018-04-16T13:05:17,747][WARN ][o.e.m.j.JvmGcMonitorService] [search-02] [gc][3961647] overhead, spent [880ms] collecting in the last [1.2s]
[2018-04-16T13:13:39,747][WARN ][o.e.l.LicenseService     ] [search-02]
已邀请:

kennywu76 - Wood

赞同来自: laoyang360 taogger

已经微信沟通过,初步判断是mapping里将keyword类型的数据设置成了数值型引起的查询性能问题,参考  https://elasticsearch.cn/article/446
 

要回复问题请先登录注册