简体繁体 English

Apache（在Ubuntu上）随着时间的推移会产生大量任务。为什么和应该/应该如何解决？

[英]Apache (on Ubuntu) spawns large number of tasks over time. Why & should/how I fix it?

原文 2015-11-11 01:06:54 6 1 apache/ performance/ ubuntu/ amazon-web-services/ load

I have about 4-6 files (per user), mostly static ones being requested from an AWS instance. 我大约有4-6个文件（每个用户），大多数是从AWS实例请求的静态文件。 Peak traffic is about 1K visitors at a time (via Google Analytics Real Time). 高峰流量一次（通过Google Analytics（分析）实时）大约有1000名访客。

The instance is powerful enough and it rarely peaks at 100% CPU, has free RAM. 该实例足够强大，几乎不会在100％CPU达到峰值，并具有可用RAM。 mpm_prefork is adjusted for 1480 simultaneous connections. 已将mpm_prefork调整为1480个同时连接。 After restarting Apache2, there are around 200 processes running during peaks. 重新启动Apache2之后，高峰期间大约有200个进程在运行。

However, over a couple of days the amount of processes seems to swell up and requires me to restart apache in order to negate the timeout errors for users. 但是， 几天后，进程数量似乎膨胀了 ，需要我重新启动apache才能消除用户的超时错误。

Am I missing anything? 我有什么想念的吗？ Is this expected? 这是预期的吗？

1 个解决方案

You could set the MaxSpareServers directive to something very low like 10 or 20. When instances are not needed anymore (not serving anything or in a keep-alive state), you should only get 10 or 20 idle workers. 您可以将MaxSpareServers伪指令设置为非常低的值，例如10或20。当不再需要实例（不再提供任何服务或处于保持活动状态）时，您应该只获得10或20个空闲工作线程。

If the workers number remains the same even when serving no client, then you've got some problem. 如果即使不为客户服务，工作人员人数也保持不变，那么您就会遇到问题。 Check your keep-alive timeout setting, or disable keep-alive altogether and see it has some effect. 检查您的Keep-alive超时设置，或完全禁用Keep-Alive，并观察到它会起作用。 I've notice keep-alive is sometimes counterproductive, especially in busy environments. 我注意到保持活力有时适得其反，特别是在繁忙的环境中。