Redmine性能不一致
我在 Linux x86_64 系统上部署了一个 redmine 实例,但我正在努力解决一个有趣的性能不一致问题。它的行为是这样的:
将其单独放置几个小时(根本没有向应用程序发出任何请求),然后建立连接,需要很长时间才能响应前几个请求(可能长达 15 秒以上,有时每个大约 1 分钟响应)。在最初的几个请求之后,它的性能非常快,几乎是立即的,我从日志中可以看到,响应时间不到 100 毫秒。
我尝试了几种使用 mongrel 运行它的选项,并将其部署在乘客 Nginx 和乘客 Apache 上。该行为可以简单地重现。我测试发现,在繁忙的情况下,完全不会有问题。
这是怎么回事?我该如何改进这个?
I have a redmine instance deployed on a Linux x86_64 system and I am struggle with a funny performance inconsistency issue. It behave like this:
Leave it alone for a couple of hours (no request to the application at all) then establish a connection, it takes a very long time to response to the first few requests (can be as bad as 15 seconds+, sometimes every for response at about 1 minute). After the first few requests, it performance very quickly, almost instantly where I can see from the log that it takes less than 100ms to response.
I have tried a couple of options to run it using mongrel, deployed it on passenger-Nginx, and on passenger-Apache. The behaviour simply reproduceable. I realised from the testing when it is keeping busy, there will be no problem at all.
What is it all about? How can I improve this?
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
该问题很可能是由乘客池闲置引起的。您可以将 PassengePoolIdleTime 增加到更大的数字,或者如果您根本不希望它超时,则将该值设置为 0。类似下面的 Apache 配置应该可以解决问题
The issue is most likely caused by the the Passenger pool idleling out. You can either increase the PassengePoolIdleTime to a larger number, or if you don't wan it to time out at all set the value to 0. An Apache config something like below should fix the issues