插槽交换或配置更改后，繁忙的Web应用程序关闭 [英] Busy web app goes down after slot swap or config change
问题描述
我有一个繁忙的Web应用程序在P3v2大小的实例(从1自动缩放到4)上作为Azure Web应用程序运行.大多数情况下，应用程序运行完美.如果我将广告位换成生产版或更改应用设置，那么会时不时地(尤其是在繁忙时段) 在生产广告位中，该应用完全消失. CPU和内存看起来不错，但所有请求都将在230秒后超时，事件日志或应用洞察中不会记录其他错误.重新启动应用程序无效，向外扩展无效.有时候这 通过再次换回插槽来解决.如果不解决，最终这个问题将自行解决(但这可能需要一个小时的时间)，并且该应用程序再次可以正常运行.
I have a busy web app running as an Azure web app on P3v2 size instances (1 autoscaling to 4). Most of the time the app runs perfectly. Every now and again, especially during busy periods, if I swap slots to production or make a change to the app settings of the production slot, the app completely dies. CPU and memory look fine but all requests will time out after 230 seconds, no other error is logged in the event log or app insights. Restarting the app has no effect, scaling out has no effect. Sometimes this is solved by swapping the slots back again. If left, eventually, this problem solves itself (but this can take up to an hour) and the app starts working fine again.
如果我在停机期间进行配置文件，我会看到很多502/503错误，但没有太多其他错误.简档中的许多请求具有最慢管道事件". ="CLRThreadPoolQueue".我不确定这意味着什么...我认为该应用程序很挣扎 随着流量的突然增加.如果是这样，该如何缓解呢?
If I take a profile during the down time I see lots of 502/503 errors but not much else. Lots of the requests in the profile have "Slowest Pipeline Event" = "CLRThreadPoolQueue". I'm not sure what this means... I assume the app is struggling with the sudden increase in traffic. If so, how do I mitigate this?
推荐答案
将广告位换成生产版不会导致停机(整个广告位的想法). 请在此处提供站点名称(或 )，更重要的是，当您执行插槽交换并观察到停机时间时，是UTC.
Swap slot to Production should not cause the downtime (the whole idea for slot). Please provide sitename here (or indirectly) and, more importantly, the UTC when you performed slot swap and observed downtime.
但是，更改生产版位的appSettings可能会导致停机，因为必须重新启动网站以获取更改. 这是设计使然.
However, changing the Production slot appSettings will potentially incur downtime since site will have to be restarted to pick up the changes. That's by design.
这篇关于插槽交换或配置更改后，繁忙的Web应用程序关闭的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
