>I have this very strange problem that popped up a few days ago. When I
>restart the CFMX server, and then hit my application, the first execution
>is really slow, between 30-50 seconds. I've tracked down the templates that
>run long and then removed them by adding a <cfexit> at the top of each for
>debugging. When I do this they go down to 0ms and some other seamingly
>random template's execution time goes up to 25 seconds or so.
>
>This is a huge mystery to me at this point. My website is made up of 2
>applications (as defined by a cfapplication tag) and the other application
>starts up fine and CFMX is serving other pages fine while processing the
>startup of the other application. The processor goes to 100%.
>
>What could possibly be going on here? Not much happens when this app
>starts. Yes, there is some data written to the application scope, but FAR
>less than on the <application> which starts up in an expected timeframe.
>
>Any pointers are greatly appreciated. As it sits, if I need to cycle CFMX
>during the day and their is a load it can cause serious problems right now.
>The problem is exacerbated under load.
>

The issue, in my experience, is that there is an inherent startup delay when
accessing MX resources the first time after restarting. While your
particular application may or may not be doing something process intensive
up front, MX is an application that runs inside of a JRun instance. As such,
there are initial compilation and caching processes going on between JRun
and CFMX. Even after you are able to pull up CF pages in your browser, you
may notice a big of "sluggishness" for the first few minutes while the
background processes are acquiring the appropriate memory resources.

In short, it's just something you deal with.

Regards,
Dave.
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings]

Reply via email to