2003. There is nothing wrong with that ... it's simply how the threading
"Ohaya" <Ohaya@NO_SPAM.cox.net> wrote in message
news:%23mmZAWP%23DHA.340@tk2msftngp13.phx.gbl...
> Hi,
>
> We are running a website on a Windows 2000 Advanced Server machine with 4
> CPUs and Hyperthreading enabled (i.e., 8 virtual processors). The web
> server is IIS5.
>
> Our application uses ASP.net, and we originally were testing the
application
> on a test system with a single CPU and no hyperthreading.
>
> When we moved the application to the production machine (the one with 4
> CPUs), we noticed a strange (to me) thing.
>
> Our application is such that we kind of expected processing a request
would
> not be quick (the processing for a single request is somewhat
long-running),
> but when we tested on the SMP server, we noted that when a request was
sent
> to the web server, we'd see CPU Usage on 1 processor go up (to about
100%),
> then after awhile, CPU Usage on the next processor would go up and CPU
usage
> on the 1st processor would drop to almost 0%. Then after awhile, CPU
Usage
> on the next (3rd) processor would go up, and CPU Usage on the 2nd
processor
> would drop to almost 0%, etc. This was all during the processing for a
> single request to the web server.
>
> I've been doing some searching, trying to figure out what's going on, and
I
> ran across the following post from 3/4/2004:
>
>
http://groups.google.com/groups?q=g:thl3654827175d&dq=&hl=en&lr=&ie=UTF-8&oe=UTF-8&selm=uTr9i5n4CHA.2324%40TK2MSFTNGP10.phx.gbl
>
> The above post seems to be indicating a problem where ASP.net would
"queue"
> up processing, but it seems like it's mostly talking about a situation
where
> multiple (i.e., 2) requests come into the web server, and processing of
the
> 2nd request would, under some circumstances, get held up because of this
> problem.
>
> We haven't tried the workarounds mentioned in the above post yet (don't
have
> access to the production system), but I'm wondering if anyone here had
> experience with this problem, and if this problem might also manifest
itself
> to look something like what we're seeing?
>
> Thanks in advance!!
>
> Jim
>
>