[racket] Running into severe scaling issues with plt-web-server

I have very long running instance of the Web server... running for
months and months with pretty steady access. For instance, the DrDr
Web app never needs to be restarted.

I don't want to be too dismissive of questions like this, but I
haven't noticed what you're talking about as being rooted in the
server itself. (Every once in a while I discover little things that
could be the root of problems like that. For example, there was an
additional thread per connection that was removed before the last
release.)

As for things going on in servlets, it can be easy to make
continuations too big (but you say you're not using them). It can be
easy to leave the server side connection hanging, I'd use lsof to see
if that's the case.

I'd say the first thing to do is gather data... such as the output of
'uptime' and memory usage of the server and the servlet (using
current-memory-use)... until you get a crash

Jay





On Wed, Apr 10, 2013 at 12:10 PM, Danny Yoo <dyoo at hashcollision.org> wrote:
> I wanted to check with other folks to see if anyone else has been
> having problems with long-running instances of PLT web server.  I'm
> not using continuations, but just the standard request/response style
> of webapp.
>
> It's the compiler server for wescheme: I'm running into severe
> problems where the server eventually hoses, either exhausting memory
> or cpu or capacity.
>
> Like any complex system, there are several places that may be
> contributing to the problem.  It may be my servlet.  I'm right in the
> middle of putting a racket/sandbox around the servlet process now to
> see if that detects something unusual.
>
> What else can I do to help debug the situation?
> ____________________
>   Racket Users list:
>   http://lists.racket-lang.org/users



-- 
Jay McCarthy <jay at cs.byu.edu>
Assistant Professor / Brigham Young University
http://faculty.cs.byu.edu/~jay

"The glory of God is Intelligence" - D&C 93