Salut,
Thomas Gideon wrote:
> I am still working on JFA's suggestion of how to use Grizzly's Comet
> support to move file uploads to their own threads. Our application is
> often used by an entire classroom at the same time and in a lab
> exercise all of the students may attempt to upload larger multimedia
> files over a slow link. We've definitely seen this consume our
> available request processing threads in Glassfish.
>
> I have mostly plumbed together the necessary pieces but am having
> trouble making my code work reliably under Glassfish 2.1. Sometimes
> it works perfectly, other times it encounters errors in the Coyote
> request classes.
Do you have the exact error?
>
> My annotated sources, a smaller project just to try to build a proof
> of concept, are here: http://bitbucketlabs.net/files/upload-comet.zip
>
> Am I missing something necessary to reliably access a Coyote
> HttpServletRequest implementation from a separate thread? The errors
> also seem to multiply the more times I use the upload form. Am I not
> closing out the CometHandler and the original request/response
> correctly?
>
> I'd very much like to be able to deploy a solution like this as it
> avoids the potential problems of creating a custom file upload server
> socket on a separate port.
>
> Any help, either from Jean-Francois or anyone else on the list, is appreciated.
I'm looking at it now.
Thanks!
-- Jeanfrancois
>
>
> Thanks! Thomas
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_grizzly.dev.java.net
> For additional commands, e-mail: users-help_at_grizzly.dev.java.net
>