> I suspect the problem is that GF is actually closing the connection
> right after firing interrupt() on thread.
> So possible the browser receives connection terminate signal
> before/during 200 response, that's why it reissues the request.
Oho? I thought along the lines of some obscure HTTP caching effect,
but I'm not an expert in the field.
> Which GF version are you using?
GlassFish Server Open Source Edition 3.1 on WinXP
> I can prepare the patch for you to make sure it's the case.
Please do. Unless you can reproduce the phenomenon on your box?
In any case, it would be nice to have this fixed in the upcoming
next release of GF.
Also, what about long-running Multipart-MIME uploads? Currenly, the 'Request
Timeout' affects long-running
uploads, even if they are not "slow" by any measure. (And the timeout
currently causes ANOTHER identical
upload, eek!) In my opinion, there should be no timeout whatsoever, as long
as data is read at a reasonable speed.
If there absolutely has to be a timeout, it should be different, and have a
default value of several hours.
Thanks for caring!
--
[Message sent by forum member 'tmpsa']
View Post: http://forums.java.net/node/855551