WebLogic threading in the webcontainer

Question

Hi.
How does a long running servlet execution work with regards to threading in WebLogic6.1sp3.
I know there is a thread pool and that new incoming requests are put in a workqueue
to await to be assigned to a thread.
If a servlet has a long running request can there be switching of the thread instance
assigned to a request? With a request being "interrupted" after a while of execution,
the "workload" returned to the workqueue and made ready to be assigned a different
running thread until completion?
The alternative would be the guaranteed execution of the request by the same thread
instance. (Wouldn't this destroy the notion of having more concurrent requests
being processed, not queued, than the total number of threads assigned to the
server? )

Answer 1

Yup, you can use it with long running requests :
http://groups.google.com/groups?selm=3c1fab45%40newsgroups.bea.com&oe=UTF-8&
output=gplain
On the other hand this is only useful if you can execute request
asynchronously (server push for example).
"Trond Strømme" <[email protected]> wrote in message
news:[email protected]..
>
But is this done with long lived requests?
trond
"Dimitri I. Rakitine" <[email protected]> wrote:
"Trond Strømme" <[email protected]> wrote in message
news:3d934796$[email protected]..
Hi.
How does a long running servlet execution work with regards to
threading
in WebLogic6.1sp3.
I know there is a thread pool and that new incoming requests are putin a
workqueue
to await to be assigned to a thread.
If a servlet has a long running request can there be switching of thethread instance
assigned to a request?Yes, there is a way to detach ServletResponse from an execute thread,
but,
unfortunately, this
functionality is not documented ;-(
With a request being "interrupted" after a while of execution,
the "workload" returned to the workqueue and made ready to be assigneda
different
running thread until completion?
The alternative would be the guaranteed execution of the request bythe
same thread
instance. (Wouldn't this destroy the notion of having more concurrentrequests
being processed, not queued, than the total number of threads assignedto
the
server? )
Dimitri
Dimitri