I don't think that there is something like a general best
place. It depends on design of the application.
I like to keep the timeframe as short as possible where
resources like databases or memory are allocated. That has
for us the implication that the allocation of a resource
is as near as possible to it's use and so is the deallocation.
So I don't like the idea to use a filter to allocate and
deallocate resources.
Filter are for me tools to modify the request or the response
or a place to do authentication and things like that.
I admit that this is a quite personal standpoint.
Some ways to get a better idea what happens:
- If you are using unix/linux send a kill -QUIT <pid> to
the vm. Then you will get a stacktrace for each thread in
the core file. (I'm not shure if this is right in the
windows world: start tomcat in a console so that the
console stays open and hit something like control-z in the
console window)
- Attach a debugger and look at the stack for the threads
- Use log messages to get a better idea what's the last
thing that happens in each request.
> -----Urspr�ngliche Nachricht-----
> Von: Reto Badertscher [mailto:[EMAIL PROTECTED]]
> Gesendet: Donnerstag, 24. Januar 2002 09:34
> An: Tomcat Users List
> Betreff: AW: Tomcat - Request Handling
>
>
> Thanks for your help.
> I thought that a filter would be the best place to release resources?
> I'll check for deadlocks - that seems to be pretty hard.
>
> Reto
>
--
To unsubscribe: <mailto:[EMAIL PROTECTED]>
For additional commands: <mailto:[EMAIL PROTECTED]>
Troubles with the list: <mailto:[EMAIL PROTECTED]>
