Thanks for the feedback - I've merged that in :)

On Sun, Jul 20, 2025 at 7:06 PM Richard Zowalla <r...@apache.org> wrote:

> Hi,
>
> I agree with Mark + Markus. Proposal lgtm.
>
> Gruß
> Richard
>
> > Am 17.07.2025 um 14:35 schrieb Jonathan Gallimore <
> jonathan.gallim...@gmail.com>:
> >
> > Hi All
> >
> > Looking at the definition of a ManagedExecutorService, such as below, it
> > looks like there isn't a way to force the service to have no queue, and
> > always use more threads, up to the max threads.
> >
> > If you consider this definition:
> >
> >    <Resource
> >        id="MyManagedExecutor"
> >        type="javax.enterprise.concurrent.ManagedExecutorService">
> >        Core=5
> >        Max=150
> >        KeepAlive=60s
> >        Queue=0
> >        WaitAtShutdown=5s
> >    </Resource>
> >
> > Per the logic here:
> >
> https://github.com/apache/tomee/blob/758e64e7fb69862c94233669531902b65bc1ab80/container/openejb-core/src/main/java/org/apache/openejb/resource/thread/ManagedExecutorServiceImplFactory.java#L77-L81
> > ,
> >
> > If size is specified (>0), anything executed will be added to the queue,
> > and the threadpool will only increase from the core size to the max size
> > when the queue is full.
> >
> > If size is 0 or negative, the same behaviour applies, but the queue is
> > unbounded (so the thread pool will not grow at all).
> >
> > There's no way to specify a zero sized queue (and always grow the thread
> > pool up to max; and reject work if all threads are busy and the pools is
> at
> > the max size).
> >
> > What are peoples thoughts on a change like this one?
> > https://github.com/apache/tomee/pull/2011
> >
> > This does change the behaviour, and I'm not necessarily a fan of "magic"
> > numbers - an additional boolean setting to control the behaviour might be
> > an alternative?
> >
> > What do you think?
> >
> > Jon
>
>

Reply via email to