The name issue wouldn't be quite so bad if we introduced some contract
for each subsystem to introduce a unique name for their pools, and maybe
another optional level of naming if they have more than one pool
"namespace" in a subsystem. Then a centralized view could be pretty
intuitive: "Thread Pools -> EJB -> ..."
On 10/20/2011 10:12 AM, Brian Stansberry wrote:
I struggled with that global view a bit over the last week.
I was thinking about approaching it as having it be a set of resources
in the threads subsystem that are basically just views onto the real
resource, which would exist in JCA, EJB etc.
That's not very satisfying. Leads to things like having to avoid
conflicts in pool names across subsystems. Plus it would be pretty hard
to implement.
I was noodling a bit this morning about separating the UI issue from the
core model a bit. The core model (in the threads subsystem) could just
provide resources addresses (and maybe a bit more metadata) about pools
configured elsewhere. The resources for pools would be consistent
throughout the model (e.g. a bounded-queue-thread-pool that supports
blocking would have the same attributes, ops, etc everywhere). The
console could use the info from the threads subsystem to discover all
the pool configs and then build a nice UI.
I suspect that's a problem for JON though. The CLI of course wouldn't be
able to produce a nice unified view, but IMHO that's not a critical problem.
On 10/20/11 8:16 AM, Jason Greene wrote:
> Right. What happened is we found out that having thread pools be global
> promotes sharing, which is usually an error since each subsystem needs
> specific configurations. We were planning on adding a global view that
> would see all the subsystems' thread pools so the administrator can
> easily identify which were in use however I don't think it was completed
> yet. Even though we neat them in the subsystem we still want them to be
> consistent.
>
> Sent from my iPhone
>
> On Oct 20, 2011, at 5:58 AM, Heiko Braun<hbraun(a)redhat.com
> <mailto:hbraun@redhat.com>> wrote:
>
>>
>>
>> no, but if possible they should rely on the default schema (xsd) for
>> defining inlined thread pool configurations.
>>
>> On Oct 20, 2011, at 12:50 PM, David Bosschaert wrote:
>>
>>> It seems that the latter defines pools that are used in
>>> /subsystem=ejb3/service=*
>>> Should they not be using thread pools defined in the threads subsystem
>>> instead?
>>
>> _______________________________________________
>> jboss-as7-dev mailing list
>> jboss-as7-dev@lists.jboss.org<mailto:jboss-as7-dev@lists.jboss.org>
>>
https://lists.jboss.org/mailman/listinfo/jboss-as7-dev
>
>
> _______________________________________________
> jboss-as7-dev mailing list
> jboss-as7-dev(a)lists.jboss.org
>
https://lists.jboss.org/mailman/listinfo/jboss-as7-dev