Re: [rules-users] Drools Flow 5.1.0.M1 - Pooling of WSHumanTaskHandler and MinaTaskClient - IoSession
Yes, definitely making sense.
In your case, it seems like the task completion event (registered using
an event listener) should not necessarily be handled by the session that
created the task, but that it would be useful to have a pool of session
for handling user requests and one for handling system events (like
timers or task completion events). We are hoping to provide this kind
of flexibility (decoupling the event registration) when providing that
more generic communication between the engine and the task service.
But if it makes sense for now to share the same physical connection, you
can definitely tweak the WSHumanTaskHandler implementation. These work
item handlers are simply the binding between the engine and the
underlying service and easy to change.
Kris
Quoting Vijay K Pandey <VPandey(a)mdes.ms.gov>:
Thanks Kris. I did saw the connect method and saw that it only
creates a new connection if the client is null, but my question was
more at the level of "session" (StatefulKnowledgeSession) and
MinaTaskClient creation.
StatefulKnowledgeSession ksession =
JPAKnowledgeService.newStatefulKnowledgeSession( kbase, null, env );
ksession.getWorkItemManager().registerWorkItemHandler("Human
Task", new WSHumanTaskHandler());
So with each StatefulKnowledgeSession - we will have a new "connect".
I think pooling the WSHumanTaskHandler would not make any sense bcoz
the executeWorkItem method registers the MinaTaskClient with the
MinaTaskServer for several events.
Now another main component for us is the MinaTaskClient where we will
be performing lots of operation where we won't register for any
events. In that case I think we can pool the MinaTaskClient - may be
even 1 MinaTaskClient can work for all the non registered client
operations - as MINA IoSession is thread safe, need to make sure the
IoFilter to be thread safe - but to be on better side we can pool
MinaTaskClient.
I hope I am making some sense? Do let me know your recommendation.
Thanks
Vijay
-----Original Message-----
From: rules-users-bounces(a)lists.jboss.org
[mailto:rules-users-bounces@lists.jboss.org] On Behalf Of Kris
Verlaenen
Sent: Sunday, November 15, 2009 1:47 PM
To: Rules Users List; Vijay K Pandey
Subject: Re: [rules-users] Drools Flow 5.1.0.M1 - Pooling of
WSHumanTaskHandler and MinaTaskClient - IoSession
One WSHumanTaskHandler should cache the client, so that it does not
recreate a new connection every time. Note that, in the connect()
method, you only create a new connection if the client is null.
Kris
Quoting Vijay K Pandey <VPandey(a)mdes.ms.gov>:
> Hi,
>
> We have a web based application that currently deals with MQ
Workflow
> for all the WSHT related processes/tasks. We are in the process of
> migrating to Drools Flow. In our current scenario we pool MQ
workflow
> session (through apache commons pooling) - and our hundreds of web
> users use these sessions to get their word done (like starting a
new
> process, retrieving a task, complete/forwarding/terminating tasks
> etc) - what we have seen is that in an 8hr window we have around
> 20,000 calls to mq workflow (around 42 calls/minute) - and it
scales
> pretty good with mq workflow session pooling in place.
>
> Should I go ahead and extend WSHumanTaskHandler and MinaTaskClient
so
> that they don't go through "connect" on every operation and get
the
> MINA IoSession from some pool - as it might consume quite a bit of
> resources/time consuming to have a separate physical connections
for
> each operation.
>
> Looking forward for your recommendations.
>
> Thanks
> Vijay
>
Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm