[rules-users] Possibly race condition in a persisted Drools + JBPM during a StatefulKnowledgeSession

Mauricio Salatino salaboy at gmail.com
Tue May 29 11:59:50 EDT 2012 

Comments inline,

On Tue, May 29, 2012 at 12:24 PM, Alberto R. Galdo <argaldo at gmail.com>wrote:

>    Sure!. I didn't mention that we're running in a OSGi environment and
> we're getting the session from JPAKnowledgeService like that:
>
>
>
> EntityManagerFactory emf =
> Persistence.createEntityManagerFactory("com.package");
>
> Environment env = KnowledgeBaseFactory.newEnvironment();
>
> env.set(EnvironmentName.ENTITY_MANAGER_FACTORY, emf);
>
> env.set(EnvironmentName.TRANSACTION_MANAGER,(TransactionManager)
> bundleContext.getService(this.tmServiceRef));
>
> env.set(EnvironmentName.TRANSACTION, (UserTransaction)
> bundleContext.getService(this.userTransactionServiceRef));
>
> env.set(EnvironmentName.GLOBALS, new MapGlobalResolver());
>
> env.set(EnvironmentName.OBJECT_MARSHALLING_STRATEGIES, new
> ObjectMarshallingStrategy[] {
> MarshallerFactory.newSerializeMarshallingStrategy() });
>
> final StatefulKnowledgeSession ksession =
> JPAKnowledgeService.newStatefulKnowledgeSession(kbase, sessionConfig, env);
>
>
>    We are just following the manual ( and some acquired knowledge through
> the days ... ) .... I understand that inserting an event and starting a
> process at the same time would lead to the current situation, but Isn't
> this the reason why we are using drools with drools fusion + jbpm to build
> a CEP?
>
I'm not sure about your reasons :)

>
>    We chose Drools and a statefulsession in a fireUntilHalt loop just
> because we would be able to have our rules and processes reacting to events
> that enter the system at their own pace ... and found ourselves in this
> nightmare when trying to make the system fault-tolerant through
> persistence... :(
>
> FireUntilHalt and Persistence will cause "conflicts", because
fireUntilHalts by default creates a different thread == race conditions.


>    Do you think we should change our fault-tolerance strategy by
> reconstructing the knowledge session, process and tasks state by our own
> means leaving Drools and JBPM JPA persistence appart?
>
>  A common problem is to think that a session will do everything for you.
I'm not saying that you need to separate your processes from your rules,
I'm just saying that probably decoupling responsibilities will help you to
tackle down some of the problems that you are having. If you have in memory
processes and fusion entry points, you can keep those together and in a
separate persistence session handle all the human task interactions and
long running processes. If you are getting events in real time and you want
to process them, you cannot expect to persist all the session state and
have no performance impacts.

>
> Alberto R. Galdo
> argaldo at gmail.com
>
>
> Cheers

>
> On Tue, May 29, 2012 at 4:43 PM, Mauricio Salatino <salaboy at gmail.com>wrote:
>
>> Ok, so probably I'm not understanding your scenario, but are you using
>> the JPAKnowledgeService to create persistent sessions?
>> If you use that everything is aware of everything :) Drools and jBPM5
>> will run using the same persistence resources. For each interaction:
>> startProcess, insert, update, fireAllRules a transaction will be created.
>> If you are starting a process and at the same time inserting a Fusion event
>> you will be generating two transaction that will fight for the resources
>> (the same resources because you have only one session).
>> Obviously if you take the persistence mechanisms outside of the
>> discussion everything will work perfectly fine, because you don't have any
>> transactional resource to fight for :)
>>
>> Hope it helps.
>> Cheers
>>
>>
>> On Tue, May 29, 2012 at 11:36 AM, Alberto R. Galdo <argaldo at gmail.com>wrote:
>>
>>>    Your answer is confusing me, I think I'm not getting it right.
>>>
>>>    It seems that we you're saying that will have to modify JBPM or
>>> Drools behaviour to get our architecture done ... but I don't think we are
>>> doing nothing special than what Drools & JBPM are able to do in a
>>> non-persisting enviroment ( at least without JBPM persisted ).
>>>
>>>    Maybe deeping a bit more the architecture will lead to a better
>>> understanding of the problem:
>>>
>>>    We have a Drools StatefulKnowledgeSession wich is populated with
>>> packages of rules and different BPMN 2.0 process definitions. That session
>>> in Drools is persisted using JPA and our JTA provider. Our solution is a
>>> CEP that uses events that get into the session using Drools Fusion. There,
>>> several rules fire and then several consequences start processes wich
>>> contain both Tasks and HumanTasks. Those rule activations are already in a
>>> queue, the Agenda. Given that JBPM acts as the processmanager provider for
>>> Drools, our processes get started in JBPM, and it is persisted using it's
>>> own JPA manager.  Just rules, that start processes in response to events in
>>> a fireUntilHalt loop in a persistent enviroment, both for Drools & JBPM.
>>> Then some processes interact with the knowledge session using BussinessTask
>>> nodes( which make other rules fire and maybe launch a different set of
>>> processes ) and other simply get to the Stop node. I see nothing wrong
>>> here, in fact, taking the persistence appart from the equation, everything
>>> runs as beautifuly as expected. We've tested that.
>>>
>>>    I'm just trying to undersand the problem, I'm pretty 90% sure I'm
>>> wrong, but what I think is happening here is that Drools isn't aware that
>>> JBPM may be using the same resources and vice-versa, Isn't it?
>>>
>>>    Shouldn't the solution be to make Drools & JBPM aware of each other
>>> in what is related to the syncronization of the transaction management to
>>> avoid race conditions?
>>>
>>>
>>>
>>>
>>> Alberto R. Galdo
>>> argaldo at gmail.com
>>>
>>>
>>> On Tue, May 29, 2012 at 3:58 PM, Mauricio Salatino <salaboy at gmail.com>wrote:
>>>
>>>> I'm in no way saying this:
>>>> "  Let me see if I'm getting this right, What you mean is that there is
>>>> no way for the JPA persistence of Drools & JBPM to coexist in the same
>>>> knowledge session because of race conditions between them competing for
>>>> their shared resources? I'm afraid so.. :(
>>>>
>>>> They can coexist. As you can see they are working together. If you take
>>>> a look at the underlying layers you will see that each operation that you
>>>> run against the session is wrapped in a transaction. That leads to your
>>>> architecture, If you have a single entry point for your session everything
>>>> will work correctly. You can achieve this by receiving all the operations
>>>> in a queue and then execute one after the other.
>>>> If you have multiple threads interacting with the session, you can
>>>> implement a retrying mechanism if the transaction gets rolled back and as
>>>> soon as the transaction win the right to commit it will work. It really
>>>> depends on the problem that you are trying to solve.
>>>>
>>>> Cheers
>>>>
>>>> On Tue, May 29, 2012 at 10:42 AM, Alberto R. Galdo <argaldo at gmail.com>wrote:
>>>>
>>>>>    Thanks Mauricio for your kind response, and kick, as usual ;-)
>>>>>
>>>>>    Sure, we see lots of rolled back transactions ... Everywhere!
>>>>>
>>>>>    Let me see if I'm getting this right, What you mean is that there
>>>>> is no way for the JPA persistence of Drools & JBPM to coexist in the same
>>>>> knowledge session because of race conditions between them competing for
>>>>> their shared resources? I'm afraid so.. :(
>>>>>
>>>>>    If it is so, Are there any plans for the integration of a shared
>>>>> JPA persistence management of both Drools & JBPM, wich may be the solution
>>>>> for this problem? ... Will we be better thinking to change our architecture
>>>>> ASAP?
>>>>>
>>>>> Greets,
>>>>>
>>>>> Alberto R. Galdo
>>>>> argaldo at gmail.com
>>>>>
>>>>>
>>>>>
>>>>> On Tue, May 29, 2012 at 1:54 PM, Mauricio Salatino <salaboy at gmail.com>wrote:
>>>>>
>>>>>> Ok, so what is happening is that two or more threads are fighting for
>>>>>> the resources, causing that two or more transactions wants to be completed,
>>>>>> but just one can win. All the other must be retried. Are you seeing rolled
>>>>>> back exceptions in your stack trace?
>>>>>> My question to you is if you really need to handle everything in just
>>>>>> one big persistent session. I've solved similar issues in the past using
>>>>>> more than one session, for example one session to run business
>>>>>> processes(persistent) and one or more for receiving and reacting to events.
>>>>>> If you have an async way to communicate both sessions there should be no
>>>>>> problem.
>>>>>> Hope it helps!
>>>>>> Cheers
>>>>>>
>>>>>> On Tue, May 29, 2012 at 7:30 AM, Alberto R. Galdo <argaldo at gmail.com>wrote:
>>>>>>
>>>>>>> Hi all,
>>>>>>>
>>>>>>>    We've been for long time now developing a complex event
>>>>>>> processing system that ( simplified version ahead !! ) involves a set of
>>>>>>> rules that in turn activates a set processes that fulfill human tasks and
>>>>>>> other kind of tasks.. This all is running in a StatefulKnowledgeSession
>>>>>>> with JPA persistence configured both for Drools and JBPM. We are running an
>>>>>>> stack of Drools, JBPM, Drools Integration, Drools fussion, etc..
>>>>>>>
>>>>>>>    We've been able to persist Drools sessions & JBPM processes in
>>>>>>> the same Persistence Context  ( not without pain :( ) using different JTA
>>>>>>> implementations including ( but not limited to ) Bitronix & Atomikos.
>>>>>>>
>>>>>>>     What we are observing here is what seems to be some kind of race
>>>>>>> condition between Drools and JBPM when running the knowledge session when
>>>>>>> JPA&JTA persistence is configured. Very often, as soon as after 2-3
>>>>>>> processes get created as rule's consequences are fired in response of
>>>>>>> events inside the session we see how JBPM finds its instance nullified by
>>>>>>> Drools when it tries to end a process and persist it.
>>>>>>>
>>>>>>>    We've been able to find where Drools decides to delete an
>>>>>>> instance of the process ... at a given time Drools executes
>>>>>>> JPAProcessInstanceManager.clearProcessInstances() [1] when it finalizes a
>>>>>>> SingleSessionCommand wich in turn calls disconnect() for *all* the
>>>>>>> local-stored processinstances ( wich gets populated with instances of
>>>>>>> processes every time a process is started in the knowledge session ):
>>>>>>>
>>>>>>>     public void clearProcessInstances() {
>>>>>>>
>>>>>>>
>>>>>>>         for (ProcessInstance processInstance: new ArrayList<ProcessInstance>(processInstances.values())) {
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>             ((ProcessInstanceImpl) processInstance).disconnect();
>>>>>>>
>>>>>>>         }
>>>>>>>     }
>>>>>>>
>>>>>>>
>>>>>>>     So, Drools decides to disconnect all process instances in it's
>>>>>>> JPA context without taking in account the state the process is in, and when
>>>>>>> an processinstance that is not stopped gets removed then JBPM finds it's
>>>>>>> NullPointerException...
>>>>>>>
>>>>>>>     We've modified the code to make Drools aware of the state of the
>>>>>>> process before wiping it from the context   ( no problem here, there will
>>>>>>> be no leak as a running processinstance will be removed in future calls of
>>>>>>> "clearProcessInstances" given the process is closed ). But unfortunatelly
>>>>>>> this seems to resolve this problem, but lots of other problems ( wich seems
>>>>>>> also race conditions arise :  for instance: Drools closes connections to
>>>>>>> the database and JBPM finds the connection closed,
>>>>>>>
>>>>>>>
>>>>>>>     So, we are really worried about using Drools & JBPM in a
>>>>>>> persisted environment. Maybe our asumptions are wrong...  Is it possible to
>>>>>>> have an scenario like ours given the current Drools & JBPM integration
>>>>>>> status for a persistent statefulKnowledge Session?  Did anyone build a
>>>>>>> complex event processing system like ours in a unaltered persistence
>>>>>>> environment such as provided in Drools and JBPM by default?
>>>>>>>
>>>>>>>
>>>>>>> Greets,
>>>>>>>
>>>>>>> [1]
>>>>>>> https://github.com/droolsjbpm/jbpm/blob/master/jbpm-persistence-jpa/src/main/java/org/jbpm/persistence/processinstance/JPAProcessInstanceManager.java
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Alberto R. Galdo
>>>>>>> argaldo at gmail.co <argaldo at gmail.com>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> rules-users mailing list
>>>>>>> rules-users at lists.jboss.org
>>>>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>> --
>>>>>>  - MyJourney @ http://salaboy.wordpress.com
>>>>>>  - Co-Founder @ http://www.jugargentina.org
>>>>>>  - Co-Founder @ http://www.jbug.com.ar
>>>>>>
>>>>>>  - Salatino "Salaboy" Mauricio -
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> rules-users mailing list
>>>>>> rules-users at lists.jboss.org
>>>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> rules-users mailing list
>>>>> rules-users at lists.jboss.org
>>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>>  - MyJourney @ http://salaboy.wordpress.com
>>>>  - Co-Founder @ http://www.jugargentina.org
>>>>  - Co-Founder @ http://www.jbug.com.ar
>>>>
>>>>  - Salatino "Salaboy" Mauricio -
>>>>
>>>>
>>>> _______________________________________________
>>>> rules-users mailing list
>>>> rules-users at lists.jboss.org
>>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>>
>>>>
>>>
>>> _______________________________________________
>>> rules-users mailing list
>>> rules-users at lists.jboss.org
>>> https://lists.jboss.org/mailman/listinfo/rules-users
>>>
>>>
>>
>>
>> --
>>  - MyJourney @ http://salaboy.wordpress.com
>>  - Co-Founder @ http://www.jugargentina.org
>>  - Co-Founder @ http://www.jbug.com.ar
>>
>>  - Salatino "Salaboy" Mauricio -
>>
>>
>> _______________________________________________
>> rules-users mailing list
>> rules-users at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/rules-users
>>
>>
>
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>
>


-- 
 - MyJourney @ http://salaboy.wordpress.com
 - Co-Founder @ http://www.jugargentina.org
 - Co-Founder @ http://www.jbug.com.ar

 - Salatino "Salaboy" Mauricio -
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-users/attachments/20120529/df9682d0/attachment-0001.html

More information about the rules-users mailing list