Hi all, Try to execute many threads with the next source code (simple simple code). You will see that only one CPU-thread is used in a mutithreaded CPU system. It seems to be a system lock inside newStatefulKnowledgeSession(). This is a big disappoint because we can't use StatefulKnowledgeSession in multithreaded environment. All is executing sequentially because of this problem. Yes, we can asure newStatefulKnowledgeSession() is multithread (no error is returned) but we can't asure newStatefulKnowledgeSession() executes in parallel. Do you know any problem about this? Thank's in advance PROBLEMATIC SOURCE CODE: public class KnowledgeSessionThread extends Thread { private static final int ITERATIONS = 1000000; private KnowledgeBase kbase; public KnowledgeSessionThread() throws Exception { kbase = readKnowledgeBase(); } public void run() { try { for( int i = 0; i < ITERATIONS; i++ ) { StatefulKnowledgeSession session = kbase.newStatefulKnowledgeSession(); session.dispose(); } System.out.println("Thread finished"); } catch( Exception e ) { e.printStackTrace(); } } private KnowledgeBase readKnowledgeBase() throws Exception { KnowledgeBuilder kbuilder = KnowledgeBuilderFactory.newKnowledgeBuilder(); kbuilder.add(ResourceFactory.newClassPathResource("Sample.drl"), ResourceType.DRL); kbuilder.add(ResourceFactory.newClassPathResource("Sample.bpmn"), ResourceType.BPMN2); KnowledgeBuilderErrors errors = kbuilder.getErrors(); if (errors.size() > 0) { for (KnowledgeBuilderError error: errors) { System.err.println(error); } throw new IllegalArgumentException("Could not parse knowledge."); } KnowledgeBase kbase = KnowledgeBaseFactory.newKnowledgeBase(); kbase.addKnowledgePackages(kbuilder.getKnowledgePackages()); return kbase; } } -- View this message in context: http://drools.46999.n3.nabble.com/StatefulKnowledgeSession-construction-e... Sent from the Drools: User forum mailing list archive at Nabble.com. _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

-----Original Message----- From: rules-users-bounces(a)lists.jboss.org [mailto:rules-users- bounces(a)lists.jboss.org] On Behalf Of Juan Carlos Fernández Sent: 11 October 2011 11:49 To: rules-users(a)lists.jboss.org Subject: Re: [rules-users] StatefulKnowledgeSession construction executes sequentially (no multithread) We create one StatefulKnowledgeSession for every client (if 20 clients at same time, 20 StatefulKnowledgeSession are created) because every client has different environment for rule evaluation. Facts for one client does not have to collide with facts of other clients so this is the reason why we create multiple sessions. Let me explain all this in greater detail. When a client executes a SOAP operation, one StatefulKnowledgeSession is created and associated to the transaction. Every log generated by the transaction is inserted as a Fact. When transaction finishes, we fire all rules for the current transaction knowledge session in order to correlate all transaction logs. This transaction logs do not have to collide with other transaction logs so every thread (transaction) has its own StatefulKnowledgeSession. Our app locks and begin to use only one CPU thread because of newStatefulKnowledgeSession(). Do you think entry-points would be the solution? Is there any other solution? One solution is to use StatefulKnowledgeSession pool who creates instances but we have to retract all facts when returning session to pool. Concurrent executions and throughput is a key feature of our SOAP application. Thank you very much -- View this message in context: http://drools.46999.n3.nabble.com/StatefulKnowledgeSession-construction- executes-sequentially-no-multithread-tp3412107p3412222.html Sent from the Drools: User forum mailing list archive at Nabble.com. _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

Here you have my pooling solution. Testing this source code reaches 100% CPU usage!! Used apache commons pooling. borrowObject() creates stateful session and pool it. returnObject() retracts all facts. Can we extract a conclusion with my tests? I'm going to implement pooling solution inside my SOAP app and do performance tests (throughput). public class KnowledgeSessionThread extends Thread { int iterations = 500; int numberFacts = 1000; private ObjectPool pool; public KnowledgeSessionThread(ObjectPool pool) throws Exception { this.pool = pool; } public void run() { try { for( int i = 0; i < iterations; i++ ) { StatefulKnowledgeSession session = (StatefulKnowledgeSession)pool.borrowObject(); for( int j = 0; j < numberFacts; j++ ) { Message msg = new Message(); msg.setMessage("Hello world"); msg.setStatus(Message.HELLO); session.insert(msg); } session.fireAllRules(); pool.returnObject(session); } System.out.println("Thread finished " + this.currentThread().getName()); } catch( Exception e ) { e.printStackTrace(); } } } @Override public Object makeObject() throws Exception { System.out.println(Thread.currentThread().getName() + " - makeObject() called" ); return kbase.newStatefulKnowledgeSession(); } @Override public void destroyObject( Object obj ) { StatefulKnowledgeSession session = (StatefulKnowledgeSession)obj; session.dispose(); } @Override public void passivateObject(Object obj) { StatefulKnowledgeSession session = (StatefulKnowledgeSession)obj; for( FactHandle handle : session.getFactHandles() ) { session.retract(handle); } } -- View this message in context: http://drools.46999.n3.nabble.com/StatefulKnowledgeSession-construction-e... Sent from the Drools: User forum mailing list archive at Nabble.com. _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users