[jboss-user] [JBoss Messaging] - Re: Opening a new JMS session for an already existing transa
The client is a webapp running on the server hosting JBM. We tried to add the aop libs inside the /lib of our server folder. Nothing better... View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4221188#4221188 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4221188 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
[jboss-user] [JBoss Messaging] - Re: Opening a new JMS session for an already existing transa
May you please elaborate on the classpath error? It seems that the state of the connectionState of the clientConnectionDelegate of the XAConnection is null. connectionState does not appear to be serializable. Does that mean that the same XAConnection cannot be shared between the nodes? If it is the case, how can we obtain an XARessource that accept the Xid we declared in the failed node? thx olivier View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4220999#4220999 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4220999 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
[jboss-user] [JBoss Messaging] - Re: Opening a new JMS session for an already existing transa
Hello, thx for your answer, we finally had time to try a version using XA. We tried to share in cluster the XAConnection so that when a node fails the transaction can continue on the other nodes using this shared connection: | XASession session = mySharedXAConnection.createXASession(); | XAResource resource = session.getXAResource(); | resource.start(mySharedXID, XAResource.TMJOIN); | Nevertheless, we got the following error when doing the mySharedXAConnection.createXASession(); | | 18:17:52,375 INFO [STDOUT] java.lang.NullPointerException | 18:17:52,375 INFO [STDOUT] at org.jboss.jms.client.container.FailoverValveInterceptor.invoke(FailoverValveInterceptor.java:87) | 18:17:52,375 INFO [STDOUT] at org.jboss.aop.advice.PerInstanceInterceptor.invoke(PerInstanceInterceptor.java:105) | 18:17:52,375 INFO [STDOUT] at org.jboss.jms.client.delegate.ClientConnectionDelegate$createSessionDelegate_6052335267724906805.invokeNext(ClientConnectionDelegate$createSessionDelegate_6052335267724906805.java) | 18:17:52,375 INFO [STDOUT] at org.jboss.jms.client.container.ClosedInterceptor.invoke(ClosedInterceptor.java:170) | 18:17:52,375 INFO [STDOUT] at org.jboss.aop.advice.PerInstanceInterceptor.invoke(PerInstanceInterceptor.java:105) | 18:17:52,375 INFO [STDOUT] at org.jboss.jms.client.delegate.ClientConnectionDelegate$createSessionDelegate_6052335267724906805.invokeNext(ClientConnectionDelegate$createSessionDelegate_6052335267724906805.java) | 18:17:52,375 INFO [STDOUT] at org.jboss.jms.client.delegate.ClientConnectionDelegate.createSessionDelegate(ClientConnectionDelegate.java) | 18:17:52,375 INFO [STDOUT] at org.jboss.jms.client.JBossConnection.createSessionInternal(JBossConnection.java:269) | 18:17:52,375 INFO [STDOUT] at org.jboss.jms.client.JBossConnection.createXASession(JBossConnection.java:198) | (We are using EAP 4.3.0) May someone help us to understand what's wrong with this method? If we create again a new XAConnection on the same JBM cluster, we do not get the same XARessource (i.e. our Xid is not recognized when doing the JOIN). Is it possible to get back an XARessource that accept the Xid we declared in the failed node? olivier View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4220711#4220711 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4220711 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
[jboss-user] [JBoss Messaging] - Re: Making a Filtering Bridge in a JBM Cluster
Thanks but in fact I need a hook to process the messages (like a transformation for instance) - sorry if the term filter does not suit the need ... View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4218147#4218147 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4218147 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
[jboss-user] [JBoss Messaging] - Re: Making a Filtering Bridge in a JBM Cluster
ok thanks a lot for the hint View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4218214#4218214 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4218214 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
[jboss-user] [JBoss Messaging] - Re: Making a Filtering Bridge in a JBM Cluster
OK, in fact I have some specific questions that should help me to begin : I could not find if the JBM bridges can process some filters on the forwarded messages. Is it possible? If not I will certainly do the bridge job with an MDB and XA. In all cases, I worry about the failover management : if a node fails during filtering+forward with bridge or mdb, all the XAresources holding the TX for the consumed message will also fail. Is the consumed message going to be stuck in the first queue (as in transaction)? (I am using the trial of EAP 4.3) Thanks View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4217879#4217879 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4217879 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
[jboss-user] [JBoss Messaging] - Making a Filtering Bridge in a JBM Cluster
Hello, I would like to forward messages on-the-fly from a local clustered queue A to another local clustered queue B after applying a set of filters. This process should occur on all the nodes of a JBM cluster. If filtering+forward does not complete, the message should be back on the queue A. If the node on which filtering+forward is processed fails, the message should be back on the queue A and being processed by another cluster node (failover should be fast - a 5mn transaction timeout will not fit) Any advice on how to do that without affecting performance and stability (hundreds of queues to process) ? Thank you for you help olivier View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4217372#4217372 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4217372 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
[jboss-user] [JBoss Messaging] - Opening a new JMS session for an already existing transactio
Hi, Is it possible to open a new JMS session for an existing transaction with JBM 1.4? I imagine that it is what happens when a connection generated by a ClusteredConnectionFactory manages the failover of a JBM cluster node, does it? Let me introduce the context of my question: We have a cluster of JBossMessaging instances. For technical limitations of our legacy system, we need to transmit some messages to our cluster via http. In this case, a client sends a message via http, a web application based on each node of our cluster retrieves the messages coming from http, transforms it in jms and send it to JBM. (JBM and the web app are on the same JBOSS instance and both are replicated on each node of our cluster). We want to manage transactions in http as we do in jms. In order to enable that feature, when a transaction is done via http, the web app stores all the messages of the transaction in Session. When the transaction finishes (commit), the stored messages are sent in a jms transaction on JBM and the stored messages are removed from the Session. The Session content of the web app is replicated in the cluster (sticky Session is used on the loadBalancer). When a node fails, all the messages of the transaction are still available: weâve done the failover. However, this solution is not safe since we cannot ensure (functionally) that the number of messages of the transactions will not overflow the cluster node memory. An alternative solution consists in keeping in the content of the replicated http-session a jms- session created for the transaction. Then the messages received in the transaction via http are sent on-the-fly to JBM (in jms). The jms-transaction is committed or rolled-back when one of these operations is received via http. With this solution, the memory problem is delegated to JBM. Nevertheless, it is no more possible to make the failover: the jms-session replicated in the http-session cannot be reused by other threads (and thus cluster nodes) than its creator. Since it is not possible to use the replicated jms-session object if the creator node fails, is it possible to open a new JMS session on a living cluster node for the already started transaction? Thanks for your help ! olivier View the original post : http://www.jboss.org/index.html?module=bbop=viewtopicp=4214961#4214961 Reply to the post : http://www.jboss.org/index.html?module=bbop=postingmode=replyp=4214961 ___ jboss-user mailing list jboss-user@lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-user
