JMS Cache Synchronization - Merge or Database Read?

When an object in one JVM is changed does TopLink' JMS cache synchronization mechanism merge the changes into the other JVMS in a cluster or does it just signal that the object is dirty and the object is re-read from the database?
If the answer is 'merge' does a post merge event get called on the object being merged?
thanks
Steve

In TopLink 10.1.3 both can happen, the changes can be merged or an object can be marked invalid and TopLink will refresh the object as needed. In 9.0.4.X the changes will be merged into the other caches (if an object of correct version exists in that cache).
In the case of a merge postMerge() event will be called.
--Gordon

Similar Messages

10.1.3 JMS Cache Synchronization on OC4J

Hi,
We're using stand-alone OC4j 10.1.3.0 along with TopLink 10.1.3.0 (packaged with OC4j). In enabling JMS cache synchronization, I get the following error message:
Exception [TOPLINK-12000] (Oracle TopLink - 10g Release 3 (10.1.3.0.0) (Build 060118)): oracle.toplink.exceptions.CommunicationException
Exception Description: Error Sending connection service to jms/Topic/ToplinkCacheSync.
Internal Exception: javax.jms.JMSException: TopicSubscriber[Oc4jJMS.Consumer.arkansas.6e2ec2b0:1117159d1ca:-7fff.65,Topic[ToplinkCacheSync],null,null,false]: cannot invoke method "setMessageListener" within the J2EE container.
at oracle.toplink.exceptions.CommunicationException.errorSendingConnectionService(CommunicationException.java:40)
at oracle.toplink.remote.CacheSynchronizationManager.addRemoteConnection(CacheSynchronizationManager.java:175)
at oracle.toplink.remote.jms.JMSClusteringService.retreiveRemoteSessions(JMSClusteringService.java:198)
at oracle.toplink.remote.jms.JMSClusteringService.run(JMSClusteringService.java:175)
Caused by: javax.jms.JMSException: TopicSubscriber[Oc4jJMS.Consumer.arkansas.6e2ec2b0:1117159d1ca:-7fff.65,Topic[ToplinkCacheSync],null,null,false]: cannot invoke method "setMessageListener" within the J2EE container.
at com.evermind.server.jms.JMSUtils.make(JMSUtils.java:1034)
at com.evermind.server.jms.JMSUtils.toJMSException(JMSUtils.java:1114)
at com.evermind.server.jms.JMSUtils.toJMSException(JMSUtils.java:1085)
at com.evermind.server.jms.JMSUtils.assertNotContainer(JMSUtils.java:1500)
at com.evermind.server.jms.EvermindMessageConsumer.setMessageListener(EvermindMessageConsumer.java:217)
at oracle.toplink.remote.jms.JMSClusteringService.connectBackToRemote(JMSClusteringService.java:53)
at oracle.toplink.remote.CacheSynchronizationManager.addRemoteConnection(CacheSynchronizationManager.java:168)
... 2 more
It is, therefore, my assumption that Toplink is using JMS 1.0.2b while OC4J is using JMS 1.1. Who can:
1) Verify or debunk my assumption?
2) Point me to a work-around?
Perhaps my config is incorrect? Here's a sessions.xml snippet, in case it helps:
<cache-synchronization-manager>
<clustering-service xsi:type="jms-clustering">
<naming-service-url>ormi://localhost:23791/ems</naming-service-url>
<jndi-user-name>oc4jadmin</jndi-user-name>
<jndi-password>oc4j</jndi-password>
<naming-service-initial-context-factory-name>com.evermind.server.rmi.RMIInitialContextFactory</naming-service-initial-context-factory-name>
<jms-topic-connection-factory-name>jms/Topic/ToplinkCacheSyncConnectionFactory</jms-topic-connection-factory-name>
<jms-topic-name>jms/Topic/ToplinkCacheSync</jms-topic-name>
</clustering-service>
<remove-connection-on-error>false</remove-connection-on-error>
</cache-synchronization-manager>
Many thanks in advance,
--Todd

Thank you very much for your contribution! Yes, indeed, that was my main issue. A sessions.xml section that looks like the following (instead of the <cache-synchronization-manager> section above) does fix the problem:
<remote-command>
<commands>
<cache-sync>true</cache-sync>
</commands>
<transport xsi:type="jms-topic-transport">
<topic-host-url>ormi://localhost:23791/ems</topic-host-url>
<topic-connection-factory-name>jms/Topic/ToplinkCacheSyncConnectionFactory</topic-connection-factory-name>
<topic-name>jms/Topic/ToplinkCacheSync</topic-name>
<jndi-naming-service>
<user-name>oc4jadmin</user-name>
<password>oc4j</password>
<initial-context-factory-name>com.evermind.server.rmi.RMIInitialContextFactory</initial-context-factory-name>
</jndi-naming-service>
</transport>
</remote-command>
Thanks again,
--Todd

JMS Cache Synchronization Manager

Hi all,
I am trying to implement distributed cache synchronization in TopLink using JMS.
On http://download-west.oracle.com/docs/cd/A97688_12/toplink.903/b10064/enterpri.htm#1022254 , there is an example given which has a snippet of the sessions.xml file.
The example snippet shows a tag <naming-service-initial-context-factory> under the <cache-synchronization-manager> tag. But on the DTD given on http://download-west.oracle.com/docs/cd/A97688_12/toplink.903/b10064/a-sessio.htm#634251 , I cant find an element <naming-service-initial-context-factory>. Please let me know which one is wrong - the DTD or the example?
Also, I tried to use JMS for distributed cache synchronization. My JMS Server is running on a seperate VM and I have configured info about it in the sessions.xml file. But TopLink is complaining that it cannot find the top and connection factory. Yet, I am able to look them up from a standalone program.
Does anyone have any experience setting this up? Any examples?
Thanks,
Binil

Hi Doug,
Thanks for the clarifications! :)
I am not able to get the JMS Cache Synchronization work still :( I am new to JMS & Toplink - so I will try to explain what I am doing and please let me know if there is something I am doing wrong.
I have an EJB application which is a Stateless Session bean which has two methods : one to fetch an entity and another to update an entity. Also, I have a weblogic domain in which I have three managed servers - appServerOne, appServerTwo and jmsServer. On appServerOne and appServerTwo I have deployed the above mentioined application. On jmsServer, I have configured my ConnectionFactory and Topic and the JNDI names of those are provided to the EJB application.
With this setup, I run two client programs. I will list the steps below:
1. Client1 fetches an entity whereby it is loaded to the TopLink cache. The DB logs show that a SELECT query is issued to the DB.
2. Client2 fetches the same entity whereby it is loaded to the cache; again a SELECT query is issued.
3. Client1 updates that entity using UOW; the logs dont show a DB UPDATE statement though. But the updates are indeed made to the DB when I check them.
4. Client1 fetches the entity again; the updates are visible and the JDBC logs show that a trip to the DB is not done.
5. After waiting for sometime Client1 fetches the same entity again and I get the stale data! :(
I have enabled logging, so I am hoping that TopLink will log a message when a JMS notification is posted. Also, I have made the cache synchronization synchronous, so that the transaction shouldnt commit before the notification being send.
I tried making the stateless session bean stateful and it still didnt work. Is there anything I need to do.
Cheerio,
Binil

JMS & Cache synchronization

We have setup Toplink to do cache synchronization using Tibco JMS. We are using Tomcat Servlet environment. We are using the pre-login event in Toplink to specify the necessary information required for cache synchronization using JMS. For some reason Toplink connects to JMS using the name "anonymous" in addition to connecting with the user name we specify in the code. Toplink also seems to connect to some queue in addition to the Topic that is mentioned in the code. Does anyone know why Toplink would behave like this?

Hello,
TopLink uses JNDI to get the TopicConnectionFactory, and will login using the username/password you have specified. It then calls createTopicConnection() that will use the default userid to get a connection to a Topic, so this could be what you are referring to. The login specified in the cache synch setup is only to access JNDI.
That said, TopLink does not use a queue that I have ever seen. The source code for the JMSClusteringService is shipped with TopLink and can be used to fill in parts left out of the documentation, as well as allow for customizations. In this case, it would be good to review it to get an understanding of the calls TopLink does to connect. If this does not help, I'd need to see the configuration used to set up cache synchronization.
Best Regards,
Chris Delahunt

JMS Exception during cache synchronization

Hi,
I am using JMS cache synchronization fro two applications . During cache synchronization I got this exception
Local Exception Stack:
Exception [TOPLINK-0] (OracleAS TopLink - 10g (9.0.4.5) (Build 040930)): oracle.toplink.exceptions.CommunicationException
Exception Description: javax.jms.JMSException: enqMessage at oracle.toplink.remote.jms.JMSConnection.processCommand(JMSConnection.java:54) at oracle.toplink.internal.remote.ChangeSetPropagator$1.run(ChangeSetPropagator.java:64)
Has this something to do with the message size ?
Appreciate any help.
Thanks
Priya

would not expect TopLink to connect twice at all. Is it possible that another service from within the ApplicationServer is connecting? I would recommend changing the JNDI location or the service port of the TIBCO JMS service and altering the TopLink JMS Clustering Service URL to reflect this change. Then verify that multiple connections are still being acquired.
--Gordon

UnitOfWork.refreshObject(); don't fire event for Cache Synchronization

My WebApp use JMS for cache synchronization.
and I need to refresh session Cache by using method unitOfWork.refreshObject() but it don't publish any message to JMSserver.
sinse My JMS cache synchronization work well when I do UnitOfWok transaction(insert,update,delete)
How to force cache synchronization ?
thank you,
Kowit Laison

TopLink cache synchronization only synchronizes changes made through a unit of work, it does not synchronize reads or refreshes.
You will need to perform the refresh on each server, or make your own JMS message to notify your other server's to refresh.
You may also wish to investigate the TopLink 10.1.3 Preview release that adds support for Cache Invalidation and Cache Invalidation Synchronization. You may be able to broadcast an invalidation command for the refresh.

Cache Synchronization with JMS and a Message Driven Bean (MDB)

If I understand correctly, Toplink ueses the following configuration info to find the target to publish its session messages to a JMS server (having its URL as below) which provides the JNDI service that Toplink needs to look up the topic:
<cache-synchronization-manager>
<clustering-service> oracle.toplink.remote.jms.mdb.JMSPublishingOnlyClusteringService
</clustering-service>
<should-remove-connection-on-error>false</should-remove-connection-on-error>


<jms-topic-connection-factory-name> jms/TopLinkTopicConectionFactory
</jms-topic-connection-factory-name>
<jms-topic-name>jms/TopLinkCacheSynchTopic</jms-topic-name>


<naming-service-url>ormi://localhost:23791</naming-service-url>
<naming-service-initial-context-factory> com.evermind.server.rmi.RMIInitialContextFactory </naming-service-initial-context-factory></cache-synchronization-manager>
The MDB also uses the same JNDI service provided by the JMS server to listen to the topic which is named in the ejb-jar.xml file:
<enterprise-beans>
<message-driven>
<display-name>TopLink clustering MDB</display-name>
<ejb-name>TopLinkClusteringMDB</ejb-name>
<ejb-class>com.mycompany.mdb.ClusteringMDB</ejb-class>
<transaction-type>Container</transaction-type>
<message-driven-destination> <destination-type>javax.jms.Topic</destination-type>
</message-driven-destination>
<env-entry>
<description>TopLink session name</description>
<env-entry-name>tl_session_name_for_mdb</env-entry-name>
<env-entry-type>java.lang.String</env-entry-type>
<env-entry-value>Employee_Session</env-entry-value>
</env-entry>
<resource-ref> <description>description</description>
<res-ref-name>jms/TopLinkTopicConectionFactory</res-ref-name>
<res-type>javax.jms.TopicConnectionFactory</res-type>
<res-auth>Application</res-auth>
<res-sharing-scope>Shareable</res-sharing-scope>
</resource-ref>
<resource-env-ref>
<resource-env-ref-name> jms/TopLinkCacheSynchTopic</resource-env-ref-name>
<resource-env-ref-type>javax.jms.Topic</resource-env-ref-type> </resource-env-ref>
</message-driven></enterprise-beans>
To do this, the EJB container that hosts the above MDB needs to be configured to know where the JMS server is by using the above URL same as in the Toplink sessions.xml file.
I am not clear that who provides the JNDI service for the Toplink session name lookup used by the MDB:
// Obtain the Session name configured in the environment variable
String sessionName = (String) myEnv.lookup(TOPLINK_SESSION_NAME);
session = SessionManager.getManager().getSession(sessionName);
so that the MDB can find the currently running Toplink session and communicate with it.
Can anybody help answer my above question and correct my understandings?
Many thanks!

Although TopLink is running in the application serverâs VM, the only notable integration points with the server are for JTA and user-defined data sources. All of the work performed to get a session via SessionManager occurs solely within TopLink. When the call is made to retrieve a session, you arenât getting the âactiveâ session identified by âEmployee_Sessionâ that is registered with the application server, rather a new session is created using the definition that exists for âEmployee_Sessionâ in sessions.xml.
When the call âsession = SessionManager.getManager().getSession(âEmployee_Sessionâ)â is made, the SessionManager simply looks for sessions.xml at the root level of its deployed jars or classpath, then looks for the âEmployee_Sessionâ tag within sessions.xml. If a session exists in the file with that name then a TopLink session object is created and returned.
Note that setting an environment entry as mentioned above simply allows for the session name to be declared dynamically. The call to âmyEnv.lookup(âtl_session_name_for_mdbâ)â simply asks the beanâs environment for the <env-entry-value> that matches the <env-entry-name>. That string is then used in SessionManager (no different than a hard coded value would be) to identify the session in sessions.xml that is to be loaded.
I hope this makes sense.

Distributed Cache synchronization with JMS

Hello,
The documentation on Cache Synchronization using JMS lists steps to setting up JMS in the JMSClusteringService but does not list steps as to how to connect or setup connections to the various different servers inorder to synchronize between them. It lists steps for connecting to servers when using RMI but not JMS. Appreciate any additional documentation on doing this in JMS. Many thanks

JMS, by its nature, provides the connections between the different 'servers'. If all servers connect to the same JMS service then there are no additional actions required to have the 'servers' synchronize.
--Gordon

Setting up cache synchronization using JMS on Tomcat

I'm trying to enable cache synchronization using JMS on Tomcat 5.0; Does anyone have any tips on how to do this?

Configure the TIBCO service to provide a Topic and a TopicConnectionFactory for TopLink to use.
You will need to register for a pre-login event configured on the session that will configure the JMSClusteringService for the TIBCO service.
ie.
JMSClusteringService clusteringService = new JMSClusteringService(event.getSession());
event.getSession().setClusteringService(clusteringService);
Hashtable env = new Hashtable();
env.put(javax.naming.Context.INITIAL_CONTEXT_FACTORY , "TIBCOFactoryClassName");
env.put(javax.naming.Context.PROVIDER_URL, "url");
env.put(javax.naming.Context.SECURITY_PRINCIPAL, "userName"));
env.put(javax.naming.Context.SECURITY_CREDENTIALS,"password");
clusteringService.setInitialContextProperties(env);
--Gordon

JMS and Data Cache Synchronization

We have setup Toplink to do cache synchronization using Tibco JMS. For some reason Toplink connects to JMS using the name "anonymous" in addition to connecting with the user name we specify in the code. Toplink also seems to connect to some queue in addition to the Topic that is mentioned in the code. Does anyone know why Toplink would behave like this?

would not expect TopLink to connect twice at all. Is it possible that another service from within the ApplicationServer is connecting? I would recommend changing the JNDI location or the service port of the TIBCO JMS service and altering the TopLink JMS Clustering Service URL to reflect this change. Then verify that multiple connections are still being acquired.
--Gordon

JMS Cache updates in existing transactions

We have two application instances running, with synchronization over JMS. In the following scenario what will be the outome?
- AppInstance1 does changes to Object1 and commits. These changes are sent to the JMS queue.
- AppInstance2 is in the middle of one transaction when it recieves the cache updates. The current transaction include Object1 (which updates are received for).
Does the transaction in Instance2 get aborted upon receiving the cache sync or are the mechanisms at work in TopLink smart enough to merge the non conflicting fields?
Thanks,
Anders,

This depends on your locking mechanism and has little to do with cache-synch. Cache synchronization does not replace the requirement for a locking mechanism.
If you are using no locking, then no matter if you have a single server, multiple servers with or without cache synch, then there is the potential that concurrent transactions can overwrite each others data.
If you are using optimistic locking, then no matter if you have a single server, multiple servers with or without cache synch, your second transaction will get an optimistic lock error and its transaction will be rolled-back.
If you are using pessimistic locking, then there is not much point in using cache synch as you will be accessing the database anyway. The second transaction will block on its read of the object until the first transaction completes, or throw an error on read if using no_wait.
The benefit that cache synchronization provides it that because changes are synchronized between servers the likely-hood of getting an optimistic lock exception is greatly reduced. It can also be used with infrequently updated objects that do not use any locking, or object to which locking is not an issue.

Cache Synchronization Exception Handling

I am setting up cache synchronization on WebLogic using JMS. I am looking into writing an exception handler to handle cache synchronization exceptions. The following statement in the documentation caught my attention:
As mentioned above, the TopLink cache does not begin the merge or update process until the database transaction has already been committed. This is quite beneficial in that it avoids letting uncommitted data into the shared cache, but should be recognized where transactional synchronization is considered. In cases where a merge may have failed there is no way to roll back the changes made to the database (although it is questionable whether this would be a good idea in any case). As a consequence, failures during remote merging can leave the cache in an inconsistent state. This makes it important to handle any errors that occur by performing cache normalization actions, such as resetting the cache, or even the server.
Suppose we have two app server instances, A and B. Instance A commits a change and sends the update notification to instance B. The merge fails on instance B with an optimistic lock exception.
- I assume that a CacheSynchronizationException will be thrown. Will it be thrown on instance A, B, or both?
- At this point which cache is inconsistent; A, B, or both?
- The documentation suggests resetting the cache, but this seems a rather severe way of dealing with the problem. Is it feasible / effective to just refresh the objects in the change set of the CacheSynchronizationException?
This is a simple exception handler. On receiving a CacheSynchronizationException it resets the cache and re-throws the exception.
class ToplinkExceptionHandler implements ExceptionHandler{
     public Object handleException(RuntimeException exception) throws DatabaseException{
          if(exception instanceof CacheSynchronizationException){){
               myServerSession.initializeAllIdentityMaps();
               throw exception;
          }else{
               throw exception;
- Is this approach effective?
- Once the cache is reset should the exception be re-thrown or should I consider it handled and swallow it? If I do swallow it what should the method return?
Any advice would be appreciated.

I had to fix a few errors (in our code) before getting it working. Since I have multiple sessions, I had to specify different multicast ports (actually I used a different multicast address itself) for each of the sessions. I was not setting the announcement delay on one of the sesssions and that was why it was announcing immediately upon coming up. Also I had to explicitly set asynchronous to false because it is true by default.
BTW, we are not using CMP (entity beans) rather Java Objects. Our deployment has two Apache/Tomcat machines and two WebLogic machines. The WebLogic servers are in a round-robin cluster. The cache synchronization seems to be working fine for this configuration.
Thanks.
Anand R

Error handling for distributed cache synchronization

Hello,
Can somebody explain to me how the error handling works for the distributed cache synchronization ?
Say I have four nodes of a weblogic cluster and 4 different sessions on each one of those nodes.
On Node A an update happens on object B. This update is going to be propogated to all the other nodes B, C, D. But for some reason the connection between node A and node B is lost.
In the following xml
<cache-synchronization-manager>
<clustering-service>...</clustering-service>
<should-remove-connection-on-error>true</should-remove-connection-on-error>
If I set this to true does this mean that the Toplink will stop sending updates from node A to node B ? I presume all of this is transparent. In order to handle any errors I do not have to write any code to capture this kind of error .
Is that correct ?
Aswin.

This "should-remove-connection-on-error" option mainly applies to RMI or RMI_IIOP cache synchronization. If you use JMS for cache synchronization, then connectivity and error handling is provided by the JMS service.
For RMI, when this is set to true (which is the default) if a communication exception occurs in sending the cache synchronization to a server, that server will be removed and no longer synchronized with. The assumption is that the server has gone down, and when it comes back up it will rejoin the cluster and reconnect to this server and resume synchronization. Since it will have an empty cache when it starts back up, it will not have missed anything.
You do not have to perform any error handling, however if you wish to handle cache synchronization errors you can use a TopLink Session ExceptionHandler. Any cache synchronization errors will be sent to the session's exception handler and allow it to handle the error or be notified of the error. Any errors will also be logged to the TopLink session's log.

More Control of Cache Synchronization in a Cluster?

If I understand correctly, the cache synchronization manager, by default, sends out a changeSet message to the network for any UOW write commit. Since I have a control on certain objects not using Toplinkâs cache, I actually do not need the cache synchronization manger to broadcast any update on these objects. In other words, I do not want the cache synchronization manger to send out the changeSet message to the network if I only update those objects that are configured to always read from the database. In this way, I may be able to avoid some unnecessary changeSet message propagations.
My question is that whether I have a way, either in sessions.xml or in Java API, to tell the cache synchronization manger not to send out the changeSet message for a given specific UOW write commit (although I still need the data update to be committed to the database).

Many thanks for the reply and glad to see something better is on the way again!
If I choose to set the cache synchronization manager to null, do the UOW write for a given class which is always refreshed from the database, and then reset the cache synchronization manger back to the previous instance, do I need to do the following afterwards:
          // must login/re-login to initialize cache synchronization
          if (session.isConnected()) {
               session.logout();
          session.login();
to ensure the manager be reset properly?
Since the propagation of the changeSet message is triggered by the UOW, can I have more control at the UOW level to turn on/off the cache synchronization manager?
If I understand correctly, in Toplink 10.1.3, I can configure the cache synchronization manger globally at the session level, and also have the option in the same application to set a given number of classes not be cache synchronized in a cluster at the descriptor level to overrule the session global settings. I should not have to choose one over the other. Once these are done in configurations, what is the reason that I still have to choose either to send the changeSet or to invalidate the message?

JMS Cache Sync Problems

All,
We're running into some strange problems synchronizing Toplink caches via JMS.
In our environment, we have two different applications running under the same app server - different code in general, but they use the same Toplink mapping and mapped classes. Each of these connects to the database independently of the other using Toplink. We've configured sessions.xml to keep the caches between the two synchronized.
Now when an object is updated on one side or the other, the message gets sent out across JMS. However, when both applications read the message back in, we get strange errors like this one:
ServerSession(1204462)--Thread[PointcastTopicConnection,5,main]--EXCEPTION [TOPLINK-10000] (TopLink - 9.0.3.5 (Build 436)): oracle.toplink.exceptions.JMSProcessingException
EXCEPTION DESCRIPTION: (There is no English translation for this exception.)
INTERNAL EXCEPTION: javax.jms.MessageFormatException: Unable to deserialize object: java.lang.ClassCastException: Assigning instance of class oracle.toplink.internal.sessions.ObjectChangeSet to field oracle.toplink.internal.sessions.ObjectReferenceChangeRecord#newValue
Has anyone seen something like this before? We're using Toplink 9.0.3.5 with OC4J 9.0.2.2 (build 021224.1834).
(Apologies if this isn't Toplink-related. It's hard to tell if this is the App Server or Toplink failing here)
Thanks
- Don

I found this on http://download.oracle.com/docs/cd/B10464_05/web.904/b10313/cache.htm
Note:
When you select JMS as the transport mechanism in the clustering-service element, OracleAS TopLink ignores the discovery setting.
Does ignoring the discovery setting mean that it is not used?
Thanks.

JMS Cache Synchronization - Merge or Database Read?

Similar Messages

Maybe you are looking for