Hi,
For test purposes, I've to get knowledge on JBoss Clustering, fail-over, replication...

I've been turning around the following issue for some days now,
reading JBoss Clustering (Edition 7), Wiki, forum and bug reports. I
just hope it's something stupid I cannot pinpoint.

Setting : JBoss 4.0.1 ; 2 Linux servers (Suse) in the Cluster.

Beans are :
- HelloMDB : basic message driven bean listening on 'queue/hello'
- HelloServletJMS : servlet to send messages to 'queue/hello'

Note :
- both servers are multihomed
- Oracle is the common Database used (runs on Server1) ( deployed using ConfigJBossMQDB wiki page )
- use of DefaultPartition with default ports.

Server2 is started prior to Server1 (thus Server2 is the HA-JNDI server.)

Inside HelloServletJMS's, the code to send the message to HelloMDB (all try/catch and close removed):

InitialContext context = new InitialContext();
 Queue queue = (Queue) context.lookup("queue/hello");
 QueueConnectionFactory queueFactory = (QueueConnectionFactory) context.lookup("XAConnectionFactory");
 QueueConnection queueConnection = queueFactory.createQueueConnection();
 QueueSession queueSession = queueConnection.createQueueSession(false, javax.jms.Session.AUTO_ACKNOWLEDGE);
 QueueSender queueSender = queueSession.createSender(queue);
 TextMessage message = queueSession.createTextMessage();
 message.setText(text);
 queueSender.send(message);

When doing an http request (from a third machine on another LAN) on HelloServletJMS :
on Server2 => OK : message treated by HelloMDB ( on Server1 or Server2 )
on Server1 => KO

Error on server1 is :

javax.naming.NameNotFoundException: queue not bound
 at org.jnp.server.NamingServer.getBinding(NamingServer.java:491)
 at org.jnp.server.NamingServer.getBinding(NamingServer.java:499)
 at org.jnp.server.NamingServer.getObject(NamingServer.java:505)
 at org.jnp.server.NamingServer.lookup(NamingServer.java:249)
 at org.jnp.interfaces.NamingContext.lookup(NamingContext.java:544)
 at org.jnp.interfaces.NamingContext.lookup(NamingContext.java:520)
 at javax.naming.InitialContext.lookup(InitialContext.java:347)
 at hello.web.HelloServletJMS.doGet(HelloServletJMS.java:135)
 ...

META-INF/jboss.xml of the jar is ( message-driven part only ) :

<message-driven>
 <ejb-name>HelloMDB</ejb-name>
 <destination-jndi-name>queue/hello</destination-jndi-name>
 </message-driven>

/opt/jboss/server/all/deploy/jms/hajndi-jms-ds.xml (JMS part) :

<mbean code="org.jboss.jms.jndi.JMSProviderLoader"
 name="jboss.mq:service=JMSProviderLoader,name=HAJNDIJMSProvider">
 <attribute name="ProviderName">DefaultJMSProvider</attribute>
 <attribute name="ProviderAdapterClass">
 org.jboss.jms.jndi.JNDIProviderAdapter
 </attribute>
 <!-- The combined connection factory -->
 <attribute name="FactoryRef">java:/XAConnectionFactory</attribute>
 <!-- The queue connection factory -->
 <attribute name="QueueFactoryRef">XAConnectionFactory</attribute>
 <!-- The topic factory -->
 <attribute name="TopicFactoryRef">XAConnectionFactory</attribute>
 <!-- Access JMS via HAJNDI -->
 <attribute name="Properties">
 java.naming.factory.initial=org.jnp.interfaces.NamingContextFactory
 java.naming.factory.url.pkgs=org.jboss.naming:org.jnp.interfaces
 java.naming.provider.url=[Local IP of Server2]:1100,[Local IP of Server1]:1100
 </attribute>
 </mbean>

JGroup uses UDP ; from /opt/jboss/server/all/deploy/cluster-service.xml :

<UDP mcast_addr="228.1.2.3" mcast_port="45566"
 ip_ttl="32" ip_mcast="true"
 mcast_send_buf_size="800000" mcast_recv_buf_size="150000"
 ucast_send_buf_size="800000" ucast_recv_buf_size="150000"
 loopback="false" bind_addr="[LocalIP Due To Multi Homed Server]" />

Where local IP is correctly set to the ethernet card in use on the LAN.

Things verified :
- /etc/hosts :
"127.0.0.1 localhost" and local IPs are on 2 different lines.

- /etc/modules.conf (just in case since IPv6 is not used) :
"alias net-pf-10 off"

- multicast route is fine :
"netstat -nr | grep 224" gives the correct multicast route on both devices ( and on the correct ethernet card device )
224.0.0.0 0.0.0.0 240.0.0.0 U 40 0 0 eth0

- I do start with "-c all" and the --host=[LocalIP] options

I tried different combinations with "jnp://..." and "localhost:1100" for java.naming.provider.url ( even if loopbackIP:1100 was not opened )

I checked traffic on Server2:1100 . Port is used by Server1 at startup ( first in java.naming.provider.url ), but no traffic afterward ( neither on the autodiscovery port 1102 ).

I checked traffic on 230.1.2.3 : ok (heartbeat packets go through)

I checked traffic on autodiscovery multicast ip 230.0.0.4 : ok (membership packets go through)

If Server1 becomes HA-JNDI server, lookup on 'queue/hello' from Server1 are ok, not lookup from Server2. (only difference in config files is the bind_addr due to multi-homed servers)

What did I miss here ?
If more src/config files are needed to solve this, i'll be more than happy to send them.

Regards,
Alex