Hi,
I would like to report something strange I've noticed.
I have been happily managing an application for over a year, using 2 servers (1 production machine and 1 test machine), each of which running Linux and JBoss 3.2.7 in their "default" configuration (i.e. no cluster)
Now I want to upgrade the production evironment, so I have set up 2 new Linux servers with JBoss 4.0.3SP1 in 1 cluster ("all" configuration); I would like to point out that, in setting up the new clustered environment, I have not changed any of the default values (i.e. the partition name is DefaultPartition, etc.).
I have a client application, which uses 2 Jboss libraries
- jboss.jar (that I've copied from jboss-4.0.3SP1/server/default/lib)
- jbossall-client.jar (from jboss-4.0.3SP1/client)
The jndi.properties of the client application is as follows
java.naming.factory.initial=org.jnp.interfaces.NamingContextFactory
java.naming.provider.url=<DNS name of the test server>:1099
java.naming.factory.url.pkgs=org.jboss.naming:org.jnp.interfaces
But when I run the client application, I see the production data, not the test ones.
Furthermore, when the test client is running, doing a "netstat" on the test server I see established connections between ports 4444 and 1098 of the test server and random ports of the production servers.
In other words, it seems that the test server, although it's in the "default" configuration, takes somehow part in the cluster.
In fact, shutting down the production servers solves the problem (obviously), and so does setting a partition name other than "DefaultPartition" in the production servers, using the jboss.partion.name property.
I would like to know whether this behavior is by design; if it is a bug instead, I'll be happy to help you determine what the problem is.
Kind regards
--
Having been asked to provide a reference for this man, I can confidently state that you will be very lucky indeed if you can get him to work for you.
.~. Roberto Neri
/V\ Registered Linux User 88943.
/( )\ Palermo, Italy.
^^-^^
