Proper way to Configure JBoss MQ Queues

[anupama]

Hi Guys,

I need some assistance in configuring JBoss MQ queues.
I need to use Jboss Queues in cluster server
environment where nodes of cluster are different
boxes. I am following these steps to do asynch
processing : placing messages into queue A where
message will trigger MDB which will process request
and place response into Queue B. In this case sender
needs to integrate responses to prepare 1 response,
for this sender pinging Queue receive for x millisecs.
The problem I have right now is sender is waiting for
X ms before even looking for response in Queue B and
also it needs to integrate all responses for requests
that it made, so right now I am n't able to parallel
receive since it needs to integrate all responses
there by I amn't able to do non-blocking receives at
all. I am using JBoss 4.0.1SP1 and using default
JBoss Queue config for this, my questions are

1. What would be the best practice to do parallel
receive to achieve non-blocking receive and also
be able to integrate responses

2. I am deploying Jboss queues as HAJNDI, but do I
need to do any configuration for making it work in cluster

3.I read that it's not good for production to use
default HSQLDB for JMS persistence is that true if so
and if I configure Oracle database for that what
should be database schema, can you please throw some
light on this and if there any other useful links
could you please mention so




Thanks alot,
Anu

[anupama]

Hi Guys,

I haven't heard any thinh about my posting, as follow up we decided to use local queues for time being and seems like working fine

Local Queue Questions:
1. If we have a local queue and switch to Oracle then assume that we have x boxes in cluster, if box1 places message into it's loca queue is it possible for box2 to receive that and process request. I guess my question is if we have local queues for each box in cluster(with Oracle as persistent storage)is it possible for other boxes to see that message.

HA-Singleton:
I need to deploy my Queues as HA-Singleton to run only on 1box in cluster, I followed Wiki for HA-Singleton and using all-has singleton configuration only but seems like only master node is able to process messages from queues but not other nodes though they are up and running. I browsed net for a long time and couldn't find any concrete example. So could any of you please throw some light on what's going and I am getting following exception if I hit node other than Master node and try to place message into queue.

Caused by: javax.naming.NameNotFoundException: XAConnectionFactory not bound
 at org.jnp.server.NamingServer.getBinding(NamingServer.java:514)
 .. 42 more

Any help is greatly appreciated.

Thanks alot

[gcoleman]

If you are looking up XAConnectionFactory you will need to use HAJNDI.

See JBossHAJNDIUseCluster on the Wiki.

[anupama]

Hi gcoleman,

Thanks for your reply, my server is configured to use HA-JNDI by default and it have jndi.properties as mentioned in Wiki in conf directory. When I am accessing ConnectionFactory within server I am just calling new InitialContext() since it is within server do I have to explicitly include properties within Context to get HA-JNDI context. All I am doing is
Context x = new InitialContext()
x.lookUp("ConnectionFactory")


So to get SINGLETON HA-JMS ConnectionFactory do I have to say anything else when I am getting IntialContext. server-all-deploy-has-singleton has all files that are shipped with server.


Any help is greatly appreciated.

Thanks alot,
Anu