I wonder if it might help to diagnose whats going on by starting off
running the pair of brokers separately, outside of JBoss and try
taking up/down different bits. As right now there's alot of moving
parts (JBoss, JCA, RA, ActiveMQ client and the brokers).

On 9/1/06, dooart <[EMAIL PROTECTED]> wrote:


I am trying to setup an environment with a clustered JBoss and ActiveMQ in
JDBC Master/Slave mode. I'm going to use JMS in our application to help in
some issues with clustering, like indexing and local files synchronization.

I have only two machines to run the entire environment and I want to run an
embedded ActiveMQ broker per application server to help in application
maintenance, so the final environment will be something like this:

http://www.nabble.com/user-files/235837/cluster_activemq.png

At first it seemed to be working fine, but when I started to test the
behaviour of the environment shutting down the nodes and restarting them, I
started to have some problems with message duplication and lost messages.
Here are the results of some tests I made (errors are marked with red
color):



First Test



Step    Action  Result


1       Servlet at NODE01 sends message "node01_test01" Both EJB containers
receive the message "node01_test01"


2       Servlet at NODE02 sends message "node02_test01" Both EJB containers
receive the message "node02_test01"


3       NODE02 is shut down and restarted       NODE02 leaves and enters the 
cluster
again


4       Servlet at NODE01 sends message "node01_test02" Both EJB containers
receive the message "node01_test02"


5       Servlet at NODE02 sends message "node02_test02" Both EJB containers
receive the message "node02_test02"


6       NODE02 is shut down     NODE02 leaves the cluster


7       Servlet at NODE01 sends message "node01_test03" Only the EJB container 
at
NODE01 receives the message "node01_test03" (ok!)


8       NODE02 is started       NODE02 enters the cluster and receives the 
message
"node01_test03" (ok!)


9       NODE01 is shut down     NODE01 leaves cluster and NODE02 receives again 
the
message "node01_test03" (twice)


10      Servlet at NODE02 sends message "node02_test03" Only the EJB container 
at
NODE02 receives the message "node02_test03" (ok!)


11      NODE01 is started       NODE01 enters the cluster and receives the 
message
"node02_test03" (ok!)


12      Servlet at NODE01 sends message "node01_test04" Both EJB containers
receive the message "node01_test04"


13      Servlet at NODE02 sends message "node02_test04" Both EJB containers
receive the message "node02_test04"


14      NODE02 is shut down     NODE02 leaves cluster and NODE01 receives again 
the
message "node02_test03" (once)


15      Servlet at NODE01 sends message "node01_test05" Only the EJB container 
at
NODE01 receives the message "node01_test05" (ok!)


16      NODE01 is shut down     Cluster is empty


17      NODE02 is started       NODE02 enters the cluster but doesn't receive 
the
message "node01_test05" as expected


18      NODE01 is started       NODE01 enters the cluster


19      NODE02 is shut down and restarted       NODE02 enters the cluster again 
but
still doesn't receive the message "node01_test05"




Second Test



Step    Action  Result


1       Servlet at NODE01 sends message "node01_test01" Both EJB containers
receive the message "node01_test01"


2       Servlet at NODE02 sends message "node02_test01" Both EJB containers
receive the message "node02_test01"


3       NODE02 is shut down and restarted       NODE02 leaves and enters the 
cluster
again


4       Servlet at NODE01 sends message "node01_test02" Only the EJB container 
at
NODE01 receives the message "node01_test02" (now it didn't worked although
it worked at the previous test)


5       Servlet at NODE02 sends message "node02_test02" Only the EJB container 
at
NODE02 receives the message "node02_test02" (now it didn't worked although
it worked at the previous test)





I've run the first test with MySQL 5.0 and the second test with MSSQL Server
2000 as datasources for JDBC Master/Slave, but both configurations failed as
you can see at the two tables above. I'm almost sure that the steps 4 and 5
from the second test failed not because of MSSQL, but because the
instability of the entire environment.

Does anyone have a clue of what can be wrong? I don't know if this helps,
but I've attached my configuration files here.

I'm using JBoss 4.0.4 GA and ActiveMQ 4.1 (I was using 4.0.1 before, but it
didn't work too).

Thanks,

Thiago Duarte

--
View this message in context: 
http://www.nabble.com/ActiveMQ%2C-Clustered-JBoss-and-JDBC-Master-Slave-tf2204736.html#a6105885
Sent from the ActiveMQ - User forum at Nabble.com.




--

James
-------
http://radio.weblogs.com/0112098/

Reply via email to