I have a JEE application (simple EJB application which puts a message to queue) which needs to connect to a uniform cluster queue managers (QMC1, QMC2) and PUT messages. The JEE application runs in a Wepshpere application server,and uses CCDT, with the following commands to consider the queue managers as a group.
DEFINE CHANNEL(QMC1.CHL) CHLTYPE(CLNTCONN) CONNAME('192.168.1.12(9001)') QMNAME(QM) AFFINITY(NONE) TRPTYPE(TCP) REPLACE
DEFINE CHANNEL(QMC2.CHL) CHLTYPE(CLNTCONN) CONNAME('192.168.1.12(9002)') QMNAME(QM) AFFINITY(NONE) TRPTYPE(TCP) REPLACE
I have done the connection factory setting with *QM as the queue manager name. Also in the WAS connection factory -> advanced properties -> Client reconnect options, I have selected
RECONNECT.
My issue is, when both queue managers are running I can put messages to a single the queue, (I have increased the queue work load rank so that all the messages go to a single queue in the cluster (say QMC1).
But if QMC1 is down, the JEE application is unable to connect to the surviving queue manager (QMC2) as it is trying to re-connect to QMC1, not based on the queue manager group.
Have I misunderstood anything here? Will the RECONNECT will try to reconnect to the same queue manager, not the available queue manager based on CCDT? Shouldn't a uniform cluster queue manager support this fail-over behavior via a CCDT queue manager group?
Appreciate any feedback.