I have a Java client that connects to MQ with 10 connections. These remain open for the duration that the Java client runs. For each thread we create a message, create a session, send the message and close the session. We are using the Spring CachingConnectionFactory and have a sessionCacheSize of 100. We have been told by our MQ engineering team is that that our queue manager has a max connections of 500 and that we are exceeding this. The QM.ini file has:
maxChannels=500
maxActiveChannels=256
maxHandles=256
What I have observed in MQ explorer is that the open output count on the queue remains static at 10, however if we load balance across 2 queues it's 10 on each, even though we still only have 10 connections. So what I'd like to know is what do jms connections and sessions equate to in MQ terminology?
I did think that a connection equates to an active channel and a session is a handle, so it was the handles that we were possibly exceeding as the amount of sessions we open (and close) run into hundreds or thousands, whereas we only have 10 connections. Although going against this, the snippet below from IBM's Technote "Explanation of connection pool and session pool settings for JMS connection factories", suggests that the max channels should be greater than the max sessions, however we never know this as it depends on the load (unless this should be greater than the sessionCacheSize?).
Each session represents a TCP/IP connection to the queue manager. With the settings mentioned here, there can be a maximum of 100 TCP/IP connections. If you are using WebSphere MQ, it is important to tune the queue manager's MaxChannels setting, located in the qm.ini file, to a value greater than the sum of the maximum possible number of sessions from each JMS connection factory that connects to the queue manager.
Any assistance on how best to configure MQ would be appreciated.
