We're deploying a set of MDB's that are configured to talk to a set of WebSphere MQ destinations. There are about 70 MDBs in total, talking to about 35 destinations on two separate WSMQ queuemanagers.
All the MDBs are bundled inside a single EJB-JAR. When shutting down the application (to shutdown the server, or re-deploy), each MDB takes around 5 seconds to stop, and they seem to be stopped serially, so a re-deploy takes 3-4 minutes to complete, rather than a few seconds.
If I use the JBossMQ provider instead of our "real" provider, things are snappy and shutdown occurs in a few seconds. We are currently using this during development.
Does anyone know why the shutdown of the EJB's is done serially, and if that can be avoided? Disconnecting from WSMQ is going to take a few seconds no matter what, but it's a bit painful that everything can't be done in parralel.
This question *might* be more appropriate in another forum, but I'm not sure which one. If you know, please feel free to move it.