Thanks. Hmm. We've had bugs in the past which cause message delivery from a
queue to come to a halt, but I can't think of one which causes it to go very
slowly like this.
You mention that only one of your seven queues behaves like this. Can you
identify anything different about this queue? Are you absolutely certain that
the slowdown is because messages are being delivered slowly and not simply
because your MDB's onMessage() method is taking a long time to return?
Can you describe your setup on more detail. Is this a single instance or a
cluster? Is the broker LOCAL or EMBEDDED? What storage are you using for
message persistence? (The defailt file store, or a JDBC database?)
When in the "slow delivery" state, how many active MDBs are in use, and what
are they doing? (A thread dump would tell you this).
When in the "slow delivery" state, have you tried restarting the MDB
application, whilst leaving the GlassFish instance running?
Nigel
--
[Message sent by forum member 'nigeldeakin']
View Post: http://forums.java.net/node/814250