I've wasted so much time on trying to get the scheduling feature to work correctly, I feel I need to warn other users before they attempt it.
I've encountered several bugs with it that have been fixed (thanks!) but there are more problems that are harder to reproduce consistently and therefore I can't even create a bug that will get worked on. Here's the brief description of problems: 1. Sometimes the scheduler throws an exception (I don't know why) and that causes the scheduler to stop working. A restart of the broker is required to start it back up. 2. On a true master/slave configuration, on failover the slave seems to pick up ok but messages that were sent while the slave was inactive are suddenly available to be sent. I end up with tons of duplicate messages. 3. Again on failover, the admin always shows "scheduler not started" on the Scheduled tab on the slave server even though it is started and runs. 4. I restarted the broker and it sent 7 messages even though none were queued and the next scheduled time was hours away. And I only send one at a time. 5. If the broker is down and say 2 scheduled times pass (using delay/repeat) when it's restarted only one message is sent and the next one doesn't get sent until after the delay period. I would think they should have both been sent immediately. 6. The master/slave issues are compounded by the Broker restarting itself without logging any error or warning. After this event occurs, the clients are actually still listening to the master but the schedule is running on the slave. Recovery is impossible. -- View this message in context: http://activemq.2283324.n4.nabble.com/Rant-on-scheduling-feature-tp3516236p3516236.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
