HornetQ-2.1.2-Final lose messages when cluster and the message's expirationTime is specified?

[shooeugenesea.tw]

Hi,

I use HornetQ-2.1.2-Final.

I setup cluster environment in machines whose ips are 192.168.0.17 and 192.168.18.107.

There is a consumer connect to 192.168.18.107.

There is a producer connect to 192.168.0.17

There is a queue: ExampleQueue in both machine's HornetQ server.

The ExampleQueue's expiry-address is also ExampleQueue.

The machine 192.168.18.107 time is earlier than the machine 192.168.0.17 about 1 minute.

For example, when 192.168.18.107 is 15:27, the 192.168.0.17 is 15:28.

When the producer send a text message with expirationTime 3000 millis,

I expect the consumer will receive the message, but it didnt.

When I change the expirationTime to 2 minutes, the consumer will receive the message correctly.

I monitor HornetQ through jconsole,

and saw the ExampleQueue's MessagesAdded and MessageCount are not changed in both HornetQ server.

I can't find such information in the community and google, so post here.

I attached my environment image, configuration and test code.

Is my configuration wrong?

Thanks

[ataylor]

you will need to sycn the clocks on your machine or set the expiry time to something big enough

[shooeugenesea.tw]

Thanks for your response.

I know I can solve the problem by sync the clock or extend the expiry time.

But I have no idea why the message is lost..

I think the message is not expired yet.

The message is lost when HornetQ redistribute message from 192.168.0.17(the time is 15:28) to 192.168.18.10(the time is 15:27)

I hope the message is expired, because I setup the expired queue.

If the message is expired, I can receive the message correctly.

It's strange because the message is lost, not expired..

Thanks for your response again.

[ataylor]

if at the point the message is redistributed it has expired then it will be routed to the expiry address.

[shooeugenesea.tw]

but the message is not routed to the expiry address.

the queue's mBean attributes "MessagesAdded" and "MessageCount" are not increased.

If the message is expired, I think I should be able to see it in the expired address.

At least I should see the expiry address's MessagesAdded should increase 1, right?

BTW, do you see my first post?

the MessageProducer send message to ExampleQueue.

the ExampleQueue's expiry address is also ExampleQueue.

I attached the configuration files.

If the message is routed to the expiry address, the ExampleQueue should have one message.

[ataylor]

Its hard for me to say either way you havent explained your full topology.  do both nodes have an expiry address with a queue, if so check both nodes. I think the fact that if you up the expiry time it works tells us what is happening.

[shooeugenesea.tw]

I draw image in the attached file to describe my deploy environment.

There are two machines in LAN.

I startup HornetQ in each machine.

Both nodes have expiry address with a queue, and I can't see the message is both nodes.

I think I have no more information for you, it's my fault.

Thanks for your response anyway.

[ataylor]

I cant say for sure whats happening, stck logging on and trace where the message goes to

[shooeugenesea.tw]

it seems the message is expired in the HornetQ's cluster core bridge,

but the HornetQ's cluster core bridge has no expiry-address, so the message is lost.