Missing messages when target node of bridge is started later than source node

Hi Clebert,
I have found possible new problem with bridge. See following scenario which was tested with NB r10291.

1. Start node with input queue and bridge definition
2. Start producer and push all messages into input queue, e.g. 100 messages
3. Start node with output queue (bridge is configured input queue → output queue)

After step 3 messages from input queue are consumed out (MessagesCount = 0, Consumers=1, MessagsAdded=100, which is correct) but messages are not available on the output queue (MessagesCount = 0,MessagsAdded=0). Consumer connected to output queue is not able to read any message.
If I restart producer again (100 messages and both nodes are online) pushed messages are successfully delivered to output queue and consumer is able to read all 100 messages.

If I remember correctly I have not seen this issue before and it seems like regression. I am going to test this scenario with previous NB and check if it is regression or no. I will prepare simple reproducer for you tomorrow.

I think that this issue represents problem for LODH.

Bridge (hornetq-configuration.xml):
<connector name="remote-connector">
<factory-class>org.hornetq.core.remoting.impl.netty.NettyConnectorFactory</factory-class>
<param key="host" value="127.0.0.3"/>
<param key="port" value="5445"/>
</connector>
...
<queues>
<queue name="jms.queue.inputQueue">
<address>jms.queue.inputQueue</address>
</queue>
</queues>
...
<bridges>
<bridge name="my-bridge">
<queue-name>jms.queue.inputQueue</queue-name>
<forwarding-address>jms.queue.outputQueue</forwarding-address>
<static-connectors>
<connector-ref>remote-connector</connector-ref>
</static-connectors>
</bridge>
</bridges>