Multiple brokers in duplex mode: Expected Behavior?

[kpalania <kp...@yahoo.com>]

This is my setup:
* I have 3 brokers configured and running: 1 broker in the cloud and 2
brokers in the LAN.
* There is no network connector defined in broker 1. The transport connector
is set to use TCP on 61616.
* The network connectors on both the brokers in the LAN have "duplex" set to
TRUE and the URI points to broker 1 (tcp://IPAddress:61616).
* I start all the brokers.
* Send a message to broker 1 in the cloud over TCP from a local client
(running on the same machine).
* The message does not get sent to either of the other 2 brokers and I see
the following exception in the log of one of the brokers in the LAN:

"Network connection between vm://localhost#<something> and
tcp://IPAddress:61616 shutdown due to a remote error: java.IO.EOFException"

* Now, I shutdown the broker in the LAN that reports this exception. 
* Send another message over TCP to the broker in the cloud (same as before).
* The message is sent to the only broker that is now running in the LAN.

Questions:
===========

1. Why is this the case? And what is wrong with my configuration?
Essentially, I do not want to see any exceptions and want one of the brokers
to pick up the message in the earlier case. Is that not possible?

2. Also, I noticed that at times I see duplication of messages (and it is
quite frequent). When I go to the ActiveMQ admin UI and click refresh on the
queue, I see the same message repeated several times (though the client just
sent it once).

Thanks.

-- 
View this message in context: http://www.nabble.com/Multiple-brokers-in-duplex-mode%3A-Expected-Behavior--tp18611706p18611706.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Re: Multiple brokers in duplex mode: Expected Behavior?

[kpalania <kp...@yahoo.com>]

If I remove the "duplex" attribute from both the brokers in the LAN, I get an
InvalidClientIDException on one of the brokers..


kpalania wrote:
> 
> This is my setup:
> * I have 3 brokers configured and running: 1 broker in the cloud and 2
> brokers in the LAN.
> * There is no network connector defined in broker 1. The transport
> connector is set to use TCP on 61616.
> * The network connectors on both the brokers in the LAN have "duplex" set
> to TRUE and the URI points to broker 1 (tcp://IPAddress:61616).
> * I start all the brokers.
> * Send a message to broker 1 in the cloud over TCP from a local client
> (running on the same machine).
> * The message does not get sent to either of the other 2 brokers and I see
> the following exception in the log of one of the brokers in the LAN:
> 
> "Network connection between vm://localhost#<something> and
> tcp://IPAddress:61616 shutdown due to a remote error:
> java.IO.EOFException"
> 
> * Now, I shutdown the broker in the LAN that reports this exception. 
> * Send another message over TCP to the broker in the cloud (same as
> before).
> * The message is sent to the only broker that is now running in the LAN.
> 
> Questions:
> ===========
> 
> 1. Why is this the case? And what is wrong with my configuration?
> Essentially, I do not want to see any exceptions and want one of the
> brokers to pick up the message in the earlier case. Is that not possible?
> 
> 2. Also, I noticed that at times I see duplication of messages (and it is
> quite frequent). When I go to the ActiveMQ admin UI and click refresh on
> the queue, I see the same message repeated several times (though the
> client just sent it once).
> 
> Thanks.
> 
> 

-- 
View this message in context: http://www.nabble.com/Multiple-brokers-in-duplex-mode%3A-Expected-Behavior--tp18611706p18613104.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.