no, if the trafic is one way, producer to ha pair, a network bridge
just on the embedded producer broker is suffice.
you say that the consumers don't get messages when the ha pair master
is stopped.
Are the consumers using failover: with both master and slave urls present?
do you see the
there is a periodic checkpoint of broker state that writes to the
journal. You can configure a longer checkpointInterval to verify that
this is what is happening.
see: http://activemq.apache.org/kahadb.html
The checkpoint state contains a producer audit it that is written even
if it is not
the default thread pool allow their threads to exit after 30 seconds
of inactivity so you may be just hitting that boundary.
Unfortunately that value is not configurable at the moment.
If you could enable trace level logging for the broker, it should be
possible to determine what component is
Even with no consumer attached to MSSD5D101, it still gets flooded.
On Wed, May 9, 2012 at 3:09 PM, Chris Robison chrisdrobi...@gmail.comwrote:
I have a very simple network of brokers. Their names are as follows (to
help understand the log):
MSTMIP101
MSSD5D101
They each reside in