On 10/26/2010 4:24 AM, Gary Tully wrote:
That looks like a bug in the peer transport factory. I uses
localhost:0 as the tcp transport url, but localhost maps to the
loopback address since 5.3[1] so the connection is refused.
The peer transport factory should use the wildcard address, 0.0.0.0
For me, the problem has been a little more complicated than this.
Switching PeerTransportFactory to using 0.0.0.0 sends my message, but
sends my non-DNS computer name as the target host, so the alive message is:
%localhost%Michael-Dell:49225
If I comment out that section of TcpTransportServer, I can get 0.0.0.0
instead of Michael-Dell, but that's not much more helpful.
If I commit to a network interface, I can get the discovery message to
be useful, but I'm not aware of a "right" way to pick the network
interface when several are available.
Is there a reason why the MulticastDiscoveryAgent doesn't consider the
source IP of a discovery message?
michael
Can you raise a jira issue for this? It is an easy fix, so we can get
it quickly into a new snapshot.
As a work around, you would need to use an xml configuration file for
each broker and the xbean
url scheme: http://activemq.apache.org/broker-xbean-uri.html
[1] https://issues.apache.org/activemq/browse/AMQ-2094
