AFAIK there will not be duplicates (even rare ones!) - in 4.1 and
especially 5.0
cheers,
Rob
On Jul 16, 2007, at 11:59 PM, ksu wrote:
There is a message posted in the user forum back in early 2006
regarding the
limitation on the configuration of a network of brokers
(http://www.nabble.com/Network-of-Brokers-...-
tf1117526s2354.html#a2921710).
The post suggests any topology involving cyclic graph may result in
duplicate messages been delivered. Does this limitation still
exist in 4.1x
/ 5.0x?
I am working with the latest 5.0 snapshot and I can't get duplicate
message
to occur in a star configuration where every node is statically
connected to
every other node in a 4 node broker network with ports 62001,
62002, 62003,
and 62004:
62002
/ | \
/ | \
62001 ----+----62003
\ | /
\ | /
62004
e.g. Network connector def for 62003:
<networkConnector
uri="static:(tcp://localhost:62001,tcp://localhost:62002,tcp://
localhost:62004)"/>
If anyone know if this configuration may cause (rare?) problem with
message
duplication in 4.1x / 5.0x, please let me know.
P.S.: One problem I noted, as described in a separate post to this
forum
earlier, is if any of the 4 nodes dies, it may cause
javax.jms.InvalidClientIDException exception in the reminding nodes
when it
try to come back online. The way to reliably stabilize the system
whenever
this happens is to terminate and restart all nodes in the broker
network.
--
View this message in context: http://www.nabble.com/Network-of-
Brokers-configuration-limitation--tf4093369s2354.html#a11639335
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
