On 8/11/06, Eugene Prokopiev <[EMAIL PROTECTED]> wrote:
I tried to use one clustered database for some brokers in assumption
that every broker connected to database (but not connected into network)
will see the same messages in queues and so there are no difference to
which broker consumer will be connected.
But my assumption is wrong. I can send message to one queue on broker,
but another broker will not refresh the same queue relaing on database
content.
Exactly. Each broker has its own dataabase.
So, there are no way to create network of brokers in which every broker
is ready to serve clients with same queue contents, am I right?
You can create a network of brokers each using their own database. In
a network each broker has its own queues and its own messages etc.
Only master-slave replicates messages to multiple brokers.
It's possible to use network topology in which only one broker has
message, so this message can be loosed on broker failure.
Only if you loose the persistent store. If you are using a clustered
JDBC database this is not the case
Another option
is copy every message from master broker to slave broker, but in this
case I can work with only one broker (master) at any point in time.
This is Master Slave. You can put multiple master-slave pairs into a
network if you really need that many brokers.
In last release I can use only standart Master/Slave configuration
described in http://activemq.org/site/masterslave.html, am I right?
Yes
When release with Shared File System Master Slave and JDBC Master Slave
will be available for production use?
When the 4.1 release goes out which should be soon.
Where can I download snapshots
with it support?
Grab the 4.1-SNAPSHOT (see the link on the download page)
http://people.apache.org/maven-snapshot-repository/org/apache/activemq/apache-activemq/4.1-incubator-SNAPSHOT/
--
James
-------
http://radio.weblogs.com/0112098/
