On Mon, Nov 16, 2009 at 3:10 PM, johnluo1 <johnl...@gmail.com> wrote:
>
> Hi,
>
> I wonder if following is possible.
>
> We have broker1 and broker2 configured as a broker cluster. On our load
> balancer, we define a VIP, which will map to broker1 and broker2's physical
> IP addresses. Clients will connect to the VIP, and load balancer will
> redirect the connections to either broker1 or broker2. A client can connect
> to broker1 through VIP and publish messages there, while another client can
> connect to broker2 through VIP and consume messages published by first
> client, since broker1 and broker2 are in a cluster.
>
> Is there any way we can achieve HA in such setup? That is, if broker1 is
> down, clients will still be able to connect to broker2 through VIP and
> consume all messages. The only way this can happen is that all messages need
> to be replicated between broker1 and broker2, like a master/slave setup.
>
> Is this something that can be configured?
A master/slave configuration for each broker1 and broker2 should
provide what you need. With a master/slave configuration, if broker1
goes down then the slave will become the new master and there will be
no message loss. The same applies to broker2. This assumes that the
load balancer is also aware of the slaves.
Bruce
--
perl -e 'print
unpack("u30","D0G)u8...@4vyy9&5R\"F)R=6-E+G-N>61E<D\!G;6%I;\"YC;VT*"
);'
ActiveMQ in Action: http://bit.ly/2je6cQ
Blog: http://bruceblog.org/
Twitter: http://twitter.com/brucesnyder
