Hello, Ok, It's a little long to explain, but to sum up, imagine two groups, group G1 is composed of nodes A and B and group G2 is composed of nodes C and D. G1 must deal with information I1 and G2 with I2. Each node from the same group acts the same way and they've a local broker and persits locally so queues in nodes A and B store information I1 (mirror nodes, if one fails the other one takes over and the service continues) and the same thing happens in the group G2. When some data is requested it must be removed from all the group's nodes so the storage media from a group should be shared (db or file system). At some point node D fails and the system controller reconfigure the nodes and force the node B to join group G2 (it will be joined to G1 as well) so B must synchronize the node D persistence storage in order to mirror this node (node B will be a node A mirror and a node D mirror). Because a shared file system scheme won't be used for the information persistence,a shared db will be tested. G1 and G2 will use their own shared db (DB1 and DB2). If node D fails and B must join to group G2, it seems easy to start a different broker in B with shared db DB2 as the persistence media. Anyway, there are many other ways to get the same behavior but now I'm checking if this is feasible.
Greetings James.Strachan wrote: > > On 7/10/07, bobsponge <[EMAIL PROTECTED]> wrote: >> Hello James, >> >> I need two or more brokers working at the same time in the same machine >> because the application is composed of several machines (nodes). Every >> node >> belongs to a group. In general data is transferred between nodes from the >> same group, no problem. But a node can bind to another group so the same >> node can communicates with two (or more) groups. The information must >> separately persists in this node so two brokers are needed, because of >> the >> persistence issues... > > I still don't get it; why 2 brokers in the same JVM on the same box? > > Usually for high performance a broker persists to the local hard > drive; so I don't understand the benefit from having 2 different > brokers writing to 2 different directories (or even different DBs) > >> I tried to configure the brokers with different names building an URI >> like >> this: tcp://localhost:50000?broker.brokerName=broker1 , but it didn't >> work >> at all. > > > You configure the broker names in the activemq.xml. See the brokerName > atttribute in the one that comes with the binary distro > > -- > James > ------- > http://macstrac.blogspot.com/ > > -- View this message in context: http://www.nabble.com/Multiple-broker-instances-tf4035160s2354.html#a11517382 Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.
