On 4/29/07, DavidR <[EMAIL PROTECTED]> wrote:
Ok, so if we go back to Master Slave, then we have the following issue: Currently we are using the JDBC Master Slave but we see this uses a lot of CPU. So, we would want to go to pure Master Slave. So, can you help us understand the following: It says in the docs that if one goes down "A failed master cannot be re-introduced without shutting down the the slave broker". Can it be re-introduced as a slave without any restart of the running queue?
The issue is we don't have a real time broker-broker synchronization mechanism yet for pure master/slave. So you can start up 2 brokers in a master/slave; if the master fails the slave takes over just fine. But if you wish to introduce another slave/master to the master after failover, you need to stop the broker & copy its files to the new location. We don't yet have a way to dynamically bring a new cluster member online to a running cluster.
Also, if the slave had items in the queue not processes and then came up as slave, what would happen with those messages?
All kinds of issues would arise if you brought up an old master/slave into a running master/slave cluster without synchronizing. e.g. ACKs could be missed so messages could be duplicated - plus messages could be lost. -- James ------- http://macstrac.blogspot.com/
