Hi, I am currently running ActiveMQ brokers using the Shared kahadb based Master/Slave topology. This setup currently uses NFS/Filer for kahadb state and the primary and failover brokers contend for lock on shared state such that only one broker is active at a time. However, I now want to avoid using NFS dependency and shared state between primary and failover brokers and instead use local disk for all access. I have the following queries in this regard:
1. I was wondering if ActiveMQ provides some replication support from primary to failover brokers in real time so that even if I lose primary broker (and its persistent store), my failover broker can continue with it's own replicated state. Looks like ActiveMQ provides Replicated leveldb store. Is there such a replication possible with kahadb based persistent store? 2. Does replicated leveldb store replicate from primary to failover broker synchronously or asynchronously? Would the producer block waiting for messages to be replicated from primary to failover? 3. I read about the master-slave discovery in network of brokers (http://activemq.apache.org/networks-of-brokers.html). Could somebody please help me understand this in more detail? Specifically -- Does master-slave network of brokers ensure only one master is active at a time and all clients are connected to the same master broker? What happens when the master broker goes down? Will one of the slaves get reelected as the new master? Do all the slaves replicate from master in real time? synchronously or asynchronously? Thanks a lot!! -Deepak -- View this message in context: http://activemq.2283324.n4.nabble.com/Avoiding-shared-state-between-master-and-slave-brokers-tp4686401.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
