Replication is for local replication, not for a latent network, and even so it's for a single node replication.
There is this JIRA here: https://issues.apache.org/jira/browse/ARTEMIS-31 It will need some design though, as if we do this at the journal level we would need the same level of confidence we have at local replication. With data centers you have a not so reliable network, so maybe we need a different solution, maybe replicate messages at the queue level. The way we do replication on Artemis, we sync the nodes data while live is running, and having a non reliable network could have issues on that operation. Someone proposed last week divert the messages to the other datacenter and having them being consumed again at the other datacenter. you could do that by diverting the queue to a different queue and bridging that queue. It's a bit manual process but something that could be achieved now, something that could be done with Virtual Topics in ActiveMQ5. On Wed, Nov 9, 2016 at 5:01 PM, Quinn Stevenson <qu...@pronoia-solutions.com> wrote: > I’m trying to figure out if how to configure an Artemis cluster to meet my > customer’s requirement for message availability. > > The basic setup is two datacenters with two Artemis brokers running in each > datacenter for a total of for brokers. All of the brokers will be in the > same cluster. > > What I’m struggling with is trying to figure out how to configure the > replication. I would like to ensure that a broker in datacenter A will alway > replicate to a broker in datacenter B so that if I lose an entire datacenter, > I haven’t lost any messages. > > Is that possible in Artemis? If so, can you provide some advice on how I’d > go about configuring the cluster? > > > > -- Clebert Suconic
