0 Replies Latest reply on Jun 19, 2012 9:51 AM by Andy Taylor

    replication and backups connecting to live

    Andy Taylor Master

      Currently when a back up server is configured using replication you need to set a live connector for it, something like:

       

      <live-connector-ref connector-name="netty-live-connector"/>

       

      For the up and coming Beta release I think that this is enough, however we need to look at providing a better way to do this.

       

      a) we need to support Discovery.

       

      The issue here is how do we configure which actual node the backup will connect to, I think we can either:

       

           1) allow the backup to connect to any live that isnt currently replicating, currently we just stop trying but we could change it to try other live nodes as they become available.

          

           2) Use some configuration to detect this, maybe via <name>live1</name> or something like <live-connector-ref discovery-name="dg1" node-name="live1"/> . i.e the backup only connects to a live with the same name?

       

           3) or both, if a name is supplied use it, if not connect to any node.

       

      b) Allow the use of multiple connector names, if you have multipole backups on the same live server then you wont know which one will actually be live, in this case the backup tries each one until it finds its live.

       

       

      thoughts?

       

      ps I have raised Jiras for these issues