[martyn-taylor]

Here is a couple of examples of what the configuration may look like when separating out the HA Policy with the node behaviour.

The policy type in this instance represents what is trying to be realised.  In the examples below the nodes are trying to realise a replication policy.  The nodes behaviour can be different depending on the role they wish to take within the context of the replication policy.  The 3 node roles are: "master", "slave", "co-location" (master and slave). 

In the first example we can see that the HA policy this node wants to realise is replication.  The node is attempting to realise this policy in the context of the cluster.  This node will act as a slave in the context of the replication policy.  Another node in the cluster may also be trying to achieve replication but could be acting as a master.  The combination of both master and slave realise the desired replication policy.

In the 2nd example.  The behaviour of the node is to act as both a master and a slave and so it's behaviour here is "co-located".  The bigger picture is still to achieve replication across the cluster.  It contributes towards this overall goal by acting as a master and also a slave to another master.

<ha-policy>
   <replication><!-- Policy Type -->
      <slave><!--node behaviour-->
         <group-name>tiddles</group-name>
         <max-saved-replicated-journals-size>22</max-saved-replicated-journals-size>
         <clustername>33rrrrr</clustername>
         <restart-backup>false</restart-backup>
         <scale-down>
            <!--a grouping of servers that can be scaled down to-->
            <group-name>boo!</group-name>
            <!--either a discovery group-->
            <discovery-group>wahey</discovery-group>
         </scale-down>
      </slave>
   </replication>
</ha-policy>


<ha-policy>
   <replication><!--policy type-->
      <colocated><!--node behaviour-->
         <backup-request-retries>44</backup-request-retries>
         <backup-request-retry-interval>33</backup-request-retry-interval>
         <max-backups>3</max-backups>
         <request-backup>false</request-backup>
         <master>
            <allow-failback>true</allow-failback>
            <group-name>purple</group-name>
            <check-for-live-server>true</check-for-live-server>
            <failback-delay>1111</failback-delay>
            <clustername>abcdefg</clustername>
         </master>
         <backup-port-offset>33</backup-port-offset>
         <slave>
            <group-name>tiddles</group-name>
            <max-saved-replicated-journals-size>22</max-saved-replicated-journals-size>
            <clustername>33rrrrr</clustername>
            <restart-backup>false</restart-backup>
            <scale-down>
               <!--a grouping of servers that can be scaled down to-->
               <group-name>boo!</group-name>
               <!--either a discovery group-->
               <discovery-group>wahey</discovery-group>
            </scale-down>
         </slave>
      </colocated>
   </replication>
</ha-policy>

[ataylor]

I am happy with either way, could you all review and comment and i will go with the consensus

[jbertram]

I like the way Martyn's is organized.  Makes more sense to me.