Allow maintaining of address order with auto recovery

[ValSmith]

According to the documentation you can use an AddressResolver to choose a desired host first. I created a resolver that returns [host1,host2,host3] but then found that the connections were all to host2 even though host 1 was available. After a while I discovered that when autoRecovery=true the address list is always shuffled. Could there be a possibility to maintain the list order when using auto recovery? Or is there a very intentional reason this is not allowed?

[michaelklishin]

This is by design.

Depending on the order of endpoints doesn't sound like a good idea. Any client can connect to any cluster node. If a "preferred" node is unreachable, attempts to connect to it sequentially will simply increase the amount of time it takes to reach to a reachable endpoint with a "lower priority".