Description
The openwire TwoBrokerFailoverClusterTest test fails sporadically. It is balancing 30 connections between 2 brokers with randomized connect, and fails if <35% hit one node. I've only seen it fail because 33.33% hit one node, i.e the difference between 10 and 11 connections. It doesnt seem that anything in play ensures >= 35% actually end up on one node, and the test is just making a slightly optimistic take on what randomizing between 2 things could do. Make it a bit less optimistic to help cut out the sporadic failures while still validating there is a spread.