FailoverTransport With maxReconnectAttempts Results In High CPU and Thread Count

If you define a network connector with the maxReconnectAttempts and there is no broker listening at the remote endpoint, the result is increasing thread count and excessive cpu usage by the broker.

So given the NC below, which does not have a remote broker at 10.0.0.5:61616, you'll note that a few minutes after starting the broker, its thread count begins to rise and it pegs the CPU at > 100%.

<networkConnector name="my-connector" uri="static:(failover:(tcp://10.0.0.5:61616)?maxReconnectAttempts=0)" userName="admin" password="admin" />

The value assigned to maxReconnectAttempts can be 0 or greater. The higher the value, the longer it seems to take for the problem to occur.

Attached thread dump and top output taken while problem was occurring.