Enhance timeout and make it configurable

While debugging a customer environment, we saw that DNS appeared to have some kind of rate limiting. While it worked when testing lookups from shell etc, it was likely failing during rebalance. It caused rebalance failure due to calls timing out. While we can't prevent it from affecting us, it is worthwhile to give larger timeout to calls that may hit DNS - perhaps 15s instead of current 5s when reaching out across network to other nodes. It would also be useful to have this as configurable.