The current zone routing behavior is such that if the parallel requests do not
satisfy the "required reads", we fall back to the other zone to satisfy our
requests. So even-though the zone-read-count is set to 0, we will be going to
the other zone to satisfy our requests. A better solution is to fail fast
rather than wait for the other zone to reply back.
Original issue reported on code.google.com by rsumb...@gmail.com on 8 Feb 2011 at 2:38
Original issue reported on code.google.com by
rsumb...@gmail.com
on 8 Feb 2011 at 2:38