>>>>> "Simon" == Simon Cocking <simon@stripped> writes:
>> What timeout value have we missed?
Simon> I believe Sasha hit the nail on the head -- a separate timer for
Simon> replication queries, to allow for extremely large data transfer between
Simon> servers. Ideally, a replication timer should not be necessary since
Simon> (bugs notwithstanding :o) replication involves two MySQL *servers*, not
Simon> one server and an unknown client. Networking being what it is,
Simon> though... :o)
We have added a replication timer for slaves in 3.23.40 (default one
hour / packet), which I am building for distribution just now.