Hi,
On 2022-07-15 13:55:32 +0900, Kyotaro Horiguchi wrote:
> Yeah, the limitation by max_sync_workers_per_subscription is performed
> on subscriber, but replication slot drops happen not on the
> subscriber, but at the termination of corresponding walsender process
> on publisher. So, there's a lag between the finish of subscription
> worker and the corresponding slot's drop. Thus, a new sync worker can
> be created while the walsenders corresponding to some already finished
> sync workers is still going to finish.
Why are we relying on the slots being dropped at the end of connection? That
doesn't seem like a good idea to me. Can't we just do that explicitly? We
still need the on-connection-close cleanup to deal with network failures etc,
but that doesn't mean we can do something else in the happy path.
Greetings,
Andres Freund
