Re: logical replication syntax (was DROP SUBSCRIPTION, query cancellations and slot handling)

On 02/05/17 15:31, Tom Lane wrote:
> Petr Jelinek <petr(dot)jelinek(at)2ndquadrant(dot)com> writes:
>> Let me expand, if we don't drop the slot by default when dropping
>> subscription, we'll have a lot of users with dead slots laying around
>> holding back WAL/catalog_xmin, that's really bad. If we do drop by
>> default like now, we need option to not do that, neither RESTRICT, nor
>> CASCADE fit that.
>
> I'm not sure which part of "you can't have an option in DROP" isn't
> clear to you. Whatever the default behavior is always has to work,
> because that is what's going to happen during DROP OWNED BY, or
> DROP DATABASE. If you want more than one behavior during DROP,
> you need to drive that off something represented as a persistent
> property of the object, not off an option in the DROP command.
>
> I agree that RESTRICT/CASCADE don't fit this, unless the model
> is that the slot is always owned by the subscription, which might
> be too restrictive.
>

What do you think of attached. I actually did add RESTRICT/CASCADE, in a
way that if there is slot RESTRICT will refuse to drop subscription and
CASCADE will try to drop it. Still all errors.

The new way to not drop slot is to set slot_name to NONE which is new
value that I invented (inspiration from OWNED BY sequences) which
basically disassociates the subscription from slot.

It's slightly less automatic this way but perhaps that's not a bad
thing, at least nobody will drop slots by mistake while we still make
sure that slots are not left over without anybody noticing.

Note that this would need catalog version bump as it removes NOT NULL
constraint from subslotname.

--
Petr Jelinek http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services