Re: Exposing the stats snapshot timestamp to SQL

Tomas Vondra <tomas(dot)vondra(at)2ndquadrant(dot)com> writes:
> I see the patch only works with the top-level snapshot timestamp, stored
> in globalStats, but since 9.3 (when the stats were split into per-db
> files) we track per-database timestamps too.

> Shouldn't we make those timestamps accessible too? It's not necessary
> for detecting unresponsive statistics collector (if it's stuck it's not
> writing anything, so the global timestamp will be old too), but it seems
> more appropriate for querying database-level stats to query
> database-level timestamp too.

I'm inclined to say not; I think that's exposing an implementation detail
that we might regret exposing, later. (IOW, it's not hard to think of
rearrangements that might mean there wasn't a per-database stamp anymore.)

> But maybe that's not necessary, because to query database stats you have
> to be connected to that particular database and that should write fresh
> stats, so the timestamps should not be very different.

Yeah. The only use-case that's been suggested is detecting an
unresponsive stats collector, and the main timestamp should be plenty for
that.

regards, tom lane