"Kevin Grittner" <kgrittn@mail.com> writes:
> That is the problem -- until the point where pg_stop_backup() was
> run PostgreSQL can't be sure of having a consistent database. It is
> waiting from enough WAL to get it there. My practice is always to
> keep the last two base backups and all WAL from the start of the
> earlier one to current. If I want to recover to a point in time
> which was during a backup, I use the previous base backup, which is
> complete and usable.
> It does seem that there is an opportunity here to make it more
> clear what is happening.
Seems like we could throw an error if the requested stop time is earlier
than the pg_stop_backup time. (That should then allow some code
simplification --- I imagine currently we're doing a max() calculation
somewhere ...)
regards, tom lane