Tom Lane wrote:
> Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> > Mart�n Marqu�s wrote:
> >> This really gives little use for recovery_target_xid. :(
>
> > Hmm, you can still use pg_xlogdump to figure it out from the actual WAL,
> > which has the correct XIDs. It's obviously a worse solution though from
> > the user's POV, because it's hard to figure out what WAL record
> > corresponds to the change you care about ...
>
> To what extent does the commit_ts infrastructure fix this?
I don't think it does at all. You could try to find out the XID using a
timestamp you obtain from the log file (knowing that the lookups are the
opposite way, i.e. you give it an XID and it returns a timestamp); but
if that's the scenario, I think it's simpler to use the timestamp in
recovery_target_time directly.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services