On Wed, 2009-12-30 at 15:31 +0200, Heikki Linnakangas wrote:
> Simon Riggs wrote:
> > The code has always been capable of starting without this, which was
> > considered a feature to be able start from a hot copy.
>
> Why is that desirable? The system is in an inconsistent state. To force
> it, you can always use pg_resetxlog.
It's not desirable, for me, but its been there since 8.0 and I have no
info on whether its used. If you have a workaround that allows us to
continue to support it, I suggest you document it and then plug the gap
as originally suggested by you.
-- Simon Riggs www.2ndQuadrant.com