On Tue, 2010-10-19 at 09:51 -0700, Jeff Davis wrote:
> On Tue, 2010-10-19 at 12:26 +0300, Heikki Linnakangas wrote:
> > Excluding pg_xlog is just a recommendation at the moment, though, so we
> > would need a big warning in the docs. And some way to enforce that
> > just_kidding is not included in the backup would be nice, maybe we could
> > remove read-permission from it?
>
> Hmm, removing the read bit would add some confidence into the process. I
> like that idea better than just assuming that the user won't copy it.
>
> I think I like this direction most, because it doesn't leave us
> guessing. If the file is there then we assume normal recovery. If we
> encounter recovery.conf we throw FATAL. If we encounter backup_label we
> can simply remove it (perhaps with a warning that there was a crash in
> the middle of a backup).
>
On second thought, this doesn't sound backpatch-friendly. We should
probably put a simpler fix in first and back-patch it. Then we can do
something like your proposal for 9.1. What do you think of my proposed
patch?
Regards,
Jeff Davis