Re: Large Database Restore

On 5/17/07, Ron Johnson <ron(dot)l(dot)johnson(at)cox(dot)net> wrote:
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 05/17/07 16:49, Michael Nolan wrote:
> > I don't know if my database is typical (as there probably is no such
> > thing), but to restore a full dump (pg_dumpall) takes over 4 hours on my
> > backup server, but to restore a low level backup (about 35GB)
>
> Meaning a tarball of $PGDATA?
>
> Actually, it's two different areas because I've got a second tablespace on
a separate physical drive for indexes, but yes, it's a tarball of all the
files that PG uses, following the procedures in section 23.3 of the
documentation.

It works very well, though I still don't understand why, if there are no
changes to the warm standby server tables, only queries, it isn't possible
to keep restoring WAL files to keep the warm standby server in parallel with
the live server. (I'm guessing there must be SOMETHING that happens at the
end of the recovery process, or some time after that, to make the next WAL
unprocessable., but I can't figure it out from the docs.)
--
Mike Nolan