On Thu, 2010-06-03 at 17:56 +0900, Fujii Masao wrote:
> On Thu, Jun 3, 2010 at 4:41 AM, Heikki Linnakangas
> <heikki.linnakangas@enterprisedb.com> wrote:
> > I don't understand why you want to use a different delay when you're
> > restoring from archive vs. when you're streaming (what about existing WAL
> > files found in pg_xlog, BTW?). The source of WAL shouldn't make a
> > difference.
>
> Yes. The pace of a recovery has nothing to do with that of log shipping.
> So to hurry up a recovery when restoring from archive seems to be useless.
When streaming drops for some reason we revert to scanning the archive
for files. There is clearly two modes of operation. So it makes sense
that you might want to set different times for the parameter in each
case.
We might think of those modes as "connected" and "degraded" modes rather
than streaming and file shipping.
-- Simon Riggs www.2ndQuadrant.com