Greg Smith wrote:
> 2) If there is a problem with the network drive, the slow/blocked
> network write is not happening where the database is waiting for it.
This is not really a consideration, is it? I was under the impression
that no other database activity blocks waiting for the archiver. That
doesn't even make sense.
> 3) In an emergency where the usual network drive is unavailable, you can
> temporarily change your archive strategy without touching the database
> server's configuration.
That's just as true with the direct write method. You simply change the
archive script to do nothing and return a failed exit status. The DBMS
will keep trying until you change it back, and then everything catches
up and proceeds as normal.
-Glen