On Tue, Jan 17, 2006 at 11:33:02AM +0100, Michael Riess wrote:
>did you read my post? In the first part I explained why I don't want to
>increase the FSM that much.
Since you didn't quantify it, that wasn't much of a data point. (IOW,
you'd generally have to be seriously resource constrained before the FSM
would be a significant source of memory consumption--in which case more
RAM would probably be a much better solution than screwing with
autovacuum.)
Mike Stone
