On Fri, 22 Aug 2003 10:45:50 -0400, Tom Lane <[EMAIL PROTECTED]>
wrote:
>One big question mark in my mind about these "partial vacuum"
>proposals is whether they'd still allow adequate FSM information to be
>maintained. If VACUUM isn't looking at most of the pages, there's no
>very good way to acquire info about where there's free space.
VACUUM has accurate information about the pages it just visited. Free
space information for pages not touched by VACUUM is still in the FSM,
unless free space on a page is too low to be interesting. VACUUM has
to merge these two lists and throw away entries with little free space
if running out of room.
Thus we might end up with new almost full pages in the FSM while there
are pages with more free space lying around that a previous VACUUM
failed to register because there was more free space at that time.
Considering that
. FSM is lossy per definitionem
. we are targeting at relations with large passive areas
. decent VACUUM shall not replace lazy VACUUM
I see no problem here.
Future advice could be: "VACCUM DECENT every hour, VACUUM daily,
VACUUM FULL once a year" where the first two could be scheduled by
autovacuum ...
Servus
Manfred
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match
