Thread:

>There is a faster, non-exclusive-locking VACUUM in the CVS now; this
>should become part of 7.2. You can download the nightly snapsot and build
>it to test it with your application.
>HTH.

OK, so from what I understand if you don't tell it to do a "FULL" vacuum, it will not minimize space by moving tuples
acrssblocks to compact a table.  But does that mean at some point you have to do a "FULL" vacuum? 

Roy.

roypgsqlgen@xemaps.com writes:
> OK, so from what I understand if you don't tell it to do a "FULL"
> vacuum, it will not minimize space by moving tuples acrss blocks to
> compact a table.  But does that mean at some point you have to do a
> "FULL" vacuum?

The theory is that you should reach a steady state where the table file
doesn't get any bigger (assuming that the number of live tuples in the
table isn't getting bigger, that is).

Of course the gap between theory and practice is sometimes large.
We'll see how it goes when 7.2 is in the field...

            regards, tom lane