Disk space problem...

Hi All,

Has anyone else run into a situation where the vacuum procedures didn't
really work?

I have maybe 500,000 rows in a database.  When using pg_dumpall, I get a 92MB
sql file.  This database was managing to fill up a 13GB partition, even
though I had a 'vacuumdb -a -z' running every night.

So, I deleted the old data and restored into an empty database freshly
created with initdb, the data only took up 256MB.

By the next morning, disk usage had grown to 650MB.  Vacuuming only seemed to
exacerbate the problem, as after vacuuming it's now 850 MB.  92 of that is in
pg_xlog, which I know will fluctuate, and am assiming will eventually go
away.  However, I just lost another 100 MB!

Several of my tables are being emptied and totally replaced (statistics about
how my website is running, that sort of thing).  Is it better to drop the
table, and rebuild it in this case?

Any other suggestions other than to pg_dumpall, initdb, and restore on a
regular basis?

--
Ken Corey, CTO    Atomic Interactive, Ltd.   ken.corey@atomic-interactive.com