Shawn <postgres@xmtservices.net> writes:
> "Kevin Grittner" <Kevin.Grittner@wicourts.gov> wrote:
>> Did you ever capture the output of VACUUM VERBOSE against this table
> vacuum verbose analyze shawns_data;
> ...
> INFO: index "shawns_data_pkey" now contains 15445 row versions in
> 35230 pages
[ and comparably bloated sizes for other indexes ]
Ouch! The table itself doesn't look nearly as bad:
> INFO: "shawns_data": found 21444 removable, 15445 nonremovable row
> versions in 770 pages
but you've got a spectacularly bad case of index bloat. An index 50
times bigger than its table is Not Good. I think you'd find that
"REINDEX TABLE shawns_data" does wonders for the situation.
The next question is how it got to be that way... what is your
vacuuming policy for this database? Maybe you need to raise
max_fsm_pages?
regards, tom lane