Tom Lane wrote:
>>Also, the
>>pg_attribute_relid_attnam_index index shrank loads, to 1% of its
>>original size.
>
> Hm, better check whether your vacuuming policy is taking care of the
> system catalogs ...
Yes, fair point. So I did:
18/11/2005 03:02:29 INFO: vacuuming "pg_catalog.pg_attribute"
18/11/2005 03:02:29 INFO: index "pg_attribute_relid_attnam_index" now
contains 2861 row versions in 11900 pages
18/11/2005 03:02:29 DETAIL: 11834 index pages have been deleted, 11834
are currently reusable.
18/11/2005 03:02:29 CPU 0.11s/0.03u sec elapsed 0.45 sec.
18/11/2005 03:02:30 INFO: index "pg_attribute_relid_attnum_index" now
contains 2861 row versions in 2942 pages
18/11/2005 03:02:30 DETAIL: 2917 index pages have been deleted, 2917
are currently reusable.
18/11/2005 03:02:30 CPU 0.01s/0.00u sec elapsed 0.22 sec.
18/11/2005 03:02:30 INFO: "pg_attribute": found 0 removable, 2861
nonremovable row versions in 48 pages
18/11/2005 03:02:30 DETAIL: 0 dead row versions cannot be removed yet.
18/11/2005 03:02:30 There were 19 unused item pointers.
18/11/2005 03:02:30 0 pages are entirely empty.
18/11/2005 03:02:30 CPU 0.12s/0.03u sec elapsed 0.72 sec.
18/11/2005 03:02:30 VACUUM
This is from our vacuum process log. The first index there is the one in
question. Hmm, any clues? Of course everything is running at full speed
currently.
--
David Mitchell
Software Engineer
Telogis