8.4 seconds is a very long time to spend looking up a single record.
Is this table bloated? What does
vacuum verbose books;
say about it? Look for a line like this:
There were 243 unused item pointers
Thanks but this table books has autovac on, and it's manually
vacuumed every hour!
Thanks Tomas.
The table may still be bloated - the default autovacuum parameters may not
be agressive enough for heavily modified tables.
My autovacuum settings:
autovacuum = on
autovacuum_vacuum_cost_delay = 20
vacuum_cost_delay= 20
autovacuum_naptime
On Fri, Nov 14, 2008 at 9:22 PM, Hoover, Jeffrey [EMAIL PROTECTED] wrote:
There were 2132065 unused item pointers.
Looks to me like a large update or insert failed on this table
Thanks. So what can I do? I have reindexed all indexes already!
--
Sent via pgsql-general mailing list
Thanks Tomas.
The table may still be bloated - the default autovacuum parameters may
not
be agressive enough for heavily modified tables.
My autovacuum settings:
autovacuum = on
autovacuum_vacuum_cost_delay = 20
vacuum_cost_delay= 20
autovacuum_naptime
Thanks Scott. Responses below.
(1) The culprit SELECT sql is (note that MYUSER in this example can
be an IP address) --
So, it can be, but might not be? Darn, If it was always an ip I'd
suggest changing types.
Yes, it can either be a registered USER ID or an IP address. I thought
of
Phoenix Kiula escribió:
Index Scan using new_idx_books_userid on books (cost=0.00..493427.14
rows=2 width=31) (actual time=0.428..8400.299 rows=1 loops=1)
Index Cond:
