Okay, I have some more info.
Some background info. This one table gets so many changes, I CLUSTER it
each night. However, after I do this. The statistics still appear to be
incorrect. Even after I do a "select pg_stat_reset();" Followed by 3
ANALYZE at default_statistics_target as 1, 10, and 10
General: +1 612 367 6224
On Tue, Feb 5, 2013 at 4:12 PM, Kevin Grittner wrote:
> James R Skaggs wrote:
>
> > Agreed, we shouldn't have so many dead rows. Our autovacuum is
> > set on but has default parameters. We are clustering today.
> > This, of course, means
Kevin -
Agreed, we shouldn't have so many dead rows. Our autovacuum is set on but
has default parameters. We are clustering today. This, of course, means
downtime and inconvenience to the users.
Here is the troublesome query:
select
sum(case when t1.cdlc < 0 then 0 else t1.cdlc end) as
