> Weirdly enough, up until a few months ago. Central DB used the index as
> well. When it started to go sequential, we added the phrase "and module
> like '%' " to the WHERE clause, and the index came back into use. Now we
> get no luck at all.
> Given that the index SELECT work in the other 3 databases, I'm at
> the point
> where , at least temporarily, going to have to delete older rows in order
> to improve performance. Even if we don't get indexing back, searching
> 400,000 rows will only take half the time , and 11 second response wil
> ldrop to about 6.
> Oh, we vacuum EVERY night, in case that has any bearing on this. Further,
> the table is updated every day with INSERT of about 3000-5000 new rows.
> Otherwise, this is a read only table. WORM, to be specific.
Have you tried using analyze <table>? It's not entirely clear (to me) that
a plain vacuum does this step.
In response to
pgsql-novice by date
|Next:||From: Manuel Sugawara||Date: 2002-05-15 17:35:16|
|Subject: Re: Catching errors inside transactions|
|Previous:||From: gerry.smit||Date: 2002-05-15 17:23:06|
|Subject: Answering my own question|