Cris,
What version are you running and what are the datatypes of your columns?
Prior to 7.3, postgres wouldn't use an index in the following case:
column = 1
if the datatype of the column "column" wasn't integer. In order to get
it to use the index you needed a cast, so if for example column was of
type int8, you would use:
column = 1::int8
so that both sides of the condition were the same datatype. Then the
index would be used.
thanks,
--Barry
Cris wrote:
> Hi!
> I've have this table:
>
> TABLE BB : There isn't any primary key, and is it more or less order (I mean,
> tt always is increased in each row, and id is nearly ordered)
> ex:
>
> id, op, atr, tt
>
> 1 0 X, 1
> 2 0 A 3
> 3 0 X 5
> ..........
> 1 0 X 51
> .......
> 85 1 l 150
> 86 2 po 155
> 2 0 X 178
> 87 3 1 189
> ....
>
> I VACUUM ANALYZE each 10.000 inserts more or less
> in my case op only can have 3 values;
> I've created an index on (id,op,tt) to improve the next query, that is executed
> very often:
> "SELECT * FROM BB WHERE id="+ id+" AND op=0 order by tt desc;";
> (because the only row I need is the one that has the highest tt)
>
> but, after an hour running the program (more than 90.000 rows), I stopped it and
> "EXPAIN SELECT * FROM BB WHERE id="+ id+" AND op=0 order by tempst desc;";
> But, my sorprise was that the index wasn't be used. Always do a Seq Scan.
>
> Have I to define the index in other columns? Have I to VACUUM ANALYZE more often?
>
> There is other option to avoid the sort that is done because of ORDER BY?
>
> Thanks a lot
>
> yours,
>
> Cris..
