On Friday 22 July 2005 16:17, Tom Lane pondered:
> Pre-8.0 tends to underestimate the correlation of a multicolumn index.
> (8.0 may too, but not as much.)
I actually upgraded to 8.0.3 now and 2 things have changed. Firstly, I don't
need to do the annoying casts anymore as the query planner now r
On Monday 25 July 2005 15:43, Michael Fuhr pondered:
> Whatever the results of your experiments, could you post the settings
> you tried and the corresponding EXPLAIN ANALYZE outputs?
I did lots of tests now that you pointed me to a useful guide, also taking
what's in the documentation into accou
On Mon, Jul 25, 2005 at 11:23:01AM +0200, Kilian Hagemann wrote:
> shared_buffers, effective_cache_size and cpu_index_tuple_cost all have their
> default values of 1000, 1000 and 0.001 respectively. From their descriptions
> I gather that's reasonable and I don't know how I would optimise these f
On Friday 22 July 2005 15:23, Michael Fuhr pondered:
> Did you run VACUUM ANALYZE or just ANALYZE? Could we see the output
> of VACUUM ANALYZE VERBOSE speed?
I just ran a plain ANALYZE then. When I VACUUM ANALYZE the table the
(inferior) sequential scan strategy is still chosen over the index sc
Kilian Hagemann <[EMAIL PROTECTED]> writes:
>> - what version are you running?
> 7.4.8, not sure if I'm ready for 8 yet.
Pre-8.0 tends to underestimate the correlation of a multicolumn index.
(8.0 may too, but not as much.)
> Also, note that set_id is strictly increasing (hence correlation of 1)
On Fri, Jul 22, 2005 at 10:46:39AM +0200, Kilian Hagemann wrote:
> > - did you run "VACUUM ANALYZE speed" lately?
>
> Yes, just before I ran all of the queries in my last email.
Did you run VACUUM ANALYZE or just ANALYZE? Could we see the output
of VACUUM ANALYZE VERBOSE speed?
> Hence I mentio
Hi there,
Thanks for your and Martijn's comments, I obviously forgot to put in some
vital detail:
> - You never want to set enable_seq off in production database.
That's what I thought...
> - did you run "VACUUM ANALYZE speed" lately?
Yes, just before I ran all of the queries in my last email
Am Donnerstag, 21. Juli 2005 17:45 schrieb Kilian Hagemann:
> Hi there,
>
> I know this subject has come up before many times, but I'm struggling for
> hours with the following problem and none of the posts seem to have a
> solution. I have a table with a good 13 million entries with
>
> station_da
On Thu, Jul 21, 2005 at 05:45:55PM +0200, Kilian Hagemann wrote:
> Hi there,
>
> I know this subject has come up before many times, but I'm struggling for
> hours with the following problem and none of the posts seem to have a
> solution. I have a table with a good 13 million entries with
Firs
