Re: [PERFORM] Poor query plan chosen in 9.0.3 vs 8.3.7

2011-05-05 Thread Tom Lane
Brian Connolly writes: > Any help that you can provide would be greatly appreciated. I'd suggest trying to get rid of the weird little subselects, like this one: > ... SELECT * FROM assayresult."c69d129_particle_size_result_fields" > WHERE (((SELECT Container FROM exp.Data WHERE RowId = DataId

Re: [PERFORM] Poor query plan chosen in 9.0.3 vs 8.3.7

2011-05-05 Thread Bosco Rama
Hey Brian, Brian Connolly wrote: > (I had to send a follow up email due the length of email restrictions on the > mailing list.) A tip for when you have this problem in the future -- turn off html mail. It will reduce your email message length by 50% - 90%. HTH Bosco. -- Sent via pgsql-perfor

Re: [PERFORM] Poor query plan chosen in 9.0.3 vs 8.3.7

2011-05-05 Thread Brian Connolly
Here is the explain analyze output for when enable_material is 'off' and information on the postgresql version, settings and server configuration (I had to send a follow up email due the length of email restrictions on the mailing list.) (I apologize for the length of these email messages. And if

[PERFORM] Poor query plan chosen in 9.0.3 vs 8.3.7

2011-05-05 Thread Brian Connolly
We have a generated query from our web application which takes far longer to complete in 9.0.4, than in 8.3.7 (>60sec in 9.0.4 ~10sec in 8.3.7) The query plan generated in 9.0, includes a Materialize step which takes the bulk of the time for the query. If I disable materialize (by running set ena