Re: [PERFORM] Forcing index usage without 'enable_hashjoin = FALSE'

2006-08-23 Thread Dan Langille
On 23 Aug 2006 at 22:30, Tom Lane wrote: > "Dan Langille" <[EMAIL PROTECTED]> writes: > > Without leaving "enable_hashjoin = false", can you suggest a way to > > force the index usage? > > Have you tried reducing random_page_cost? Yes. No effect. > FYI, 8.2 should be a bit better about this.

Re: [PERFORM] Forcing index usage without 'enable_hashjoin = FALSE'

2006-08-23 Thread Tom Lane
"Dan Langille" <[EMAIL PROTECTED]> writes: > Without leaving "enable_hashjoin = false", can you suggest a way to > force the index usage? Have you tried reducing random_page_cost? FYI, 8.2 should be a bit better about this. regards, tom lane ---(

Re: [PERFORM] Forcing index usage without 'enable_hashjoin = FALSE'

2006-08-23 Thread Dan Langille
On 23 Aug 2006 at 13:31, Chris wrote: > Dan Langille wrote: > > I'm using PostgreSQL 8.1.4 and I'm trying to force the planner to use > > an index. With the index, I get executions times of 0.5 seconds. > > Without, it's closer to 2.5 seconds. > > > > Compare these two sets of results (also pro

Re: [PERFORM] Forcing index usage without 'enable_hashjoin = FALSE'

2006-08-22 Thread Chris
Dan Langille wrote: I'm using PostgreSQL 8.1.4 and I'm trying to force the planner to use an index. With the index, I get executions times of 0.5 seconds. Without, it's closer to 2.5 seconds. Compare these two sets of results (also provided at http://rafb.net/paste/results/ywcOZP66.html sho