Re: enable_XXX options - Mailing list pgsql-performance

From Tom Lane
Subject Re: enable_XXX options
Date
Msg-id 7068.1112373403@sss.pgh.pa.us
Whole thread Raw
In response to enable_XXX options  ("Peterson, Bjorn" <Bjorn.Peterson@pearson.com>)
List pgsql-performance
"Peterson, Bjorn" <Bjorn.Peterson@pearson.com> writes:
> I have a query in my application that takes an unreasonable amount of time
> to complete (>4.5 hours execution time).   After reviewing the EXPLAIN and
> EXPLAIN ANALYZE output for that and similar queries, my colleagues and I
> determined that turning off the enable_nestloop option might help - we
> noticed dramatic speed improvements for that specific query after doing so
> (<2 minutes execution time).  I was warned not to mess with the enable_XXX
> options in a production environment, but does anyone see any problem with
> turning off the enable_nestloop option right before executing my query and
> turning it back on afterwards?

That's what it's there for ... but it would be useful to look into why
the planner gets it so wrong without that hint.  Could we see EXPLAIN
ANALYZE both ways?

            regards, tom lane

pgsql-performance by date:

Previous
From: "Keith Worthington"
Date:
Subject: Triggers with FOR EACH STATEMENT
Next
From: "Peterson, Bjorn"
Date:
Subject: Re: enable_XXX options