Re: One query run twice in parallel results in huge performance decrease - Mailing list pgsql-performance

From Jeff Janes
Subject Re: One query run twice in parallel results in huge performance decrease
Date
Msg-id CAMkU=1wyY0BCkae12-kn=hAg81TJi2Yz+9WHvPfruzzUvxbazg@mail.gmail.com
Whole thread Raw
In response to One query run twice in parallel results in huge performance decrease  (Jan Michel <Jan@mueschelsoft.de>)
Responses Re: One query run twice in parallel results in huge performance decrease  (Jan Michel <Jan@mueschelsoft.de>)
List pgsql-performance
On Fri, Nov 29, 2013 at 2:07 PM, Jan Michel <Jan@mueschelsoft.de> wrote:

When I run a second instance of this query in a unrelated database on the
same server, they take 100% CPU, no iowait and they do not finish even
after more than a day.

The planner is not aware of what else is going on in the server, so it can't change plans with that in mind.  So I think that that is a red herring.  I'd guess that the 2nd database is missing the geometry index, or has it defined in some different way such that the database doesn't think it can be used.

Are you sure that you get good plans when you run the exact same queries on the exact same database/schema one at a time?

Cheers,

Jeff

pgsql-performance by date:

Previous
From: Jan Michel
Date:
Subject: One query run twice in parallel results in huge performance decrease
Next
From: Jan Michel
Date:
Subject: Re: One query run twice in parallel results in huge performance decrease