From: Tom Lane <tgl@sss.pgh.pa.us>
To: Ben Hoyt <benhoyt@gmail.com>
cc: David Johnston <polobo@yahoo.com>, pgsql-performance@postgresql.org
Subject: Re: Re: Adding foreign key constraint holds exclusive lock for too
 long (on production database)
In-reply-to: 
 <CAL9jXCGNjjSNqhnqKQoSdWA8G=8ohaDJrpDPBUoL4=gz9mLM_w@mail.gmail.com>
References: 
 <CAL9jXCFwKXt67+fSGq2N4NDMEadAe=DtKdUGU2gFSS0t-crFxA@mail.gmail.com>
 <1383092289576-5776315.post@n5.nabble.com> <13822.1383093005@sss.pgh.pa.us>
 <CAL9jXCGNjjSNqhnqKQoSdWA8G=8ohaDJrpDPBUoL4=gz9mLM_w@mail.gmail.com>
Comments: In-reply-to Ben Hoyt <benhoyt@gmail.com>
 message dated "Thu, 31 Oct 2013 21:01:40 +1300"
Date: Thu, 31 Oct 2013 10:19:48 -0400
Message-ID: <22670.1383229188@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

Ben Hoyt <benhoyt@gmail.com> writes:
>> It appears the possible explanations for Ben's problem are:
>> 1. For some reason this query is a lot slower than the one he came up
>> with;

> Anyway, it's definitely #1 that's happening, as I build the
> RI_Initial_Check() query by hand, and it takes just as long as the ADD
> CONSTRAINT.

Huh.  Maybe an optimizer failing?  Could we see the full text of both
queries and EXPLAIN ANALYZE results for them?

> So I guess this is really a side effect of the quirky way we're
> dumping and restoring only one schema, and dropping/re-adding
> constraints on deployment because of this. Is this a really strange
> thing to do -- deploying only one schema (the "static" data) and
> dropping/re-adding constraints -- or are there better practices here?

Doesn't seem unreasonable.  One thought is that maybe you need to insert a
manual ANALYZE after reloading the data?

			regards, tom lane


-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance