From: Tom Lane <tgl@sss.pgh.pa.us>
To: Jeff Janes <jeff.janes@gmail.com>
cc: Vitaly Baranovsky <barvetalforums@gmail.com>,
        pgsql-performance@lists.postgresql.org
Subject: Re: PostgreSQL optimizer use seq scan instead of pkey index only scan
 (in queries with postgres_fdw)
In-reply-to: 
 <CAMkU=1xTHcVOpSZU2zbdUv32xwhNOdurQ5tjO9=GXPLmGheGew@mail.gmail.com>
References: 
 <CALWbrdE9pw4+sQjOpa_VFWfLX+A4-kjUrtYEWzxZ7jcX-emRLA@mail.gmail.com>
 <CAMkU=1xTHcVOpSZU2zbdUv32xwhNOdurQ5tjO9=GXPLmGheGew@mail.gmail.com>
Comments: In-reply-to Jeff Janes <jeff.janes@gmail.com>
	message dated "Mon, 06 May 2019 11:32:03 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <16477.1557157999.1@sss.pgh.pa.us>
Date: Mon, 06 May 2019 11:53:19 -0400
Message-ID: <16478.1557157999@sss.pgh.pa.us>
Precedence: bulk

Jeff Janes <jeff.janes@gmail.com> writes:
> It works the way you want in version 12, which is currently under
> development and should be released in 5 months or so.

Even in older versions, the OP would get a significantly smarter
plan after setting use_remote_estimate = on.  I think the core
issue here is that we won't generate remote parameterized paths
without that:

	/*
	 * If we're not using remote estimates, stop here.  We have no way to
	 * estimate whether any join clauses would be worth sending across, so
	 * don't bother building parameterized paths.
	 */
	if (!fpinfo->use_remote_estimate)
		return;

			regards, tom lane