Date: Wed, 8 Apr 2020 20:31:56 -0400
From: Bruce Momjian <bruce@momjian.us>
To: Andres Freund <andres@anarazel.de>
Cc: Robert Haas <robertmhaas@gmail.com>,
	"Jonathan S. Katz" <jkatz@postgresql.org>,
	Alexander Korotkov <a.korotkov@postgrespro.ru>,
	PostgreSQL Hackers <pgsql-hackers@postgresql.org>,
	Peter Geoghegan <pg@bowt.ie>,
	Alvaro Herrera <alvherre@2ndquadrant.com>,
	Thomas Munro <thomas.munro@gmail.com>,
	David Rowley <dgrowleyml@gmail.com>
Subject: Re: Improving connection scalability: GetSnapshotData()
Message-ID: <20200409003156.GA9858@momjian.us>
References: 
 <CAPpHfdthCaCt09Nnj6oCYq93_PUfJSw5QOLGgnid3qJsEWXuGA@mail.gmail.com>
 <20200406133959.viql5fqecog6mppj@alap3.anarazel.de>
 <20200406235228.lnxcau2wgocnnn22@alap3.anarazel.de>
 <20200407121503.zltbpqmdesurflnm@alap3.anarazel.de>
 <20200408124318.amzvnpsqhy747wqp@alap3.anarazel.de>
 <CAPpHfdt78x-aOhbTdY01HJ-ydM9ZSo-sRJNi0pwk7WmWd0xxAA@mail.gmail.com>
 <07c6af6c-c30d-52e4-8716-e3688c3ef265@postgresql.org>
 <CA+TgmobObXx2-JZo_eoRVLgfNF29V=AOP69U=_frcy5+YCPGvA@mail.gmail.com>
 <20200408220623.GA23414@momjian.us>
 <20200408222534.56ys4lt7xd3dvgqu@alap3.anarazel.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200408222534.56ys4lt7xd3dvgqu@alap3.anarazel.de>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Wed, Apr  8, 2020 at 03:25:34PM -0700, Andres Freund wrote:
> Hi,
> 
> On 2020-04-08 18:06:23 -0400, Bruce Momjian wrote:
> > If we don't commit this, where does this leave us with the
> > old_snapshot_threshold feature?  We remove it in back branches and have
> > no working version in PG 13?  That seems kind of bad.
> 
> I don't think this patch changes the situation for
> old_snapshot_threshold in a meaningful way.
> 
> Sure, this patch makes old_snapshot_threshold scale better, and triggers
> fewer unnecessary query cancellations.  But there still are wrong query
> results, the tests still don't test anything meaningful, and the
> determination of which query is cancelled is still wrong.

Oh, OK, so it still needs to be disabled.  I was hoping we could paint
this as a fix.

Based on Robert's analysis of the risk (no SQL syntax, no storage
changes), I think, if you are willing to keep working at this until the
final release, it is reasonable to apply it.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EnterpriseDB                             https://enterprisedb.com

+ As you are, so once was I.  As I am, so you will be. +
+                      Ancient Roman grave inscription +