To: Pavan Deolasee <pavan.deolasee@gmail.com>
cc: =?ISO-8859-1?Q?C=E9dric_Villemain?= <cedric.villemain.debian@gmail.com>,
	Greg Stark <gsstark@mit.edu>, Robert Haas <robertmhaas@gmail.com>,
	Kevin Grittner <Kevin.Grittner@wicourts.gov>,
	Florian Helmberger <fh@25th-floor.com>,
	Alvaro Herrera <alvherre@commandprompt.com>, pgsql-hackers@postgresql.org
Subject: Re: [ADMIN] pg_class reltuples/relpages not updated by
	autovacuum/vacuum
In-reply-to: <BANLkTimTJLzdPDHrVUa=56n_mPtPtO3D3w@mail.gmail.com> 
References: <4DD28F73.6010108@25th-floor.com> <25017.1306291669@sss.pgh.pa.us>
	<4DDCA801.70805@25th-floor.com> <19738.1306338472@sss.pgh.pa.us>
	<BANLkTi=973Arh-nXLO6YFP7_ZTCeNp-X7w@mail.gmail.com>
	<22497.1306341699@sss.pgh.pa.us>
	<BANLkTi=jcKSU5zR+JG2BvfLNCtezC9sb+g@mail.gmail.com>
	<15043.1306423500@sss.pgh.pa.us>
	<BANLkTikrCaoZCVrda6SNCUQFiYvvZX75BA@mail.gmail.com>
	<25097.1306426982@sss.pgh.pa.us>
	<BANLkTikxM6V5DUsXnDEs84z=oRMbyeoOgA@mail.gmail.com>
	<4DDE475B020000250003DD63@gw.wicourts.gov>
	<BANLkTikTbZ8gibj+X++vkdqkxxJCU6trWw@mail.gmail.com>
	<4DDE5033020000250003DD71@gw.wicourts.gov>
	<BANLkTikGu30z1rwYeN0NwuO8jMFWHzus+Q@mail.gmail.com>
	<4DDE7EB8020000250003DDA8@gw.wicourts.gov>
	<9007.1306446600@sss.pgh.pa.us>
	<BANLkTinREie2cX2zqZPSEfUsZsdM+GjymQ@mail.gmail.com>
	<28116.1306609295@sss.pgh.pa.us>
	<BANLkTi=6RYvEVfKn8ufPEf9meGVEu=fHsQ@mail.gmail.com>
	<11295.1306645848@sss.pgh.pa.us>
	<BANLkTimbSiUZC7+ZZJOzMWp=M2jYZZcJSg@mail.gmail.com>
	<19876.1306680027@sss.pgh.pa.us> <BANLkTi=!
	ONrsOoMD3rOdPjRnKPrirKE+fpg@mail.gm{ h@ T
	<BANLkTimTJLzdPDHrVUa=56n_mPtPtO3D3w@mail.gmail.com>
Comments: In-reply-to Pavan Deolasee <pavan.deolasee@gmail.com>
	message dated "Sun, 29 May 2011 21:56:40 +0530"
Date: Sun, 29 May 2011 13:05:53 -0400
Message-ID: <22525.1306688753@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

Pavan Deolasee <pavan.deolasee@gmail.com> writes:
> I am sorry if I sounded terse above. But my gripe is that sometimes we
> are too reluctant to listen to ideas and insist on producing some hard
> numbers first which might take significant efforts. But we are not
> equally strict when such changes are introduced initially.

The reason for not wanting to change it without some actual evidence
is that there is already evidence: the code has been in the field with
this setting since 8.4, and nobody's vacuum performance has fallen off a
cliff.  So while I'd agree that there was little testing done before the
code went in, there is more than zero reason to leave it where it is.
Without some positive evidence showing that another value is better,
I'm disinclined to change it.  I also think that you're not helping
by complaining about the code without being willing to do some work
to try to collect such evidence.

> Heikki suggested 20, Simon proposed 32 to make it a power of 2. But
> why not 16 ? Thats closer to 16 than 32. And Greg yesterday said, 8 is
> a better number based on his testings.

Greg said he had found that the read speed was the same for reading
every page vs reading every 8th page.  That's not the same as concluding
that 8 is the optimal skip distance for vacuum; or at least, he didn't
say that's what he had concluded.  vacuum isn't just reading ...

			regards, tom lane