DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=mime-version:in-reply-to:references:date:message-id:subject:from:to
	:cc:content-type:content-transfer-encoding;
	b=YhvHa0SF43bPK51l/Idot6dabmfpNv9qOp3uaiyDL/s/VHsFxmP05WcPdfS9D99bz+
	K3aRL2Sh+Z66ORvJNIIgquRi8l9Ts777lFveAR9ioLwXfPox4+T61NCNqWjZCPyEzTJE
	iPRGLP3LVaVClfY4LV2mAZO8zf4VPNBLLJ/4g=
MIME-Version: 1.0
In-Reply-To: <1306424785-sup-7743@alvh.no-ip.org>
References: <4DDB64CB.7070109@2ndQuadrant.com>
	<1306269968-sup-2890@alvh.no-ip.org>
	<4DDC1DBF.3050306@2ndQuadrant.com>
	<1306346092-sup-2903@alvh.no-ip.org>
	<4DDD6EC3.80703@2ndQuadrant.com>
	<1306424785-sup-7743@alvh.no-ip.org>
Date: Mon, 13 Jun 2011 12:11:48 -0400
Message-ID: <BANLkTi=hh4Ur_EMm_2h6wLOgBD7XrA+xYQ@mail.gmail.com>
Subject: Re: Improve warnings around CREATE INDEX CONCURRENTLY
From: Robert Haas <robertmhaas@gmail.com>
To: Alvaro Herrera <alvherre@commandprompt.com>
Cc: Greg Smith <greg@2ndquadrant.com>, pgsql-docs <pgsql-docs@postgresql.org>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Thu, May 26, 2011 at 11:52 AM, Alvaro Herrera
<alvherre@commandprompt.com> wrote:
> Excerpts from Greg Smith's message of mi=E9 may 25 17:04:03 -0400 2011:
>
>> Sure. =A0Simon's command string idea might work better, and doing some
>> extra lock decoration as you suggested in the above thread would be
>> another level of improvement. =A0We should pick up redesign later on the
>> main list. =A0You can at least count me in as someone who wants to see
>> this improved now.
>
> Great
>
>> Back to the doc patch I submitted...is that a useful step toward making
>> this issue visible enough to users for now to help?
>
> Sure, why not? =A0I thought I could choose my bikeshed color while I was
> here, how about
>
> + =A0 =A0second and third transaction. =A0All active transactions at the =
time the
> + =A0 =A0second table scan starts, not just ones that already involve the=
 table,
> + =A0 =A0have the potential to block the concurrent index creation until =
they
> + =A0 =A0finish. =A0When checking for transactions that
> + =A0 =A0could still use the original index, concurrent index creation ad=
vances
> + =A0 =A0through potentially interfering older transactions one at a time=
,
> + =A0 =A0obtaining shared locks on their virtual transaction identifiers =
to wait for
> + =A0 =A0them to complete.

Alvaro, did you commit this?

--=20
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company