Date: Mon, 2 Feb 2015 15:10:14 +0100
From: Andres Freund <andres@2ndquadrant.com>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Peter Eisentraut <peter_e@gmx.net>,
 Michael Paquier <michael.paquier@gmail.com>, Jim Nasby <jim@nasby.net>,
 PostgreSQL mailing lists <pgsql-hackers@postgresql.org>
Subject: Re: REINDEX CONCURRENTLY 2.0
Message-ID: <20150202141014.GA9201@alap3.anarazel.de>
References: 
 <CAB7nPqSTFkuc7dZxCDX4HOTU63YXHRroNv2aoUzyD-Zz_8Z_Zg@mail.gmail.com>
 <52CF45E8.6090208@nasby.net>
 <CAB7nPqTZfx20Ra2Wf4Gm0a=NqF4q2aC8Q6OS8vKN6fhnD=cUxQ@mail.gmail.com>
 <CAB7nPqRkwKFgn4BFUybqU-Oo-=Gcbq0K-8H93Gr6fX-GGRPDXg@mail.gmail.com>
 <20140827064120.GZ21544@awork2.anarazel.de>
 <CAB7nPqSdSMeQMCvtt9P+ndfao5ALpVPDEMmaBPVjNiifLzb=3Q@mail.gmail.com>
 <CAB7nPqSyZhdc-gvwm3KyueoceFJSX5ypQAx_rsJ2DWr5nHtixA@mail.gmail.com>
 <545B8AD2.8030203@gmx.net>
 <CA+TgmoZtd=sa8LD0tnWgRDzx-Zzt_g_FnPf9jtj61z_rbuse6w@mail.gmail.com>
 <CA+TgmoZ=0NMfMVjemYjO0zWFi8PS3OMZCEsqimcgc34U8UKQeg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: 
 <CA+TgmoZ=0NMfMVjemYjO0zWFi8PS3OMZCEsqimcgc34U8UKQeg@mail.gmail.com>
Precedence: bulk
Sender: pgsql-hackers-owner@postgresql.org

On 2014-11-12 16:11:58 -0500, Robert Haas wrote:
> On Wed, Nov 12, 2014 at 4:10 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> > On Thu, Nov 6, 2014 at 9:50 AM, Peter Eisentraut <peter_e@gmx.net> wrote:
> >> If REINDEX cannot work without an exclusive lock, we should invent some
> >> other qualifier, like WITH FEWER LOCKS.
> >
> > What he said.
> 
> But more to the point .... why, precisely, can't this work without an
> AccessExclusiveLock?  And can't we fix that instead of setting for
> something clearly inferior?

So, here's an alternative approach of how to get rid of the AEL
locks. They're required because we want to switch the relfilenodes
around. I've pretty much no confidence in any of the schemes anybody has
come up to avoid that.

So, let's not switch relfilenodes around.

I think if we should instead just use the new index, repoint the
dependencies onto the new oid, and then afterwards, when dropping,
rename the new index one onto the old one. That means the oid of the
index will change and some less than pretty grovelling around
dependencies, but it still seems preferrable to what we're discussing
here otherwise.

Does anybody see a fundamental problem with that approach?

Greetings,

Andres Freund

-- 
 Andres Freund	                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services


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