From: Tom Lane <tgl@sss.pgh.pa.us>
To: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
cc: jesper.pedersen@redhat.com, Robert Haas <robertmhaas@gmail.com>,
        "Jamison, Kirk" <k.jamison@jp.fujitsu.com>,
        Alvaro Herrera <alvherre@2ndquadrant.com>,
        Michael Paquier <michael@paquier.xyz>,
        "fabriziomello@gmail.com" <fabriziomello@gmail.com>,
        "pgsql-hackers@lists.postgresql.org"
 <pgsql-hackers@lists.postgresql.org>,
        Bruce Momjian <bruce@momjian.us>
Subject: Re: pg_upgrade: Pass -j down to vacuumdb
In-reply-to: <2c054085-0b68-c049-d6d8-8b74852b4c73@2ndquadrant.com>
References: <201902010958.cf2cbi3d2ule@alvherre.pgsql>
 <bef52417-2416-cb97-afce-a351601eab43@redhat.com>
 <D09B13F772D2274BB348A310EE3027C64659E9@g01jpexmbkw24>
 <CA+TgmobMGTfkMY-isz4pLw_bfywVOqTUDcW+obeAfYSeTBJM7A@mail.gmail.com>
 <03845f77-49e7-4569-7062-8ee1fb460ec2@redhat.com>
 <30470.1553551070@sss.pgh.pa.us>
 <2c054085-0b68-c049-d6d8-8b74852b4c73@2ndquadrant.com>
Comments: In-reply-to Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
	message dated "Tue, 26 Mar 2019 12:28:34 +0100"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <15728.1553605759.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Tue, 26 Mar 2019 09:09:19 -0400
Message-ID: <15729.1553605759@sss.pgh.pa.us>
Precedence: bulk

Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
> On 2019-03-25 22:57, Tom Lane wrote:
>> +	fprintf(script, "echo %sYou may wish to add --jobs=3DN for parallel a=
nalyzing.%s\n",
>> +			ECHO_QUOTE, ECHO_QUOTE);

> But then you get that information after you have already started the scr=
ipt.

Yes, but that's already true of all the info that this script prints out.

> I don't find any information about this analyze business on the
> pg_upgrade reference page.  Maybe a discussion there could explain the
> different paths better than making the output script extra complicated.

> Essentially: If you want a slow and gentle analyze, use the supplied
> script.  If you want a fast analyze, use vacuumdb, perhaps with an
> appropriate --jobs option.  Note that pg_upgrade --jobs and vacuumdb
> --jobs are resource-bound in different ways, so the same value might not
> be appropriate for both.

I have no objection to handling it that way (i.e., just as a doc fix).
Or we could do both.

			regards, tom lane