Message-ID: <550593D0.3020106@2ndquadrant.com>
Date: Sun, 15 Mar 2015 15:14:40 +0100
From: Petr Jelinek <petr@2ndquadrant.com>
User-Agent: Mozilla/5.0 (X11; Linux x86_64;
 rv:31.0) Gecko/20100101 Thunderbird/31.5.0
MIME-Version: 1.0
To: Magnus Hagander <magnus@hagander.net>,
 Andres Freund <andres@2ndquadrant.com>
CC: PostgreSQL-development <pgsql-hackers@postgresql.org>,
 Simon Riggs <simon@2ndquadrant.com>
Subject: Re: recovery_target_action = pause & hot_standby = off
References: <20150312145202.GD20199@awork2.anarazel.de>
	<20150315132707.GB19792@alap3.anarazel.de>
 <CABUevEzZaXqMzTryrEo1Z8dmQgbmL0bh6tVHvSQqyTqfmn9Wkw@mail.gmail.com>
In-Reply-To: 
 <CABUevEzZaXqMzTryrEo1Z8dmQgbmL0bh6tVHvSQqyTqfmn9Wkw@mail.gmail.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk
Sender: pgsql-hackers-owner@postgresql.org

On 15/03/15 14:51, Magnus Hagander wrote:
> On Sun, Mar 15, 2015 at 2:27 PM, Andres Freund <andres@2ndquadrant.com
> <mailto:andres@2ndquadrant.com>> wrote:
>
>     On 2015-03-12 15:52:02 +0100, Andres Freund wrote:
>      >       /*
>      >        * Override any inconsistent requests. Not that this is a
>     change
>      >        * of behaviour in 9.5; prior to this we simply ignored a
>     request
>      >        * to pause if hot_standby = off, which was surprising
>     behaviour.
>      >        */
>      >       if (recoveryTargetAction == RECOVERY_TARGET_ACTION_PAUSE &&
>      >               recoveryTargetActionSet &&
>      >               standbyState == STANDBY_DISABLED)
>      >               recoveryTargetAction = RECOVERY_TARGET_ACTION_SHUTDOWN;
>
>     While it's easy enough to fix I rather dislike the whole intent here
>     though. *Silently* switching the mode of operation in a rather
>     significant way seems like a bad idea to me. At the very least we need
>     to emit a LOG message about this; but I think it'd be much better to
>     error out instead.
>
>     <9.5's behaviour was already quite surprising. But changing things to a
>     different surprising behaviour seems like a bad idea.
>
>
> +1. Especially for "sensitive" operations like this, having
> predictable-behavior-or-error is usually the best choice.
>

Thinking about it again now, it does seem that ignoring user setting 
because it's in conflict with another user setting is a bad idea and I 
think we in general throw errors on those.

So +1 from me also.

-- 
  Petr Jelinek                  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