Re: Fast promotion, loose ends

On 24 April 2013 09:53, Heikki Linnakangas <hlinnakangas@vmware.com> wrote:
> On 24.04.2013 11:46, Simon Riggs wrote:
>>
>> On 24 April 2013 09:32, Heikki Linnakangas<hlinnakangas@vmware.com>
>> wrote:
>>
>>>> pg_ctl already checks versions, so I don't see the point.
>>>
>>>
>>>
>>> The point is, if you do "pgsql93/bin/pg_ctl -D $92DATADIR promote", it
>>> will
>>> create fast_promote file and return success. But it won't actually
>>> promote
>>> the server. I think that's bad.
>>>
>>> If pg_ctl already has a check against that, fine, but I don't think it
>>> does.
>>> Please make sure you test that before applying.
>>
>>
>> If it doesn't check, that is not the only thing that would be broken.
>> The original commit of pg_ctl promote would also be broken.
>
>
> Yeah, it would've been good if the "pg_ctl promote" patch would've added a
> version check. Nevertheless, don't you think it would be good to avoid
> changing the filename of the "promote" file, so that we don't have any more
> such breakage? I don't see any advantage in changing it.

Apart from all the reasons I already gave, no.

The filename isn't changing. We are adding a new capability and
changing the default.

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