Re: Remove non-fast promotion Re: Should we remove a fallback promotion? take 2

On 2020/04/22 10:53, Kyotaro Horiguchi wrote:
> At Wed, 22 Apr 2020 10:28:07 +0900, Ian Barwick <ian(dot)barwick(at)2ndquadrant(dot)com> wrote in
>> On 2020/04/22 6:53, Alvaro Herrera wrote:
>>> On 2020-Apr-21, Jehan-Guillaume de Rorthais wrote:
>>>
>>>> On Tue, 21 Apr 2020 15:36:22 +0900
>>>> Michael Paquier <michael(at)paquier(dot)xyz> wrote:
>>>
>>>>>> Also in that case, non-fast promotion is triggered. Since my patch
>>>>>> tries to remove non-fast promotion, it's intentional to prevent them
>>>>>> from doing that. But you think that we should not drop that because
>>>>>> there are still some users for that?
>>>>>
>>>>> It would be good to ask around to folks maintaining HA solutions about
>>>>> that change at least, as there could be a point in still letting
>>>>> promotion to happen in this case, but switch silently to the fast
>>>>> path.
>>>>
>>>> FWIW, PAF relies on pg_ctl promote. No need for non-fast promotion.
>>> AFAICT repmgr uses 'pg_ctl promote', and has since version 3.0
>>> (released
>>> in mid 2015). It was only 3.3.2 (mid 2017) that supported Postgres
>>> 10,
>>> so it seems fairly safe to assume that the removal won't be a problem.
>>
>> Correct, repmgr uses "pg_ctl promote" or pg_promote() (if available),
>> and
>> won't be affected by this change.
>
> For the record, the pgsql resource agent uses "pg_ctl promote" and
> working with fast-promote. Auxiliary tools for it is assuming
> fast-promote.

Thanks all for checking whether the change affects each HA solution!

Regards,

--
Fujii Masao
Advanced Computing Technology Center
Research and Development Headquarters
NTT DATA CORPORATION