Обсуждение: Roles for pg_basebackup

Based on a recent conversation about backups I had I propose a small tweak to
the pg_basebackup documentation.  Listing the user types in the reverse order
from today, putting superuser last, makes it IMO a little clearer that a
REPLICATION role is preferrable to using a superuser for running backups.

cheers ./daniel

On Mon, Apr 27, 2020 at 12:16:41PM +0200, Daniel Gustafsson wrote:
> Based on a recent conversation about backups I had I propose a small tweak to
> the pg_basebackup documentation.  Listing the user types in the reverse order
> from today, putting superuser last, makes it IMO a little clearer that a
> REPLICATION role is preferrable to using a superuser for running backups.

Makes sense to me.  We do that in logical-replication.sgml as well as
pg_rewind.sgml (the latter outlines superuser rights last).

Any objections from others?
--
Michael

On 2020/04/28 13:37, Michael Paquier wrote:
> On Mon, Apr 27, 2020 at 12:16:41PM +0200, Daniel Gustafsson wrote:
>> Based on a recent conversation about backups I had I propose a small tweak to
>> the pg_basebackup documentation.  Listing the user types in the reverse order
>> from today, putting superuser last, makes it IMO a little clearer that a
>> REPLICATION role is preferrable to using a superuser for running backups.
> 
> Makes sense to me.  We do that in logical-replication.sgml as well as
> pg_rewind.sgml (the latter outlines superuser rights last).

Seems there are other documentations having the similar description,
for example, pg_receivewal.sgml, func.sgml and high-availability.sgml.
Isn't it better to update also them at the same time?

> Any objections from others?

No.

Regards,

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

> On 28 Apr 2020, at 07:22, Fujii Masao <masao.fujii@oss.nttdata.com> wrote:
>
> On 2020/04/28 13:37, Michael Paquier wrote:
>> On Mon, Apr 27, 2020 at 12:16:41PM +0200, Daniel Gustafsson wrote:
>>> Based on a recent conversation about backups I had I propose a small tweak to
>>> the pg_basebackup documentation.  Listing the user types in the reverse order
>>> from today, putting superuser last, makes it IMO a little clearer that a
>>> REPLICATION role is preferrable to using a superuser for running backups.
>> Makes sense to me.  We do that in logical-replication.sgml as well as
>> pg_rewind.sgml (the latter outlines superuser rights last).
>
> Seems there are other documentations having the similar description,
> for example, pg_receivewal.sgml, func.sgml and high-availability.sgml.
> Isn't it better to update also them at the same time?

That doe make sense.  I've updated the places where I think a user configuring
the system might conceivably infer a preference from the order, but have left
the other ones as is.

cheers ./daniel

On 2020/04/28 17:15, Daniel Gustafsson wrote:
>> On 28 Apr 2020, at 07:22, Fujii Masao <masao.fujii@oss.nttdata.com> wrote:
>>
>> On 2020/04/28 13:37, Michael Paquier wrote:
>>> On Mon, Apr 27, 2020 at 12:16:41PM +0200, Daniel Gustafsson wrote:
>>>> Based on a recent conversation about backups I had I propose a small tweak to
>>>> the pg_basebackup documentation.  Listing the user types in the reverse order
>>>> from today, putting superuser last, makes it IMO a little clearer that a
>>>> REPLICATION role is preferrable to using a superuser for running backups.
>>> Makes sense to me.  We do that in logical-replication.sgml as well as
>>> pg_rewind.sgml (the latter outlines superuser rights last).
>>
>> Seems there are other documentations having the similar description,
>> for example, pg_receivewal.sgml, func.sgml and high-availability.sgml.
>> Isn't it better to update also them at the same time?
> 
> That doe make sense.  I've updated the places where I think a user configuring
> the system might conceivably infer a preference from the order, but have left
> the other ones as is.

Thanks for updating the patch! It looks good to me.

Regards,

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

On Fri, May 01, 2020 at 12:29:09PM +0900, Fujii Masao wrote:
> On 2020/04/28 17:15, Daniel Gustafsson wrote:
>> That does make sense.  I've updated the places where I think a user configuring
>> the system might conceivably infer a preference from the order, but have left
>> the other ones as is.

Yeah, I spotted a couple of places in adminpack.sgml, dropuser.sgml,
catalogs.sgml and config.sgml, but the intention looks different.  We
could always discuss more on that later on.

> Thanks for updating the patch! It looks good to me.

Fine for me as well as far as it goes.  I'll go apply it if there are
no objections.
--
Michael

On Sun, May 03, 2020 at 03:53:31PM +0900, Michael Paquier wrote:
> Fine for me as well as far as it goes.  I'll go apply it if there are
> no objections.

And done.
--
Michael