Re: [COMMITTERS] pgsql: New SQL functons pg_backup_in_progress() and pg_backup_start_tim

On 15.06.2012 17:54, Magnus Hagander wrote:
> On Fri, Jun 15, 2012 at 8:16 PM, Robert Haas<robertmhaas(at)gmail(dot)com> wrote:
>> On Fri, Jun 15, 2012 at 1:52 AM, Magnus Hagander<magnus(at)hagander(dot)net> wrote:
>>> On Fri, Jun 15, 2012 at 1:29 AM, Robert Haas<rhaas(at)postgresql(dot)org> wrote:
>>>> New SQL functons pg_backup_in_progress() and pg_backup_start_time()
>>>>
>>>> Darold Gilles, reviewed by Gabriele Bartolini and others, rebased by
>>>> Marco Nenciarini. Stylistic cleanup and OID fixes by me.
>>>
>>> How well is the term "on-line exclusive backup" really settled with
>>> people? I wonder if we need to add a specific note to the docs saying
>>> that the function doesn't consider streaming base backups at all, and
>>> that one should refer to pg_stat_replication for info about those? Or
>>> really, should the function be pg_exclusive_backup_in_progress()
>>> perhaps?
>>
>> Well, if we think that the term "exclusive backup" is not going to be
>> easily comprehensible, then sticking that into the function name isn't
>> going to help us much. I think that's just wordiness for the sake of
>> being wordy. I do agree that we could probably improve the clarity of
>> the documentation along the lines you suggest.
>
> It would alert people to the existance of the term, and thus help
> those who didn't actually read the documentation.

I'm not sure we want to expose the "exclusive backup" term to users.
It's a bit confusing. It makes sense in the limited scope in the code in
xlog.c where it's currently used, but if I wanted to explain what it is
to users, I don't think I'd choose that term.

> Which actually makes an argument for making that change *anyway*,
> because right now the function is incorrectly named. A function named
> pg_backup_in_progress() should answer the question "is a backup in
> progress". And it doesn't answer that question.

I agree that pg_backup_in_progress() is confusing, if it returns false
while you're running pg_basebackup. In the doc changes you proposed, you
call the pg_start/stop_backup() a "low level API" for taking backups.
That's not suitable for a function name, but I think we should work on
that, and find a better term that works.

Backup mode? Filesystem backup mode?

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com