On Fri, Jun 11, 2021 at 03:32:28PM +0900, Kyotaro Horiguchi wrote:
> At Fri, 11 Jun 2021 10:48:32 +0800, Julien Rouhaud <rjuju123@gmail.com> wrote in
> >
> > "far from perfect" is a strong understatement for "appears to work but will
> > randomly and silently breaks everything without a simple way to detect it".
>
> I think it's overstating. It sounds like a story of a mission critical
> world. How perfect archive_command should be depends on the
> requirements of every system. Simple cp is actualy sufficient in
> certain log? range of usages, maybe.
I disagree, cp is probably the worst command that can be used for this purpose.
On top on the previous problems already mentioned, you also have the fact that
the copy isn't atomic. It means that any concurrent restore_command (or
anything that would consume the archived files) will happily process a half
copied WAL file, and in case of any error during the copy you end up with a
file for which you don't know if it contains valid data or not. I don't see
any case where you would actually want to use that, unless maybe if you want to
benchmark how long it takes before you lose some data.
> > We should document a minimum workable setup, but cp isn't an example of that,
> > and I don't think that there will be a simple example unless we provide a
> > dedicated utility.
>
> It looks somewhat strange like "Well, you need a special track to
> drive your car, however, we don't have one. It's your responsibility
> to construct a track that protects it from accidents perfectly.".
>
> "Yeah, I'm not going to push it so hard and don't care it gets some
> small scratches, couldn't I drive it on a public road?"
>
> (Sorry for the bad analogy).
I think that a better analogy would be "I don't need working brakes on my car
since I only drive on highway and there aren't any red light there".
> Isn't it almost saying that anything less than pgBackRest isn't
> qualified as archive_program?
I don't know, I'm assuming that barman also provides one, such as wal-e and
wal-g (assuming that the distant providers do their part of the job correctly).
Maybe there are other tools too. But as long as we don't document what exactly
are the requirements, it's not really a surprise that most people don't
implement them.