Re: [PATCH] Initial progress reporting for COPY command

On 2020/06/22 17:21, Josef Šimánek wrote:
>
>
> po 22. 6. 2020 v 4:48 odesílatel Fujii Masao <masao(dot)fujii(at)oss(dot)nttdata(dot)com <mailto:masao(dot)fujii(at)oss(dot)nttdata(dot)com>> napsal:
>
>
>
> On 2020/06/21 20:33, Josef Šimánek wrote:
> >
> >
> > po 15. 6. 2020 v 6:39 odesílatel Fujii Masao <masao(dot)fujii(at)oss(dot)nttdata(dot)com <mailto:masao(dot)fujii(at)oss(dot)nttdata(dot)com> <mailto:masao(dot)fujii(at)oss(dot)nttdata(dot)com <mailto:masao(dot)fujii(at)oss(dot)nttdata(dot)com>>> napsal:
> >
> >
> >
> >     On 2020/06/14 21:32, Josef Šimánek wrote:
> >      > Hello, as proposed by Pavel Stěhule and discussed on local czech PostgreSQL maillist (https://groups.google.com/d/msgid/postgresql-cz/CAFj8pRCZ42CBCa1bPHr7htffSV%2BNAcgcHHG0dVqOog4bsu2LFw%40mail.gmail.com?utm_medium=email&utm_source=footer), I have prepared an initial patch for COPY command progress reporting.
> >
> >     Sounds nice!
> >
> >
> >      > file - bool - is file is used?
> >      > program - bool - is program used?
> >
> >     Are these fields really necessary in a progress view?
> >     What values are reported when STDOUT/STDIN is specified in COPY command?
> >
> >
> > For STDOUT and STDIN file is true and program is false.
>
> Could you tell me why these columns are necessary in *progress* view?
> If we want to see what copy command is actually running, we can see
> pg_stat_activity, instead. For example,
>
>      SELECT pc.*, a.query FROM pg_stat_progress_copy pc, pg_stat_activity a WHERE pc.pid = a.pid;
>
> If that doesn't make any sense, I can remove those. I have not strong opinion about those values. Those were just around when I was looking for possible values to include in the progress report.

I vote not to expose them. *If* we expose them, we should also
expose the options in pg_stat_progress_xxx views, for example,
the options for BASE_BACKUP command in pg_stat_progress_basebackup,
for the consistency. But I don't think that makes sense.

>
> >
> >      > file_bytes_processed - amount of bytes processed when file is used (otherwise 0), works for both direction (
> >      > FROM/TO) when file is used (file = t)
> >
> >     What value is reported when STDOUT/STDIN is specified in COPY command?
> >
> >
> > For my first patch nothing was reported on STDOUT/STDIN usage. I'll attach new patch soon supporting those as well.
>
> Thanks for the patch!
>
> With the patch, pg_stat_progress_copy seems to report the progress of
> the processing on file_fdw. Is this intentional?
>
>
> Every action using internally COPY will be included in the progress report view.
> I have spotted for example pg_dump does that and is reported there as well.
> I do not see any problem regarding this. For pg_dump it is consistent with "pg_stat_activity" reporting COPY command in the query field.

So it's better to add this kind of information into the docs?

Regards,

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