Re: User Interface for WAL usage data

Поиск
Список
Период
Сортировка
От Amit Kapila
Тема Re: User Interface for WAL usage data
Дата
Msg-id CAA4eK1JMvwBSvnPJTLipr4eKx99axxXzLjoH04q85NCFhndhyQ@mail.gmail.com
обсуждение исходный текст
Ответ на Re: User Interface for WAL usage data  (Justin Pryzby <pryzby@telsasoft.com>)
Ответы Re: User Interface for WAL usage data
Список pgsql-hackers
On Thu, Apr 2, 2020 at 12:05 PM Justin Pryzby <pryzby@telsasoft.com> wrote:
>
> On Thu, Apr 02, 2020 at 11:32:16AM +0530, Amit Kapila wrote:
> > On Thu, Apr 2, 2020 at 11:28 AM Kyotaro Horiguchi <horikyota.ntt@gmail.com> wrote:
> > >
> > > At Thu, 2 Apr 2020 00:41:20 -0500, Justin Pryzby <pryzby@telsasoft.com> wrote in
> > > > Regarding v10-0004-Add-option-to-report-WAL-usage-in-EXPLAIN-and-au.patch:
> > > > I think there should be additional spaces before "full" and before "bytes":
> > > >
> > > > >    WAL: records=2359 full page records=42 bytes=447788
> > > >
> > > > Compare with these:
> > > >
> > > >        "Sort Method: %s  %s: %ldkB\n",
> > > >        "Buckets: %d (originally %d)  Batches: %d (originally %d)  Memory Usage: %ldkB\n",
> > > >        "Buckets: %d  Batches: %d  Memory Usage: %ldkB\n",
> > > >
> > > > Otherwise "records=2359 full page records=42" is hard to parse.
> > >
> > > I got the same feeling seeing the line.
> >
> > But isn't this same as we have BufferUsage data?  We can probably
> > display it as full_page_writes or something like that.
>
> I guess you mean this:
>          Buffers: shared hit=994 read=11426 dirtied=466
>
> Which can show shared/local/temp.  Actually I would probably make the same
> suggestion for "Buffers" (if it were a new patch).  I would find this to be
> pretty unfriendly output:
>
>          Buffers: shared hit=12345 read=12345 dirtied=12345 local hit=12345 read=12345 dirtied=12345 temp hit=12345
read=12345dirtied=12345
 
>
> Adding two extra spaces "  local" and "  temp" would have helped there, so
> would commas, or parenthesis, dashes or almost anything - other than a
> backslash.
>
> So I think you're right that WAL is very similar to the Buffers case, but I
> suggest that's not a good example to follow, especially since you're adding a
> "field" with spaces in it.
>

Agreed, this is a good reason to keep two spaces as we have in cases
of Buckets.  We can display the variable as "full page writes".

-- 
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com



В списке pgsql-hackers по дате отправления:

Предыдущее
От: Thomas Munro
Дата:
Сообщение: Re: Should we add xid_current() or a int8->xid cast?
Следующее
От: Amit Kapila
Дата:
Сообщение: Re: WAL usage calculation patch