Thread: Doc: fix track_io_timing description to mention pg_stat_io

Hi,

pg_stat_io has I/O statistics that are collected when track_io_timing is

enabled, but it is not mentioned in the description of track_io_timing.

I think it's better to add a description of pg_stat_io for easy reference. 

What do you think?

Its always good to add new things. 

On Thu, Jun 27, 2024 at 5:06 AM <Hajime.Matsunaga@nttdata.com> wrote:
>
> Hi,
>
> pg_stat_io has I/O statistics that are collected when track_io_timing is
> enabled, but it is not mentioned in the description of track_io_timing.
> I think it's better to add a description of pg_stat_io for easy reference.
> What do you think?

Seems quite reasonable to me given that track_wal_io_timing mentions
pg_stat_wal. I noticed that the sentence about track_io_timing in the
statistics collection configuration section [1] only mentions reads
and writes -- perhaps it should also mention extends and fsyncs?

- Melanie

[1] https://www.postgresql.org/docs/devel/monitoring-stats.html#MONITORING-STATS-SETUP

Hi,

On Thu, 27 Jun 2024 at 14:30, Melanie Plageman
<melanieplageman@gmail.com> wrote:
>
> On Thu, Jun 27, 2024 at 5:06 AM <Hajime.Matsunaga@nttdata.com> wrote:
> >
> > Hi,
> >
> > pg_stat_io has I/O statistics that are collected when track_io_timing is
> > enabled, but it is not mentioned in the description of track_io_timing.
> > I think it's better to add a description of pg_stat_io for easy reference.
> > What do you think?
>
> Seems quite reasonable to me given that track_wal_io_timing mentions
> pg_stat_wal. I noticed that the sentence about track_io_timing in the
> statistics collection configuration section [1] only mentions reads
> and writes -- perhaps it should also mention extends and fsyncs?

Both suggestions look good to me. If what you said will be
implemented, maybe track_wal_io_timing too should mention fsyncs?

--
Regards,
Nazir Bilal Yavuz
Microsoft

> > On Thu, Jun 27, 2024 at 5:06 AM <Hajime.Matsunaga@nttdata.com> wrote:
> > >
> > > Hi,
> >
> > > pg_stat_io has I/O statistics that are collected when track_io_timing is
> > > enabled, but it is not mentioned in the description of track_io_timing.
> > > I think it's better to add a description of pg_stat_io for easy reference.
> > > What do you think?
> >
> > Seems quite reasonable to me given that track_wal_io_timing mentions
> > pg_stat_wal. I noticed that the sentence about track_io_timing in the
> > statistics collection configuration section [1] only mentions reads
> > and writes -- perhaps it should also mention extends and fsyncs?
>
> Both suggestions look good to me. If what you said will be
> implemented, maybe track_wal_io_timing too should mention fsyncs?

Thank you both for your positive comments.
I also think your suggestions seem good.


Regards,
--
Hajime Matsunaga
NTT DATA Group Corporation

From: Nazir Bilal Yavuz <byavuz81@gmail.com>
Sent: Thursday, June 27, 2024 9:01 PM
>
> Hi,
>
> On Thu, 27 Jun 2024 at 14:30, Melanie Plageman
> <melanieplageman@gmail.com> wrote:
> >
> > On Thu, Jun 27, 2024 at 5:06 AM <Hajime.Matsunaga@nttdata.com> wrote:
> > >
> > > Hi,
> > >
> > > pg_stat_io has I/O statistics that are collected when track_io_timing is
> > > enabled, but it is not mentioned in the description of track_io_timing.
> > > I think it's better to add a description of pg_stat_io for easy reference.
> > > What do you think?
> >
> > Seems quite reasonable to me given that track_wal_io_timing mentions
> > pg_stat_wal. I noticed that the sentence about track_io_timing in the
> > statistics collection configuration section [1] only mentions reads
> > and writes -- perhaps it should also mention extends and fsyncs?
>
> Both suggestions look good to me. If what you said will be
> implemented, maybe track_wal_io_timing too should mention fsyncs?

Thanks for the suggestions the other day.
I have created a patch that incorporates your suggestions.

Regards,
--
Hajime Matsunaga
NTT DATA Group Corporation

On 2024/07/03 17:51, Hajime.Matsunaga@nttdata.com wrote:
> Thanks for the suggestions the other day.
> I have created a patch that incorporates your suggestions.

-        <structname>pg_stat_database</structname></link>, in the output of
+        <structname>pg_stat_database</structname></link> and
+        <link linkend="monitoring-pg-stat-io-view">
+        <structname>pg_stat_io</structname></link>, in the output of

I'm not a native English speaker, but it seems more natural to use a comma instead of "and" before "pg_stat_io."

-   of block read and write times.
+   of block read, block write, extend, and fsync times.

"block" before "fsync" doesn't seem necessary.

Regards,

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

From: Fujii Masao <masao.fujii@oss.nttdata.com>
Sent: Wednesday, July 3, 2024 7:44 PM
>
> On 2024/07/03 17:51, Hajime.Matsunaga@nttdata.com wrote:
> > Thanks for the suggestions the other day.
> > I have created a patch that incorporates your suggestions.
>
> -        <structname>pg_stat_database</structname></link>, in the output of
> +        <structname>pg_stat_database</structname></link> and
> +        <link linkend="monitoring-pg-stat-io-view">
> +        <structname>pg_stat_io</structname></link>, in the output of
>
> I'm not a native English speaker, but it seems more natural to use a comma instead of "and" before "pg_stat_io."
>
> -   of block read and write times.
> +   of block read, block write, extend, and fsync times.
>
> "block" before "fsync" doesn't seem necessary.

I'm also not a native English speaker, so I appreciate your feedback.
I have created a patch that incorporates your feedback.


Regards,
--
Hajime Matsunaga
NTT DATA Group Corporation

On 2024/07/08 12:01, Hajime.Matsunaga@nttdata.com wrote:
> 
> 
> From: Fujii Masao <masao.fujii@oss.nttdata.com>
> Sent: Wednesday, July 3, 2024 7:44 PM
>>
>> On 2024/07/03 17:51, Hajime.Matsunaga@nttdata.com wrote:
>>> Thanks for the suggestions the other day.
>>> I have created a patch that incorporates your suggestions.
>>
>> -        <structname>pg_stat_database</structname></link>, in the output of
>> +        <structname>pg_stat_database</structname></link> and
>> +        <link linkend="monitoring-pg-stat-io-view">
>> +        <structname>pg_stat_io</structname></link>, in the output of
>>
>> I'm not a native English speaker, but it seems more natural to use a comma instead of "and" before "pg_stat_io."
>>
>> -   of block read and write times.
>> +   of block read, block write, extend, and fsync times.
>>
>> "block" before "fsync" doesn't seem necessary.
> 
> I'm also not a native English speaker, so I appreciate your feedback.
> I have created a patch that incorporates your feedback.

Thanks for updating the patch! Pushed.

Regards,

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