Re: fdatasync performance problem with large number of DB files

On 2021/03/19 14:28, Thomas Munro wrote:
> On Fri, Mar 19, 2021 at 3:23 PM Fujii Masao <masao(dot)fujii(at)oss(dot)nttdata(dot)com> wrote:
>> Thanks for updating the patch! It looks good to me!
>> I have one minor comment for the patch.
>>
>> + elog(LOG, "could not open %s: %m", path);
>> + return;
>> + }
>> + if (syncfs(fd) < 0)
>> + elog(LOG, "could not sync filesystem for \"%s\": %m", path);
>>
>> Since these are neither internal errors nor low-level debug messages, ereport() should be used for them rather than elog()? For example,
>
> Fixed.

Thanks! LGTM.

> I'll let this sit until tomorrow to collect any other feedback or
> objections, and then push the 0001 patch
> (recovery_init_sync_method=syncfs).

Understood.

> On Fri, Mar 19, 2021 at 4:08 PM Fujii Masao <masao(dot)fujii(at)oss(dot)nttdata(dot)com> wrote:
>> 0002 patch looks good to me. Thanks!
>> I have minor comments.
>
> Ok, I made the changes you suggested.

Thanks! LGTM.

> Let's see if anyone else would
> like to vote for or against the concept of the 0002 patch
> (recovery_init_sync_method=none).

Agreed. I also want to hear more opinion about the setting "none".

Regards,

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