Re: Effective IO Concurrency
От | luis.roberto@siscobra.com.br |
---|---|
Тема | Re: Effective IO Concurrency |
Дата | |
Msg-id | 330249832.8108571.1600100745617.JavaMail.zimbra@siscobra.com.br обсуждение исходный текст |
Ответ на | Re: Effective IO Concurrency (Ron <ronljohnsonjr@gmail.com>) |
Список | pgsql-general |
De: "Ron" <ronljohnsonjr@gmail.com>
Para: "pgsql-general" <pgsql-general@lists.postgresql.org>
Enviadas: Segunda-feira, 14 de setembro de 2020 13:10:23
Assunto: Re: Effective IO Concurrency
Para: "pgsql-general" <pgsql-general@lists.postgresql.org>
Enviadas: Segunda-feira, 14 de setembro de 2020 13:10:23
Assunto: Re: Effective IO Concurrency
On 9/14/20 11:03 AM, Laurenz Albe wrote:
> On Mon, 2020-09-14 at 10:39 -0300, luis.roberto@siscobra.com.br wrote:
>> In PostgreSQL 13, the way of using effective_io_concurrency has changed. Until v12,
>> I used 200 for this value (using SSD drives). Using the new formula described in
>> https://www.postgresql.org/docs/13/release-13.html gives me 1176.
>> However, in the documentation https://www.postgresql.org/docs/13/runtime-config-resource.html#GUC-EFFECTIVE-IO-CONCURRENCY
>> it says that the maximum value allowed is 1000.
> Then use the value 1000...
I think he means that the formula should take that into account.
--
Angular momentum makes the world go 'round.
> On Mon, 2020-09-14 at 10:39 -0300, luis.roberto@siscobra.com.br wrote:
>> In PostgreSQL 13, the way of using effective_io_concurrency has changed. Until v12,
>> I used 200 for this value (using SSD drives). Using the new formula described in
>> https://www.postgresql.org/docs/13/release-13.html gives me 1176.
>> However, in the documentation https://www.postgresql.org/docs/13/runtime-config-resource.html#GUC-EFFECTIVE-IO-CONCURRENCY
>> it says that the maximum value allowed is 1000.
> Then use the value 1000...
I think he means that the formula should take that into account.
--
Angular momentum makes the world go 'round.
Exactly....
For future reference, and maybe a updated documentation:
SELECT least(round(sum(OLD / n::float)),1000) FROM generate_series(1, OLD) s(n)
I don't know how to write patches, so maybe someone can do that.
Thanks.
В списке pgsql-general по дате отправления: