From: | Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com> |
---|---|
To: | "Andrey M(dot) Borodin" <x4mmm(at)yandex-team(dot)ru> |
Cc: | PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
Subject: | Re: Using read stream in autoprewarm |
Date: | 2024-11-06 14:56:41 |
Message-ID: | CAN55FZ2xGRgLm368xYCvQAqLJ+Dkx5fJ8ZdKO6G=GxaAusuqkw@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hi,
On Fri, 1 Nov 2024 at 21:06, Andrey M. Borodin <x4mmm(at)yandex-team(dot)ru> wrote:
>
> > On 1 Nov 2024, at 12:51, Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com> wrote:
> >
> > am not
> > sure whether 'BufferStrategyControl.lastFreeBuffer -
> > BufferStrategyControl.firstFreeBuffer' is safe to use.
>
> Ugh... it will work. But it seems to me too dirty hack. There's no scalable way to know size of a free list.
> Let's just comment that we might read some more buffers if database does not fit into memory?
> Alternatively we can count size of a free list on the start.
I agree that it is too dirty to hack. There is a minor problem with
the counting size of a free list on the start. There may be other
processes that fill the buffer pool concurrently, so we can still end
up doing unnecessary I/Os. That said, I believe this approach remains
an improvement.
The first patch includes the comment you suggested, and the second
patch implements counting size of a free list on the start.
--
Regards,
Nazir Bilal Yavuz
Microsoft
Attachment | Content-Type | Size |
---|---|---|
v2-0001-Use-read-stream-in-autoprewarm.patch | text/x-patch | 5.9 KB |
v2-0002-Count-free-buffers-at-the-start-of-the-autoprewar.patch | text/x-patch | 2.6 KB |
From | Date | Subject | |
---|---|---|---|
Next Message | Daniel Gustafsson | 2024-11-06 15:03:35 | Re: pg_upgrade check for invalid databases |
Previous Message | Robert Haas | 2024-11-06 14:43:27 | Re: Eager aggregation, take 3 |