Re: Streaming read-ready sequential scan code

Hi,

On 2024-04-04 22:37:39 +1300, Thomas Munro wrote:
> On Thu, Apr 4, 2024 at 10:31 PM Thomas Munro <thomas.munro@gmail.com> wrote:
> > Alright what about this?

I think it's probably worth adding a bit more of the commit message to the
function comment. Yes, there's a bit in one of the return branches, but that's
not what you're going to look at when just checking what the function does.


> From e610bc78a2e3ecee50bd897e35084469d00bbac5 Mon Sep 17 00:00:00 2001
> From: Thomas Munro <thomas.munro@gmail.com>
> Date: Thu, 4 Apr 2024 21:11:06 +1300
> Subject: [PATCH v2 2/2] Increase default vacuum_buffer_usage_limit to 2MB.
> 
> The BAS_VACUUM ring size has been 256kB since commit d526575f.  Commit
> 1cbbee03 made it configurable but retained the traditional default.
> The correct default size has been debated for years, but 256kB is
> certainly very small.  VACUUM soon needs to write back data it dirtied
> only 32 blocks ago, which usually requires flushing the WAL.  New
> experiments in prefetching pages for VACUUM exacerbated the problem by
> crashing into dirty data even sooner.  Let's make the default 2MB.
> That's 1.5% of the default toy buffer pool size, and 0.2% of 1GB, which
> would be a considered a small shared_buffers setting for a real system
> these days.  Users are still free to set the GUC to a different value.

+1.  Independent of any other changes, this improves the default vacuum
performance substantially. We might want to dynamically size the default at
some point - but we probably should overhaul the infrastructure first...


Greetings,

Andres Freund