Re: pgBufferUsage.blk_{read|write}_time are zero although there are pgBufferUsage.local_blks_{read|written}

From: Melanie Plageman <melanieplageman(at)gmail(dot)com>
To: Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: pgBufferUsage.blk_{read|write}_time are zero although there are pgBufferUsage.local_blks_{read|written}
Date: 2023-09-15 13:30:24
Message-ID: CAAKRu_azZCAZJSpMuG_F1Sk2tOnn=TvUwHcxiwct1XCrbv5iKQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Fri, Sep 15, 2023 at 9:24 AM Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com> wrote:
> I found that pgBufferUsage.blk_{read|write}_time are zero although there are pgBufferUsage.local_blks_{read|written}

Yes, good catch. This is a bug. I will note that at least in 15 and
likely before, pgBufferUsage.local_blks_written is incremented for
local buffers but pgBufferUsage.blk_write_time is only added to for
shared buffers (in FlushBuffer()). I think it makes sense to propose a
bug fix to stable branches counting blk_write_time for local buffers
as well.

- Melanie

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Daniel Gustafsson 2023-09-15 13:34:35 Re: sslinfo extension - add notbefore and notafter timestamps
Previous Message jacktby jacktby 2023-09-15 13:13:39 Re: Implement a column store for pg?