Thread

Re: Add wal_fpi_bytes_[un]compressed to pg_stat_wal

Shinya Kato <shinya11.kato@gmail.com> — 2025-10-31T02:15:03Z

On Thu, Oct 30, 2025 at 3:57 PM Michael Paquier <michael@paquier.xyz> wrote:
>
> On Tue, Oct 28, 2025 at 10:04:20PM +0900, Michael Paquier wrote:
> > I'll clean up that tomorrow, which can be summarized as something like
> > the attached (quick fix, need to double-check).
>
> Done this cleanup as d3111cb753e8, and tweaked a bit the second patch
> (order of the fields and docs) before applying it as 5ab0b6a24807.

Thank you for the fix and committing.

> On top of what has been already done here, should we also update the
> logs generated by do_analyze_rel() and heap_vacuum_rel()?

You're right, I missed them. I have created a patch that addresses the
above. Similar to EXPLAIN, I wanted to use 'fpi' and 'fpi bytes', but
the VACUUM/ANALYZE logs showed 'full pages images', so I used 'full
page image bytes'.

postgres=# vacuum (verbose) pg_class;
~snip~
WAL usage: 1 records, 1 full page images, 7935 bytes, 7816 full page
image bytes, 0 buffers full

postgres=# analyze (verbose) pg_class;
~snip~
WAL usage: 92 records, 6 full page images, 49416 bytes, 37488 full
page image bytes, 0 buffers full


-- 
Best regards,
Shinya Kato
NTT OSS Center