Horton
Menu

Re: n_ins_since_vacuum stats for aborted transactions

Sami Imseih <samimseih@gmail.com>

View thread
From: Sami Imseih <samimseih@gmail.com>
To: "David G. Johnston" <david.g.johnston@gmail.com>
Cc: Euler Taveira <euler@eulerto.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2025-04-09T18:57:44Z
Lists: pgsql-hackers 
> Forget original purpose, is there presently a bug or not?

Yes, there is a bug. Accounting rows inserted as part of an aborted
transaction in
n_ins_since_vacuum is not correct, since the same rows are being
accounted for with n_dead_tup.

> Between the two options the one where we count dead tuples makes more sense on its face.

IIUC, I am saying the same thing, if an inserted row is rolled back,
it should only count as a
dead tuple (n_dead_tup) only, and not in n_ins_since_vacuum.

--
Sami Imseih
Amazon Web Services (AWS)

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Add code comment explaining ins_since_vacuum and aborted inserts

  2. Trigger autovacuum based on number of INSERTs