Re: Regarding vacuum freeze locking mechanism - Mailing list pgsql-general

From David G. Johnston
Subject Re: Regarding vacuum freeze locking mechanism
Date
Msg-id CAKFQuwYZO8d0x=cnseo0RRZS7JAbC-=c2hkZ_mxetBheb6NELQ@mail.gmail.com
Whole thread Raw
In response to Regarding vacuum freeze locking mechanism  (Durgamahesh Manne <maheshpostgres9@gmail.com>)
List pgsql-general
On Wednesday, July 17, 2024, Durgamahesh Manne <maheshpostgres9@gmail.com> wrote:
when autovacuum runs , it will freeze the transaction ID (TXID) of the table it's working on.

This statement is incorrect.  A table as a whole does not have a txid.  Freezing makes it so individual tuples get assigned an always-in-the-past txid.  Then, the table can recompute how far away its furthest back txid is from being considered in-the-future.  That gap should increase since the furthest away txids were the ones being frozen.

Inserts might get delayed a brief moment if it just happens the page they want to insert onto is presently being worked on.  But the odds there seem low.

David J.

pgsql-general by date:

Previous
From: Kashif Zeeshan
Date:
Subject: Re: Regarding vacuum freeze locking mechanism
Next
From: "David G. Johnston"
Date:
Subject: Re: Regarding vacuum freeze locking mechanism