Re: Logical replication in PostgreSQL Amount of subscriber vs publisher WALs Vacuuming

Поиск
Список
Период
Сортировка
От PALAYRET Jacques
Тема Re: Logical replication in PostgreSQL Amount of subscriber vs publisher WALs Vacuuming
Дата
Msg-id 1748712738.372293199.1774455320476.JavaMail.zimbra@meteo.fr
обсуждение исходный текст
Ответ на Logical replication in PostgreSQL Amount of subscriber vs publisher WALs  (PALAYRET Jacques <jacques.palayret@meteo.fr>)
Список pgsql-general
Hello,

Vacuuming seems to explain these differences.
Especially for large tables, vacuuming can generate many WAL.

Regards

De: "PALAYRET Jacques" <jacques.palayret@meteo.fr>
À: pgsql-general@lists.postgresql.org
Envoyé: Mardi 24 Mars 2026 15:06:23
Objet: Logical replication in PostgreSQL Amount of subscriber vs publisher WALs

Hello,

The amount of WAL generated by an SQL INSERT or UPDATE statement applied to a table with multiple indexes can be much greater than the size of the table (table + index).
For example, an INSERT statement in an empty table (with 3 indexes) can generate WALs twice the size of the table (table + index).
This difference (even for an INSERT) may seem surprising, but it's understandable.

What's less intuitive is that, according to my tests, with logical replication in PostgreSQL, the amount of WAL generated by an SQL statement can be very different between the subscriber server (the replica) and the publisher server (the provider).
Is this accurate? Sometimes 1.5 or 2 times greater?

Regards
----- Météo-France -----
PALAYRET Jacques
DCSC/GDC
jacques.palayret@meteo.fr
Fixe : +33 561078319

В списке pgsql-general по дате отправления: