From: | Jelte Fennema-Nio <postgres(at)jeltef(dot)nl> |
---|---|
To: | Ian Lawrence Barwick <barwick(at)gmail(dot)com> |
Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Have pg_basebackup write "dbname" in "primary_conninfo"? |
Date: | 2024-02-20 00:21:56 |
Message-ID: | CAGECzQTh9oB3nu98DsHMpRaVaqXPDRgTDEikY82OAKYF0=hVMA@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Tue, 20 Feb 2024 at 00:34, Ian Lawrence Barwick <barwick(at)gmail(dot)com> wrote:
> With the addition of "pg_sync_replication_slots()", there is now a use-case for
> including "dbname" in "primary_conninfo" and the docs have changed from
> stating [1]:
>
> Do not specify a database name in the primary_conninfo string.
>
> to [2]:
>
> For replication slot synchronization (see Section 48.2.3), it is also
> necessary to specify a valid dbname in the primary_conninfo string. This will
> only be used for slot synchronization. It is ignored for streaming.
>
> [1] https://www.postgresql.org/docs/16/runtime-config-replication.html#RUNTIME-CONFIG-REPLICATION-STANDBY
> [2] https://www.postgresql.org/docs/devel/runtime-config-replication.html#RUNTIME-CONFIG-REPLICATION-STANDBY
Sounds like that documentation should be updated in the same way as
was done for pg_basebackup/pg_receivewal in commit cca97ce6a665. When
considering middleware/proxies having dbname in there can be useful
even for older PG versions.
> I can't see any reason for continuing to omit "dbname", so suggest it should
> only continue to be omitted for 16 and earlier; see attached patch.
Yeah, that seems like a good change. Though, I'm wondering if the
version check is actually necessary.
From | Date | Subject | |
---|---|---|---|
Next Message | Andy Fan | 2024-02-20 01:17:37 | Re: serial not accepted as datatype in ALTER TABLE ... ALTER COLUMN |
Previous Message | Michael Paquier | 2024-02-19 23:51:17 | Re: Fix race condition in InvalidatePossiblyObsoleteSlot() |