From: | Thom Brown <thom(at)linux(dot)com> |
---|---|
To: | Jelte Fennema <postgres(at)jeltef(dot)nl> |
Cc: | PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Allow specifying a dbname in pg_basebackup connection string |
Date: | 2023-07-05 12:43:35 |
Message-ID: | CAA-aLv6QHjBPZtGjy2mpP=1r4+vW=buigGUpU-cVqTLU=ARVNg@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Mon, 3 Jul 2023 at 13:23, Jelte Fennema <postgres(at)jeltef(dot)nl> wrote:
>
> Normally it doesn't really matter which dbname is used in the connection
> string that pg_basebackup and other physical replication CLI tools use.
> The reason being, that physical replication does not work at the
> database level, but instead at the server level. So you will always get
> the data for all databases.
>
> However, when there's a proxy, such as PgBouncer, in between the client
> and the server, then it might very well matter. Because this proxy might
> want to route the connection to a different server depending on the
> dbname parameter in the startup packet.
>
> This patch changes the creation of the connection string key value
> pairs, so that the following command will actually include
> dbname=postgres in the startup packet to the server:
>
> pg_basebackup --dbname 'dbname=postgres port=6432' -D dump
I guess my immediate question is, should backups be taken through
PgBouncer? It seems beyond PgBouncer's remit.
Thom
From | Date | Subject | |
---|---|---|---|
Next Message | Amit Kapila | 2023-07-05 12:47:54 | Re: logicalrep_message_type throws an error |
Previous Message | Thom Brown | 2023-07-05 12:38:44 | Re: Disabling Heap-Only Tuples |