Message-ID: <481ee580-ba4b-445d-bf35-7bee976543b4@netcologne.de>
Date: Mon, 30 Jun 2025 20:31:09 +0200
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: Replication and Switchover
To: pgsql-admin@lists.postgresql.org
References: 
 <CAN6TVj=01f6At2L4X7WM8g1c+18up8mSLB1GSKzJdJNAEUtaag@mail.gmail.com>
Content-Language: en-US
From: vrms <vrms@netcologne.de>
In-Reply-To: 
 <CAN6TVj=01f6At2L4X7WM8g1c+18up8mSLB1GSKzJdJNAEUtaag@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/481ee580-ba4b-445d-bf35-7bee976543b4%40netcologne.de>
Precedence: bulk

just to remind myself about the 'switchover' process ... in such a case 
as the on at hand here there is no ambition to actually switch and 
replicate back to the original main, but rather use replication 
mechanics the 'clone' the cluster (or in your case part of it), right?

In order to 'untie' the cloned cluster you would

   1. stop the replica
   2. remove the standby.signal file from the replicas $PGDATA
   3. delete the postgresql.auto.conf  on the replica
   4. start the replica anew

... is that a fair assumption? Or is the above more or less what "SELECT 
pg_promote() ;" does?

I guess #3 might not even be a hard requirement as it is being ignored 
in absence of a standby.signal


On 06.06.25 16:45, Sam Stearns wrote:
> Howdy,
>
> We have 2 databases running out of a single database cluster.  We 
> would like to set up replication from the existing host to a new host 
> and then switchover to that new host.  One of the databases is ready 
> to switchover now while the other database is going to take a bit more 
> time to be ready for the switchover.
>
> We would like to switchover the database that is ready now so we can 
> start performing load testing against it.  Is it possible to 
> switchover one database in a cluster while not switching over the other?
>
> Thanks,
>
> Sam