Re: Issue with logical replication slot during switchover

Alexander Kukushkin <cyberdemn@gmail.com>

From: Alexander Kukushkin <cyberdemn@gmail.com>
To: Fabrice Chapuis <fabrice636861@gmail.com>
Cc: shveta malik <shveta.malik@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Shlok Kyal <shlok.kyal.oss@gmail.com>
Date: 2025-11-04T13:01:45Z
Lists: pgsql-hackers
On Tue, 4 Nov 2025 at 12:15, Fabrice Chapuis <fabrice636861@gmail.com>
wrote:

> Hi Alexander,
> I understand your use case but I think it won't work like this because of
> how is implemented get_local_synced_slots()
> this function will return slots which synced flag to true only.
>

Yeah. Sorry, my bad, you are absolutely right
about get_local_synced_slots() function.
I didn't validate this part of the patch, because first I wanted to get
feedback from hackers.
However, it looks like the people who build it never run Postgres in
production and therefore don't understand the problem and the pain.
I will probably just add yet another workaround to Patroni and start
dropping logical replication slots on standby with failover=true and
synced=false.

-- 
Regards,
--
Alexander Kukushkin