Re: BUG #14227: Former master not shot down - Mailing list pgsql-bugs

From hubert depesz lubaczewski
Subject Re: BUG #14227: Former master not shot down
Date
Msg-id 20160704143353.GA30344@depesz.com
Whole thread Raw
In response to BUG #14227: Former master not shot down  (rob.nieuwenhuizen@mobpro.com)
List pgsql-bugs
On Mon, Jul 04, 2016 at 01:44:38PM +0000, rob.nieuwenhuizen@mobpro.com wrote:
> I have a master / slave setup with a hot-standby slaves.
> When I do a failover to the hot-standby slave the slave becomes master, but
> the former master is not shot down.
> This results in 2 master servers.
> Maybe this is by design, maybe there is a configuration flag I'm missing.

this is not a bug, so it shouldn't be posted to -bugs. -general mailing
list seems to be better suited.

PostgreSQL doesn't kill old master, there is no such flag, because - how
could it?

By "do a failover" I assume you created trigger file on slave, or used
pg_ctl promote.
In both cases, it happens on slave, which can't connect and kill pg on
master.
Not to mention that it wouldn't always be what user really needs.

If you want to stop Pg on old master, ssh there, and kill pg yourself.

Best regards,

depesz

--
The best thing about modern society is how easy it is to avoid contact with it.
                                                             http://depesz.com/

pgsql-bugs by date:

Previous
From: rob.nieuwenhuizen@mobpro.com
Date:
Subject: BUG #14227: Former master not shot down
Next
From: Tom Lane
Date:
Subject: Re: "insert [...] on conflict" hangs on conflict on an unmentioned gist index