Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 10 Dec 2015 20:27:56 +0100
From:      Terje Elde <terje@elde.net>
To:        Emmanuel Vadot <manu@bidouilliste.com>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: CARP demotion counter
Message-ID:  <F5A58402-FBA6-4B18-B78A-D38C4AA8D3D4@elde.net>
In-Reply-To: <ee17d45f9f7b1a24f76b59b09608d466@megadrive.org>
References:  <20151210182821.GM13477@mordor.lan> <ee17d45f9f7b1a24f76b59b09608d466@megadrive.org>

Next in thread | Previous in thread | Raw E-Mail | Index | Archive | Help

> On 10 Dec 2015, at 20:00, Emmanuel Vadot <manu@bidouilliste.com> wrote:
>=20
> You need net.inet.carp.preempt=3D1 for MASTER returning to MASTER after a B=
ACKUP state.

If all things are otherwise equal, I've come to favor having redundant equip=
ment not be designated master/slave, but rather "router 1" and "router 2" et=
c, where possible.=20

It reduced "oh, the other is just backup"-impulses, promotes caring equally f=
or either box, and reduces the idea of failover and fallback being separate t=
hings.=20

Also removes "we need to fail back to the master" as a thing to do after a f=
ailover, while the equality makes it easier to do changes as a matter of rou=
tine.=20

Terje




Want to link to this message? Use this URL: <http://docs.FreeBSD.org/cgi/mid.cgi?F5A58402-FBA6-4B18-B78A-D38C4AA8D3D4>