Thanks to those who replied to my question. Here's what they said:
--
Alan Thew alan.thew@???
Computing Services,University of Liverpool Fax: +44 151 794-4442
Thanks to:
Nigel Cass <ncass@???>
Philip Hazel <ph10@???>
Piete Brooks <Piete.Brooks@???>
From: Nigel Cass <ncass@???>
Alan.
I'm running two Exim Mailhubs with the same MX records and everything
works a treat. Essentially the only things I maintain separately are
files that my domainlist routers use, and obviously local_domains on
each box is slightly different.
Good luck - you shouldn't need it too much :)
> Date: Mon, 20 Oct 1997 20:44:12 +0100 (BST)
> From: Alan Thew <Alan.Thew@???>
> Reply-to: Alan Thew <Alan.Thew@???>
> To: Exim List <exim-users@???>
> Subject: Mirrored/paired hubs...
> Organization: The University of Liverpool
> I know this is not strictly exim but I have 2 machines both running
> 1.73 which I want to run as a pair with the same MX record etc, all main
> data files are common etc.
>
> Do have people have any advice/horror stories? If you mail me directly I
> will sumarise to the list.
>
> Thanks
>
From: Philip Hazel <ph10@???>
On Mon, 20 Oct 1997, Alan Thew wrote:
> I know this is not strictly exim but I have 2 machines both running
> 1.73 which I want to run as a pair with the same MX record etc, all main
> data files are common etc.
>
> Do have people have any advice/horror stories? If you mail me directly I
> will sumarise to the list.
The system from which I am sending this message consists of a pair of
identical Sun servers running Solaris 2.5. There are a pair of
identically numbered MX records pointing at the two machines. Each
machine has its own copy of Exim configuration files and an Exim spool,
but the user mailbox directory (/var/mail) is in fact on an entirely
separate dedicated fileserver and mounted over NFS (as are the users'
home directories). Also /opt/exim, which contains the binaries and
utility programs is on a separate fileserver, so there is only one copy
of that.
We maintain the Exim configuration files by having a shared "build"
directory in which edits are done, using SCCS to keep an audit trail.
Then rdist is used to push a copy to each of the machines. The only
thing one has to do "per machine" is to HUP the daemon when changing the
configuration or after installing a new version of Exim.
The system has been working well like this for some time, doing around
30,000 deliveries in total a day when I last looked. A week or so ago
one of the machines was out of service for some hours, but the other one
kept on running happily and mail service was not disrupted.
Summary:
Shared: /opt/exim/bin, /opt/exim/util, /var/mail, /home/*
Private: /var/spool/exim, /etc/mail/*
From: Piete Brooks <Piete.Brooks@???>
> Do have people have any advice/horror stories?
I can't see the problem ...
I suspect you need to set up commands to say "ensure things are in step".
E.g. you change a file on one machine and it need to be updated on the other.
I "normally" do all changes on one machine, and then update the "slave", but I
use "rdist -y" so that if there is a change made on the other, I get warned ...
--
* This is sent by the exim-users mailing list. To unsubscribe send a
mail with subject "unsubscribe" to exim-users-request@???
* Exim information can be found at http://www.exim.org/
