But the setup I sugested is not :
1-2-3-4-5-2-3-4-5....
It is :
1-2-3-4-5
(Mach5 will be using the same internet link but it is not the same machine as
Mach1. Mach5 will ahve no fallback relay.)
(
The packet routing solution is not feasible because those servers are located
in different cities, conected by a VPN.
Some pair are 4 thousand kilometers apart.
The VPN provider is very reliable but the internet links are not.
And to change this will take months.
There was moments when the VPN remains active and the internet links are down.
That's why I was comissioned to solve the problem this way.
There are other people addressing the packet routing angle.
)
Em 08/11/2010 17:39, Wietse Venema escreveu:
Victor Duchovni:
On Mon, Nov 08, 2010 at 03:29:49PM -0500, Wietse Venema wrote:
Luiz Antonio Emerenciano Alcoforado:
Like that : Mach1 --> Mach2 --> Mach3 --> Mach4 --> Mach5 (same link as
Mach1)
Thus, when the problem were raised by the faling link, Mach1 will
forward the pending emails to Mach2 and they could be sent.
When all links are down, you have a loop
1-2-3-4-5-2-3-4-5....
going at local network speeds.
The right solution to this problem is a packet routing solution,
not an SMTP message routing solution.
Like setting the default route to the "right" gateway. That would
be robust only if those gateways perform network address translation,
otherwise who knows what path the return packets will take.
Wietse
