How Can I Get The MTA To Try Different Client IP Addresses For Outbound Connections? (Doc ID 2621154.1)

Last updated on OCTOBER 06, 2022

Applies to:

Goal

An upcoming deployment will consist of the following:

* multiple (currently, up to 3) paths to the Internet
* none of them nearly as reliable as would normally be expected
* and each must use its own IP address (provided by the respective ISP)

When the ISPs work, connecting to the desired remote MTAs to relay mail should work.
But at any given time, until it is tried, it is not known if any given path is going to work.

One thought on this was that if the IP_ACCESS mapping allowed the source IP address to change, a 33% chance could be given to each of the respective source IPs.
By choosing one source IP over the others, the network should be able to choose the appropriate ISP. If the connection succeeds, great. If not, the next retry would have a 66% chance of trying one of the other two.

But, the documentation for the IP_ACCESS mapping table eludes to the IP address to which the client would otherwise be about to connect (refer to: https://docs.oracle.com/communications/F15020_01/doc.810/f15150.pdf#page=1342&zoom=auto,-116,651

How can the MTA be configured to try different client IP addresses for outbound connections?
Perhaps something like IP_ACCESS but to control the SOURCE IP address?

If there was a static list (ie destinations A,B,Q,X,Z are reachable via ISP1; dest C,D,... via ISP2; dest... via ISP3) then it would be a simple matter to create separate three channels, each with interfaceaddress set to the appropriate address for the respective ISP, and rewrite rules to send the mail for each destination to the appropriate channel.
But that is not the case. At any given time, one ISP may work for some destination and not others, or not work at all.

Solution

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.