Friday January 03 2020, Kees van Eeten wrote to Michael Dukelsky:
MD>> I handle the loop messages differently. If a message comes to me MD>> not for the first time, it is moved to the special area. Once a MD>> day messages are released from the area. This is done in a hope MD>> that the routing may be fixed during this delay. After the MD>> message comes for the 7th time, it is clear that the routing has MD>> not been fixed during the week and the message is bounced back to MD>> the sender (if he is in the nodelist). And the last but not the MD>> least: when the message comes for the second time and a loop is MD>> detected, I receive a message from my robot about it. So I can MD>> take some actions to help with fixing the routing.
KE> I do not know how my tosser handles loops,
This is the key phrase. Your tosser does not handle loops, it will do nothing if you do not develop a Perl script for it. All that I described above is a configuration in RNtrack.
KE> but in this case it was not KE> routing failure. Du to some unfortunate handling the messages KE> contained twoo INTL kludge lines. The first twoo systems in the loop KE> processed the first line, causing the message to be delivered to the KE> intended system. There and at the following system the second INTL KE> line was processed, causing the message to be returned to the sender.
If you received a message that some mail came to your node a second time, you could investigate the problem and would not have to wait until the same mail came 15 times.
KE> Nobody cared to make his software safe, to messages structures, that KE> should not exist.