RFC 2328, "OSPF Version 2", April 1998Source of RFC: ospf (rtg)
Errata ID: 5611
Reported By: Anil Chaitanya Mandru
Date Reported: 2019-01-22
Rejected by: Alvaro Retana
Date Rejected: 2019-01-28
Section 16.1 (4) says:
In this case, the current routing table entry should be overwritten if and only if the newly found path is just as short and the current routing table entry's Link State Origin has a smaller Link State ID than the newly added vertex' LSA.
It should say:
In this case, if the newly found path is just as short, then both the paths should be added to the routing table.
If the newly found path is just as short then both the paths should be considered for ECMP. Why should the smaller Link State ID path overwrite the current one even if the paths are equi distant?
See WG discussion here: https://mailarchive.ietf.org/arch/msg/lsr/ACVzdktoiFbIcMyQck1RCGn50es
From Acee Lindem: "This is the case where the transit network vertex corresponding to a network-LSA is newly added to the current intra-area graph and an intra-area route corresponding to the SPF in progress already exists. This implies that there was another network-LSA. So, either the network corresponding to the subnet is partitioned or there is a network configuration error with the same subnet configured for multiple multi-access networks. In either case, I don't really see the benefit of an ECMP route. In fact, it could make trouble-shooting the problem more difficult. "
Note also that the proposed text would result in a modification of the process to calculate the routing table, not just a correction. A change like that should be discussed in the WG/mailing list instead.