North South traffic was lost when the active T1 was place on the passive edge.
The active T0 was located on the active edge.
The next variable needed to produce the issue was the presence of static routes and overlapping networks existing.
This topology is not supported. East West traffic will flow as expected.
The use of overlapping networks in the IP scheme is not supported in this topology.
The details of why are in the cause section of this article.
NSX
The topology has an active T1 relocated from edge1 to edge2, and paired with the standby T0. The standby T0 on edge2 will not have the static routes, which is expected and per design.
The above diagram shows why this topology will not pass the traffic for the static routes. Note that in the active edge1, T-0 has a merged FIB.
The Service Router (SR) and Distributed Router (DR) routes are merged into one FIB. The FIB includes the static routes and the DR component of
T0 sees the static route.
Notice that T-0 on edge 2, which is in standby, does not have a merged FIB. The SR and DR components of T-0 on edge 2 do not share the same routes.
In this case, the static routes are in the standby T-0 SR component’s FIB only. The standby T0 DR component does not have the more specific static routes to reach the north. Since the overlapping subnet points to the downlink, the traffic drop occurs. This would not be a problem for non-overlapping networks. When the T0 networks don’t overlap with the T1 routes, then the traffic will use the default route in the T-0 DR to reach the SR and we will not see the traffic drop. In this specific site, overlapping subnets are present. This is the key to the problem. If there were no overlaps this issue would not be occurring. The presence of overlapping subnets will precipitate the network outage.
This is working per design. There is no product defect.