An enterprise has four diverse transports at each data center edge and wants OMP to load-share to a branch prefix over all healthy paths. The engineer sees only one path installed. Which OMP best-path and multipath behavior explains expanded forwarding over equal-cost paths?

Correct behavior is OMP best-path selection followed by installation of up to four equal-cost OMP routes into the forwarding table, which directly explains how four diverse transports at each edge can all carry traffic to the same branch prefix when healthy, with best-path tie-breakers still determining the advertised best route while multipath preserves load sharing. The three-path variant fails on scale because the default limit is four (ecmp-limit, configurable up to 16), not three, so it cannot satisfy the all-healthy-paths requirement for four transports despite correctly describing best-path selection and load sharing. The single-path-until-enabled variant fails on default behavior because it describes a BGP-like ECMP knob model requiring explicit activation, whereas the scenario asks for the expanded forwarding behavior that installs multiple equal-cost paths without that extra condition, which is why seeing only one installed path points to path inequality or policy rather than a missing multipath switch. The standby-failover variant fails on forwarding purpose because it correctly captures four-path installation but mischaracterizes use as active-standby redundancy, contradicting the explicit load-share objective where all healthy paths forward simultaneously. Exam caveat is to distinguish best-path advertisement from multipath forwarding and to verify equal cost, origin, and TLOC resolvability before assuming four paths should appear. Practical check is to compare control-plane received routes with data-plane installed routes on the edge router and confirm the four TLOCs show equal OMP cost and healthy BFD.