Hello,
can you pls help me to understand why below sentence is true ?
Because otherwise, it would be a chicken-and-egg problem. The sham-link is configured using the opposite PE's IP address in a particular VRF - it is basically a targeted OSPF session between two distant PEs. Assuming that you want OSPF to carry this IP address, you have an unsolvable situation: without knowledge where this IP address is located, the sham-link can not go up, and if the sham-link can not go up, you can not find out where this IP address is located.
Best regards,
Peter