Hello Irvan
Like standard MPLS, MPLS over FlexVPN allows the address spaces of multiple customers to be routed over an MPLS infrastructure using different VRFs. This allows customer IP address spaces to overlap. The actual addresses for each customer that can be routed across the infrastructure are advertised by the spoke routers (or PE routers in MPLS parlance).
For example, in this lesson, spoke 1 has VRF CUSTOMER defined as the address space configured on Gi0/1. This, in turn, is redistributed into BGP using the redistribute connected command. So the MPLS infrastructure learns of that customer address space and can route traffic to and from that subnet successfully. Spoke 2 is similarly configured with its CUSTOMER VRF as well.
Now, if you add the PCs and the addresses as you have done, the MPLS infrastructure knows nothing about the 192.168.100.0/24 and 192.168.200.0/24 networks. CE1 will forward such packets because it has a default gateway set up. But when spoke 1 receives such a packet, it doesn’t recognize the destination address and will drop the packet.
The same happens for packets sent from PC2.
If you want the networks behind the CE routers to be routable over the MPLS infrastructure, you must advertise those over the MPLS infrastructure as well. To do this, you must configure the CE routers to advertise their internal routes to the spoke routers. Currently in this lesson, there is no such route advertisements configured.
To see how this can be done, take a look at how the CE routers are sharing their networks in the following MPLS configuration. The config should be the same for both standard MPLS and for MPLS over FlexVPN.
I hope this has been helpful!
Laz