The design intention is to provide connectivity to VLAN X & Y should either one of the legacy cores fail, unfortunately it has to be this way with the VLANs stretched across.
I could equally create a LAG that that goes from one of the legacy cores to each of the new MLAG cores, but in the diagram below I have them directly linked.
The issue here is that I will need to block either LAG A or LAG B to stop a loop. I can't use STP as its not supported on MLAG. Don't think I can really use ELRP, although it will block the loop its not really a redundancy protocol.
Was thinking of perhaps using (SRP) software redundant port, as each of the switches are Extreme, but not sure if that concept will work or how it would work with the below:
Perhaps there is another method I am missing, but just seeing what the community thinks.
That is a good option. The legacy switches are an X670 and X650, although the problem in this case is the work evolved in reconfiguring the legacy switches is required without disruption as those VLANs currently stretch all over the network too lots of different switches.