Single DHCP scope failed after subnet failover

I have several remote sites, using different subnets for each, which are all connected to the central office, which is where the central DHCP server is located. Each site has a dedicated scope. One day, ONE of the remote site's MPLS circuit failed so I had the site rerouted over a failover device which VPN'd back into our central office,using the same remote subnet addresses as the primary circuit. (while the mpls was disconnected so it wouldn't conflict). All was OK until ONE device could not pull a DHCP address, then later another one, then later another one and so on... After an hour the Primary MPLS circuit came back up so I rerouted everything back to normal and just assumed that the devices that had the problem would then pull an IP address from the scope..... not so lucky.  It turns out that even after all was normalized, the devices still couldn't pull an ip from that subnet/scope.  I even tried adding a device to that remote site network(from spares closet) so it would pick up a NEW ipaddress from the scope and still nothing. I was trying renewals versus new request.   Called the MPLS people and kinda blamed it on them, thinking that they changed something, but that didn't work. They said the configurations did not change since before the outage.  The only thing else I could try is to delete scope and rebuild it.  That worked...?? Windows 2003 dhcp..... Any thoughts?  This kinda throws a wrench into my disaster plan... having to rebuild a scope every time this happens??