I raised this with VMware support and we went back and forth a few times trying to figure things out and eventually came to the conclusion that we would be best to to stop replication for the VM in question, remove it from SRM and put it back again. I didn't really want to do this as using Remove Replication actually deletes the files in the DR site as well - and it was on a slow enough link that doing a full sync would be a pain.
I had a bit of a play and it seems you can move the VMDK files on the recovery site while replication is working. I understand vSphere Replication actually stores the data it replicates in a separate file and then merges it back into the VMDK file, just like a snapshot is, once replication is completed. So you can move the base VMDK files out, disable replication, move the file back, enable replication and it should continue where it left off.
- SSH into one of the ESXi hosts in the DR site
- Change to the directory containing the recovery copy of the VM (you can access the datastores at /vmfs/volumes)
- Move the VMDK files out of the directory (using the mv command) - I temporarily put them one directory up so I used mv *.vmdk ../
- Go into SRM and Remove Replication for the VM and give it a couple of minutes to clean things up
- Back to the ESXi host, move the VMDK files back to the original folder (recreate it first if necessary)
- Configure replication for the VM - it should tell you that there are existing files and ask if you want to use those as the initial copy.
You should be back in business 