When you see the above message in the GoldenGate logfile there are usually orphan processes that prevent GoldenGate from locking the file (typically when your trail files are on NFS). In a case I had at a customer last week this was not the case. It could be confirmed that there are no other processes sitting on the file by doing an fuser on the file on all nodes of the cluster (This was an 8 node Exadata). What we finally needed to do was:

cd [TRAIL_DIRECTORY]
mv de000000 de000000_bak
cp de000000_bak de000000
rm de000000

Once we did this we could start the extract again and GoldenGate was happy. Hope this helps …