I have modified the scale-out data loader such that it will write on the full text index. However, it does not currently do this using the asynchronous index write pipeline. Instead the TokenBuffer is used and writes on the IClientIndexView (the scale-out view of the sharded index). This works, but it lacks the high performance of the pipelined index writes.
The index write pipeline should be integrated for the full text index before this issue is closed. Until then the use of the full text index with the cluster can not be recommended.
Committed Revision r5290.