I know we had a conversation here with Aleš about this roughly a year ago or a little less, but I can't find it. My preference is to use a hybrid haplogroup identifier, e.g., R1b1-BY35083, and that was throwing the same error. We got it corrected; no more suggestion for it in my report.
I can't fathom why R-FTD8841 would be causing a problem. I'm pretty confident that there is no validation going on against a downloaded public version of FTDNA's haplotree, but I could be wrong. My impression was that the only validation was in the syntactical construction of the haplogroup name...which is why my "R1b1-" thing was blowing it up.
I mean, in the past 12 months alone FTDNA has added 12,244 new haplotree branches (in September 2018 the entire tree consisted of 16,361 branches). It would seem inefficient to try to keep up with that, especially considering that FTDNA, YFull, and Y-DNA Warehouse aren't in complete agreement about their trees.