Comment (by Roland Haas):

How strange. Yes, access to MN5 might be useful for this, just in case this is cluster dependent issue. I likely still exist in the PRACE users databases (if PRACE still handles this) using my rhaas@aei.mpg.de email, otherwise inviting rhaas@mail.ubc.ca is likely best (still an EU citizen in case they have silly restrictions in place [some clusters apparently do]).

--
Ticket URL: https://bitbucket.org/einsteintoolkit/tickets/issues/2899/bhns-with-elliptica-assertion-all-offset