We are investigating an increased amount of errors from service endpoints.
We've rolled back the changes and the errors subsided.
