We have an open case and are pursuing it actively. We have managed to obtain debug and system and client logs from affected machines “in the act” of losing all their client settings, and have provided them to support. No conclusions to report yet, but we’ll keep the thread updated with developments.