How can the risk of UID/GID inconsistencies be avoided across UNIX/Linux systems that are sharing information over protocols other than SMB/CIFS