Module Release January 2003: Control Filesystem Mounting and Unmounting

Last month's module covered the creation of partitions and filesystems. This month we move on to the issue of mounting filesystems.

The module begins by defining key concepts, i.e., filesystem, root filesystem, mounting and mount point. In the classroom situation, this is where a tutor would emphasise the way in which Unix has a unified filestem, allowing any number of devices and their filesystems to be grafted on to arbitrarily convenient ‘mount points’ within the root filesystem's directory structure. This enables any number of extra devices or systems to be added within a single coherent structure, and avoids clumsy arrangements like Microsoft's A–Z drive mapping.

All filesystems are added to the root filesystem using the mount command and removed from it using the umount command. ‘Unmount’ may have been a more mnemonically convenient name for the latter, but the history of Unix is littered with questionable naming and syntax decisions which become too deeply embedded to change.

Important filesystems are mounted automatically at boot time, using parameters defined in the system's filesystem table (/etc/fstab). Filesystems for devices which may be attached to and detached from the root filesystem on a regular basis (e.g., CD-ROM drives, digital cameras, etc) are mounted ‘manually’. That is, a human user defines mount parameters like the device's name, its prospective mount point and the conditions under which it is mounted, e.g., writability, executability and user permissions. In most cases, only the root user will have the right to mount or unmount parts of the unified filesystem.

Obviously, given the very wide range of devices and filesystems (including non-Unix filesystems like VFAT and NTFS), a fair amount of discussion is devoted to the particular options required to specify and manage such devices.