How does it work?

OGRSH works using a classic 'shimming' technique. Using shared library techniques built into the Linux operating system, OGRSH intercepts certain calls made from the application and into lower-level dependent libraries. By intercepting calls, such as opendir, readdir, closedir, open, read, write, close, etc., OGRSH is able to provide the application with the illusion of a file system namespace that only exists virtually in the OGRSH library. In reality, this namespace is made up of a number of arbitrary 'mount' points where a configuration file has indicated which local file system directory, or which grid file system directory, to mount and under what OGRSH name. This is done via a pluggable mechanism, whereby file system provider shared libraries handle paths that resolve under their given mount points and are allowed to respond to calls made to those paths.

Because OGRSH is intercepting the calls in a standard way, any binary application that uses those calls can work within an OGRSH shim. This includes very useful legacy programs, such as ls, cat, vi, and even bash. In fact, the most common way to start OGRSH is by launching an instance of the BASH shell with an OGRSH shim. This interactive shell is then connected to the grid via configured mount points, and users can navigate and manipulate the file system structure provided there.