What is Cluster-NFS? -------------------- ClusterNFS is a based on the "Universal NFS Daemon" (UNFSD) version 2.2 Beta 47. It has been modified by Gregory R. Warnes to enable per-client file name matching . This is intended to allow multiple diskless clients to NFS mount the same root filesystem, greatly easing file system maintenance for large clusters. The original version of the UNFS is maintained by Olaf Kirch . Description ----------- When the NFS daemon is started with the command line option "-T" or "--translate-names", file lookups will be performed by attempting to match file names with various "interpreted" expressions appended. This implementation operates by appending one of a number of strings to the requested file name. The current extension strings are: $$UID=xxxx$$ user id of the user accessing the file $$GID=xxxx$$ current group id of the user accessing the file $$HOSTNAME=ssss$$ hostname of the client making the request $$IP=xxx.xxx.xxx.xxx$$ host IP number of the client making the request $$CLIENT$$ any access through the NFS server Each of these is tried, in order listed. If a match exist AND IS READABLE, it is returned, and no further attempts are made. If none of these extensions matches, the action will be attempted using the unmodified original filename. Note that if a matching file is located, but authorization is denied, this is NOT considered a match, and the next entry on the list will be attempted. To access the "base" file or directory, rather than the machine specific version, simply append "$$$$" to the filename. This also cause make two lookup attempts, first using the filename without the trailing "$$$$", then with the trailing "$$$$". (The latter, in case there is a file actually named "filename$$$$".) Note that file operations which do not use lookup, in particular mount commands and file creation, are not (yet) effected by the name mangling. Please see EXAMPLE.ClusterNFS for a simple example.