(C) 2003-2025, various authors

UNFS3 is a user-space implementation of the NFSv3 server specification.

UNFS3 supports all NFSv3 procedures with the exception of the READDIRPLUS procedure. It tries to provide as much information to NFS clients as possible, within the limits possible from user-space.

See the unfsd(8) manpage for restrictions imposed on NFS operations (section RESTRICTIONS) and for possible races with local file system activity (section BUGS).

It is not possible to export to netgroups or wildcard hostnames via /etc/exports, all other addressing methods should work. The following options are recognized in the exports file: ro, rw, root_squash, all_squash, no_root_squash, no_all_squash. If other options are present, they are ignored.

UNFS3 can be used to (re-)export part of an AFS network filesystem. Because AFS does not simulate inodes particularly well, configuring the source with --enable-afs is recommended in this scenario.

Cluster extensions compatible to the older ClusterNFS project are supported when the source is configured with --enable-cluster.

unfs3 is developed and tested on Linux, but should also compile and run on other Unix systems. In the past, versions of unfs3 have been successfully tested on NetBSD, FreeBSD, Solaris, AIX, Irix, and Mac OS X. There is also some support for running on Windows, see doc/README.win for details.

Releases are tested by trying to compile them on Linux, macOS and Windows.

If unfs3 doesn't build or work on a Unix system, a problem report is appreciated.

You will need gcc, lex (flex), yacc (bison), and libtirpc to compile UNFS3.

./bootstrap   # (only when building from git)
./configure
make
make install

Please read the manpage for information about command-line options.

man 8 unfsd

If you decide to modify the code yourself, you can run

make dep

to append dependency information to the Makefile, so that make knows which files depend on each other and recompiles all the necessary files on changes.