\section{{\em NFS\/}: The Network Filesystem}

Many of the files you use are actually stored on a disk connected to
one of Athena's {\em file servers}, machines whose primary function is
to provide disk storage for users, projects, and
``courseware.''\footnote{Some of the machines serve other functions
which don't interfere with providing file service.} Your workstation
can access them through several different ways, one of them being {\em
NFS}, the Network File System (developed by Sun Microsystems).  The
most common way to access NFS remote filesystems is through the {\tt
attach} command.  This system lets many workstations access the disk
storage on a server machine as if that disk storage were directly
attached to the workstation.

To protect your files, Athena has modified the standard {\em
NFS}\/ server machines so that in order for you to convince the
server machines that you really are who you say you are (and
therefore have access to {\em your}\/ files), you must establish
a {\it map} between your user-id on the workstation and the
appropriate user-id on the server.  The {\tt attach} command
does this automatically by presenting {\em Kerberos}\/
authentication to the server\footnote{See the {\bf Kerberos}
section of this guide.}.

Unfortunately, problems occasionally arise with the file
servers.\footnote{AFS, addressed later, can deal better with such
problems.  This is one of the main reasons Athena is adopting it.} In
spite of the preventive maintenance occasionally performed on the file
servers by their manufacturers, Athena's file servers do occasionally
crash due to hardware or software problems.  This typically results in
messages on the workstation console window about ``TIMED OUT.''  This
message will include the name of the failed server.  By default,
filesystems are {\tt attach}ed so that if a file server that you are
using crashes, should you try to access files from that server, your
workstation will hang until the file server resumes service or until a
certain amount of time has passed.  If this time period passes and the
workstation is still unable to contact the file server, the
workstation will give up.  If there are no serious problems with the
server machine, it usually takes about thirty (30) minutes to resume
service.  When the server reboots, however, it will have forgotten all
of the maps established by workstation users.  So if you wait around
until a file server has rebooted, you must manually re-establish a map
with the file server.  You do this with the command {\tt nfsid {\it
server}}.  The argument, {\it server}, should be the name of the
server with which you wish to re-establish a map.  The Hesiod section
above details how to find the correct server name.  You can remap to
all of the filesystems you have attached with the command {\tt nfsid
-a}.  You can remap to the server containing a single filesystem with
the command {\tt nfsid -f {\it filesystem}}.  If you still have
trouble accessing files that you know you are supposed to be able to
have access to, you can always try re{\tt attach}ing the filesystem
that you are having trouble with.