Two versions of NFS are currently in use. NFS version 2 (NFSv2), which has been
around for several years, is widely supported by various operating
systems. NFS version 3 (NFSv3) has several more features, including a variable
file handle size and better error reporting. Red Hat Linux supports both NFSv2 and
NFSv3, and uses NFSv3 by default when connecting with a server that
supports it.
This chapter will focus on NFS version 2, though many of the concepts
discussed also apply to version 3. Additionally, only fundamental NFS
concepts and supplemental information will be provided. For specific
instructions regarding the configuration and operation of NFS on client or
server machines, see the chapter titled Network File System (NFS)
in the Red Hat Linux Customization Guide.
9.1. Methodology
Linux uses a combination of kernel-level support and continuously
running daemon processes to provide NFS file sharing, however, NFS
support must be enabled in the Linux kernel in order to function. NFS
uses Remote Procedure Calls (RPC) to route
requests between clients and servers, meaning that the
portmap service must be enabled and active at the
proper runlevels for NFS communication to occur. Working with
portmap, the following processes ensure that a given
NFS connection is allowed and may proceed without error:
rpc.mountd — The running process that
receives the mount request from an NFS client and checks to see if
it matches with a currently exported file system.
rpc.nfsd — The process that implements
the user-space components of the NFS service. It works with the
Linux kernel to meet the dynamic demands of NFS clients, such as
providing additional server threads for NFS clients to use.
rpc.lockd — A daemon that is not necessary
with modern kernels. NFS file locking is now done by the kernel. It
is included with the nfs-utils package for
users of older kernels that do not include this functionality
by default.
rpc.statd — Implements the
Network Status Monitor (NSM) RPC
protocol. This provides reboot notification when an NFS server is
restarted without being gracefully brought down.
rpc.rquotad — An RPC server that provides
user quota information for remote users.
Not all of these programs are required for NFS service. The only
services that must be enabled are rpc.mountd,
rpc.nfsd, and portmap. The other
daemons provide additional functionality and should only be used if the
server environment requires them.
NFS version 2 uses the User Datagram Protocol
(UDP) to provide a stateless network connection between the
client and server. NFS version 3 can use UDP or TCP running over an
IP. The stateless UDP connection minimizes network traffic, as the NFS
server sends the client a cookie after the client is authorized to
access the shared volume. This cookie is a random value stored on the
server's side and is passed along with RPC requests from the client. The
NFS server can be restarted without affecting the clients and the cookie
will remain intact.
NFS only performs authentication when a client system attempts to mount
a remote file system. To limit access, the NFS server first employs TCP
wrappers. TCP wrappers reads the /etc/hosts.allow
and /etc/hosts.deny files to determine if a
particular client should be permitted or prevented access to the NFS
server. For more information on configuring access controls with TCP
wrappers, see Chapter 15 TCP Wrappers and xinetd.
After the client is granted access by TCP wrappers, the NFS server refers
to its configuration file, /etc/exports, to
determine whether the client can mount any of the exported file
systems. After granting access, any file and directory operations are
sent to the server using remote procedure calls.
 | Warning |
|---|
| | NFS mount privileges are granted specifically to a client, not a
user. Exported file systems can be a accessed by
any users on the remote machine.
When configuring the /etc/exports file, be very
careful when granting read-write permissions (rw) for
an exported file system.
|
9.1.1. NFS and portmap
NFS relies upon remote procedure calls (RPC) to function. The
portmap service is required to map RPC requests to
the correct services. RPC processes notify portmap
when they start, revealing the port number they are monitoring and the
RPC program numbers they expect to serve. The client system then
contacts portmap on the server with a particular
RPC program number. portmap then redirects the
client to the proper port number to communicate with its intended
service.
Because RPC-based services rely on portmap to make
all connections with incoming client requests,
portmap must be available before any of these
services start. If, for some reason, the portmap
service unexpectedly quits, restart portmap and any
services running when it was started.
The portmap service can be used with TCP wrappers'
hosts access files (/etc/hosts.allow and
/etc/hosts.deny) to control which remote systems
are permitted to use RPC-based services on the server. See Chapter 15 TCP Wrappers and xinetd for more information. Access control rules
for portmap will affect all RPC-based
services. Alternatively, it is possible to specify each of the NFS RPC
daemons to be affected by a particular access control rule. The man
pages for rpc.mountd and
rpc.statd contain information regarding the precise
syntax for these rules.
9.1.1.1. Trouble shooting NFS with portmap
As portmap provides the coordination between RPC
services and the port numbers used to communicate with them, it is
useful to be able to view the status of current RPC services using
portmap when troubleshooting. The
rpcinfo command shows each RPC-based service with
its port number, RPC program number, version, and IP protocol type
(TCP or UDP).
To make sure the proper NFS RPC-based services are enabled for
portmap, use the rpcinfo -p
command:
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 1024 status
100024 1 tcp 1024 status
100011 1 udp 819 rquotad
100011 2 udp 819 rquotad
100005 1 udp 1027 mountd
100005 1 tcp 1106 mountd
100005 2 udp 1027 mountd
100005 2 tcp 1106 mountd
100005 3 udp 1027 mountd
100005 3 tcp 1106 mountd
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
100021 1 udp 1028 nlockmgr
100021 3 udp 1028 nlockmgr
100021 4 udp 1028 nlockmgr |
The -p option probes the portmapper on the
specified host or defaults to localhost if no specific host
is listed. Other options are available from the
rpcinfo man page.
From this output, it is apparent that various NFS services are
running. If one of the NFS services does not start up correctly,
portmap will be unable to map RPC requests from
clients for that service to the correct port. In many cases,
restarting NFS as root (/sbin/service nfs
restart) will cause those service to correctly register
with portmap and begin working.
