1rpc.nfsd(8) System Manager's Manual rpc.nfsd(8)
2
6 rpc.nfsd - NFS server process
7
9 /usr/sbin/rpc.nfsd [options] nproc
10
12 The rpc.nfsd program implements the user level part of the NFS service.
13 The main functionality is handled by the nfsd kernel module. The user
14 space program merely specifies what sort of sockets the kernel service
15 should listen on, what NFS versions it should support, and how many
16 kernel threads it should use.
17 The rpc.mountd server provides an ancillary service needed to satisfy
19 mount requests by NFS clients.
20
22 -d or --debug
23 enable logging of debugging messages
24 -H or --host hostname
26 specify a particular hostname (or address) that NFS requests
27 will be accepted on. By default, rpc.nfsd will accept NFS re‐
28 quests on all known network addresses. Note that lockd (which
29 performs file locking services for NFS) may still accept request
30 on all known network addresses. This may change in future re‐
31 leases of the Linux Kernel. This option can be used multiple
32 times to listen to more than one interface.
33 or --scope scope
35 NFSv4.1 and later require the server to report a "scope" which
36 is used by the clients to detect if two connections are to the
37 same server. By default Linux NFSD uses the host name as the
38 scope.
39 It is particularly important for high-availablity configurations
41 to ensure that all potential server nodes report the same server
42 scope.
43 -p or --port port
45 specify a different port to listen on for NFS requests. By de‐
46 fault, rpc.nfsd will listen on port 2049.
47 -r or --rdma
49 specify that NFS requests on the standard RDMA port ("nfsrdma",
50 port 20049) should be honored.
51 --rdma=port
53 Listen for RDMA requests on an alternate port - may be a number
54 or a name listed in /etc/services.
55 -N or --no-nfs-version vers
57 This option can be used to request that rpc.nfsd does not offer
58 certain versions of NFS. The current version of rpc.nfsd can
59 support major NFS versions 3,4 and the minor versions 4.0, 4.1
60 and 4.2.
61 -s or --syslog
63 By default, rpc.nfsd logs error messages (and debug messages, if
64 enabled) to stderr. This option makes rpc.nfsd log these mes‐
65 sages to syslog instead. Note that errors encountered during op‐
66 tion processing will still be logged to stderr regardless of
67 this option.
68 -t or --tcp
70 Instruct the kernel nfs server to open and listen on a TCP
71 socket. This is the default.
72 -T or --no-tcp
74 Instruct the kernel nfs server not to open and listen on a TCP
75 socket.
76 -u or --udp
78 Instruct the kernel nfs server to open and listen on a UDP
79 socket.
80 -U or --no-udp
82 Instruct the kernel nfs server not to open and listen on a UDP
83 socket. This is the default.
84 -V or --nfs-version vers
86 This option can be used to request that rpc.nfsd offer certain
87 versions of NFS. The current version of rpc.nfsd can support ma‐
88 jor NFS versions 3,4 and the minor versions 4.0, 4.1 and 4.2.
89 -L or --lease-time seconds
91 Set the lease-time used for NFSv4. This corresponds to how of‐
92 ten clients need to confirm their state with the server. Valid
93 range is from 10 to 3600 seconds.
94 -G or --grace-time seconds
96 Set the grace-time used for NFSv4 and NLM (for NFSv2 and NFSv3).
97 New file open requests (NFSv4) and new file locks (NLM) will not
98 be allowed until after this time has passed to allow clients to
99 recover state.
100 nproc specify the number of NFS server threads. By default, eight
102 threads are started. However, for optimum performance several
103 threads should be used. The actual figure depends on the number
104 of and the work load created by the NFS clients, but a useful
105 starting point is eight threads. Effects of modifying that num‐
106 ber can be checked using the nfsstat(8) program.
107 Note that if the NFS server is already running, then the options for
109 specifying host, port, and protocol will be ignored. The number of
110 processes given will be the only option considered, and the number of
111 active nfsd processes will be increased or decreased to match this num‐
112 ber. In particular rpc.nfsd 0 will stop all threads and thus close any
113 open connections.
114
117 Many of the options that can be set on the command line can also be
118 controlled through values set in the [nfsd] section of the
119 /etc/nfs.conf configuration file. Values recognized include:
120 threads
122 The number of threads to start.
123 host A host name, or comma separated list of host names, that
125 rpc.nfsd will listen on. Use of the --host option replaces all
126 host names listed here.
127 scope Set the server scope.
129 grace-time
131 The grace time, for both NFSv4 and NLM, in seconds.
132 lease-time
134 The lease time for NFSv4, in seconds.
135 port Set the port for TCP/UDP to bind to.
137 rdma Enable RDMA port (with "on" or "yes" etc) on the standard port
139 ("nfsrdma", port 20049).
140 rdma-port
142 Set an alternate RDMA port.
143 UDP Enable (with "on" or "yes" etc) or disable ("off", "no") UDP
145 support.
146 TCP Enable or disable TCP support.
148 vers3
150 vers4 Enable or disable all NFSv4 versions. All versions are normally
152 enabled by default.
153 vers4.1
155 vers4.2
157 Setting these to "off" or similar will disable the selected mi‐
158 nor versions. Setting to "on" will enable them. The default
159 values are determined by the kernel, and usually minor versions
160 default to being enabled once the implementation is sufficiently
161 complete.
162
165 If the program is built with TI-RPC support, it will enable any proto‐
166 col and address family combinations that are marked visible in the net‐
167 config database.
168
171 nfsd(7), rpc.mountd(8), exports(5), exportfs(8), nfs.conf(5), rpc.rquo‐
172 tad(8), nfsstat(8), netconfig(5).
173
175 Olaf Kirch, Bill Hawes, H. J. Lu, G. Allan Morris III, and a host of
176 others.
177 20 Feb 2014 rpc.nfsd(8)