1RPC.STATD(8) System Manager's Manual RPC.STATD(8)
2
6 rpc.statd - NSM service daemon
7
9 rpc.statd [-dh?FLNvV] [-H prog] [-n my-name] [-o outgoing-port]
10 [-p listener-port] [-P path]
11 [--nlm-port port] [--nlm-udp-port port]
12
14 File locks are not part of persistent file system state. Lock state is
15 thus lost when a host reboots.
16 Network file systems must also detect when lock state is lost because a
18 remote host has rebooted. After an NFS client reboots, an NFS server
19 must release all file locks held by applications that were running on
20 that client. After a server reboots, a client must remind the server
21 of file locks held by applications running on that client.
22 For NFS version 2 [RFC1094] and NFS version 3 [RFC1813], the Network
24 Status Monitor protocol (or NSM for short) is used to notify NFS peers
25 of reboots. On Linux, two separate user-space components constitute
26 the NSM service:
27 rpc.statd
29 A daemon that listens for reboot notifications from other hosts,
30 and manages the list of hosts to be notified when the local sys‐
31 tem reboots
32 sm-notify
34 A helper program that notifies NFS peers after the local system
35 reboots
36 The local NFS lock manager alerts its local rpc.statd of each remote
38 peer that should be monitored. When the local system reboots, the sm-
39 notify command notifies the NSM service on monitored peers of the
40 reboot. When a remote reboots, that peer notifies the local rpc.statd,
41 which in turn passes the reboot notification back to the local NFS lock
42 manager.
43
45 The first file locking interaction between an NFS client and server
46 causes the NFS lock managers on both peers to contact their local NSM
47 service to store information about the opposite peer. On Linux, the
48 local lock manager contacts rpc.statd.
49 rpc.statd records information about each monitored NFS peer on persis‐
51 tent storage. This information describes how to contact a remote peer
52 in case the local system reboots, how to recognize which monitored peer
53 is reporting a reboot, and how to notify the local lock manager when a
54 monitored peer indicates it has rebooted.
55 An NFS client sends a hostname, known as the client's caller_name, in
57 each file lock request. An NFS server can use this hostname to send
58 asynchronous GRANT calls to a client, or to notify the client it has
59 rebooted.
60 The Linux NFS server can provide the client's caller_name or the
62 client's network address to rpc.statd. For the purposes of the NSM
63 protocol, this name or address is known as the monitored peer's
64 mon_name. In addition, the local lock manager tells rpc.statd what it
65 thinks its own hostname is. For the purposes of the NSM protocol, this
66 hostname is known as my_name.
67 There is no equivalent interaction between an NFS server and a client
69 to inform the client of the server's caller_name. Therefore NFS
70 clients do not actually know what mon_name an NFS server might use in
71 an SM_NOTIFY request. The Linux NFS client uses the server hostname
72 from the mount command to identify rebooting NFS servers.
73 Reboot notification
75 When the local system reboots, the sm-notify command reads the list of
76 monitored peers from persistent storage and sends an SM_NOTIFY request
77 to the NSM service on each listed remote peer. It uses the mon_name
78 string as the destination. To identify which host has rebooted, the
79 sm-notify command sends the my_name string recorded when that remote
80 was monitored. The remote rpc.statd matches incoming SM_NOTIFY
81 requests using this string, or the caller's network address, to one or
82 more peers on its own monitor list.
83 If rpc.statd does not find a peer on its monitor list that matches an
85 incoming SM_NOTIFY request, the notification is not forwarded to the
86 local lock manager. In addition, each peer has its own NSM state num‐
87 ber, a 32-bit integer that is bumped after each reboot by the sm-notify
88 command. rpc.statd uses this number to distinguish between actual
89 reboots and replayed notifications.
90 Part of NFS lock recovery is rediscovering which peers need to be moni‐
92 tored again. The sm-notify command clears the monitor list on persis‐
93 tent storage after each reboot.
94
96 -d, --no-syslog
97 Causes rpc.statd to write log messages on stderr instead of to
98 the system log, if the -F option was also specified.
99 -F, --foreground
101 Keeps rpc.statd attached to its controlling terminal so that NSM
102 operation can be monitored directly or run under a debugger. If
103 this option is not specified, rpc.statd backgrounds itself soon
104 after it starts.
105 -h, -?, --help
107 Causes rpc.statd to display usage information on stderr and then
108 exit.
109 -H, --ha-callout prog
111 Specifies a high availability callout program. If this option
112 is not specified, no callouts are performed. See the High-
113 availability callouts section below for details.
114 -L, --no-notify
116 Prevents rpc.statd from running the sm-notify command when it
117 starts up, preserving the existing NSM state number and monitor
118 list.
119 Note: the sm-notify command contains a check to ensure it runs
121 only once after each system reboot. This prevents spurious
122 reboot notification if rpc.statd restarts without the -L option.
123 -n, --name ipaddr | hostname
125 Specifies the bind address used for RPC listener sockets. The
126 ipaddr form can be expressed as either an IPv4 or an IPv6 pre‐
127 sentation address. If this option is not specified, rpc.statd
128 uses a wildcard address as the transport bind address.
129 This string is also passed to the sm-notify command to be used
131 as the source address from which to send reboot notification
132 requests. See sm-notify(8) for details.
133 -N Causes rpc.statd to run the sm-notify command, and then exit.
135 Since the sm-notify command can also be run directly, this
136 option is deprecated.
137 -o, --outgoing-port port
139 Specifies the source port number the sm-notify command should
140 use when sending reboot notifications. See sm-notify(8) for
141 details.
142 -p, --port port
144 Specifies the port number used for RPC listener sockets. If
145 this option is not specified, rpc.statd will try to consult
146 /etc/services, if gets port succeed, set the same port for all
147 listener socket, otherwise chooses a random ephemeral port for
148 each listener socket.
149 This option can be used to fix the port value of its listeners
151 when SM_NOTIFY requests must traverse a firewall between clients
152 and servers.
153 -T, --nlm-port port
155 Specifies the port number that lockd should listen on for NLM
156 requests. This sets both the TCP and UDP ports unless the UDP
157 port is set separately.
158 -U, --nlm-udp-port port
160 Specifies the UDP port number that lockd should listen on for
161 NLM requests.
162 -P, --state-directory-path pathname
164 Specifies the pathname of the parent directory where NSM state
165 information resides. If this option is not specified, rpc.statd
166 uses /var/lib/nfs/statd by default.
167 After starting, rpc.statd attempts to set its effective UID and
169 GID to the owner and group of this directory.
170 -v, -V, --version
172 Causes rpc.statd to display version information on stderr and
173 then exit.
174
176 Many of the options that can be set on the command line can also be
177 controlled through values set in the [statd] or, in some cases, the
178 [lockd] sections of the /etc/nfs.conf configuration file. Values rec‐
179 ognized in the [statd] section include port, outgoing-port, name,
180 state-directory-path, and ha-callout which each have the same effect as
181 the option with the same name.
182 The values recognized in the [lockd] section include port and udp-port
184 which have the same effect as the --nlm-port and --nlm-udp-port
185 options, respectively.
186
189 The rpc.statd daemon must be started as root to acquire privileges
190 needed to create sockets with privileged source ports, and to access
191 the state information database. Because rpc.statd maintains a long-
192 running network service, however, it drops root privileges as soon as
193 it starts up to reduce the risk of a privilege escalation attack.
194 During normal operation, the effective user ID it chooses is the owner
196 of the state directory. This allows it to continue to access files in
197 that directory after it has dropped its root privileges. To control
198 which user ID rpc.statd chooses, simply use chown(1) to set the owner
199 of the state directory.
200
202 Lock recovery after a reboot is critical to maintaining data integrity
203 and preventing unnecessary application hangs. To help rpc.statd match
204 SM_NOTIFY requests to NLM requests, a number of best practices should
205 be observed, including:
206 The UTS nodename of your systems should match the DNS names that
208 NFS peers use to contact them
209 The UTS nodenames of your systems should always be fully quali‐
211 fied domain names
212 The forward and reverse DNS mapping of the UTS nodenames should
214 be consistent
215 The hostname the client uses to mount the server should match
217 the server's mon_name in SM_NOTIFY requests it sends
218 Unmounting an NFS file system does not necessarily stop either the NFS
220 client or server from monitoring each other. Both may continue moni‐
221 toring each other for a time in case subsequent NFS traffic between the
222 two results in fresh mounts and additional file locking.
223 On Linux, if the lockd kernel module is unloaded during normal opera‐
225 tion, all remote NFS peers are unmonitored. This can happen on an NFS
226 client, for example, if an automounter removes all NFS mount points due
227 to inactivity.
228 High-availability callouts
230 rpc.statd can exec a special callout program during processing of suc‐
231 cessful SM_MON, SM_UNMON, and SM_UNMON_ALL requests, or when it
232 receives SM_NOTIFY. Such a program may be used in High Availability
233 NFS (HA-NFS) environments to track lock state that may need to be
234 migrated after a system reboot.
235 The name of the callout program is specified with the -H option. The
237 program is run with 3 arguments: The first is either add-client del-
238 client or sm-notify depending on the reason for the callout. The sec‐
239 ond is the mon_name of the monitored peer. The third is the call‐
240 er_name of the requesting lock manager for add-client or del-client ,
241 otherwise it is IP_address of the caller sending SM_NOTIFY. The forth
242 is the state_value in the SM_NOTIFY request.
243 IPv6 and TI-RPC support
246 TI-RPC is a pre-requisite for supporting NFS on IPv6. If TI-RPC sup‐
247 port is built into rpc.statd, it attempts to start listeners on network
248 transports marked 'visible' in /etc/netconfig. As long as at least one
249 network transport listener starts successfully, rpc.statd will operate.
250
252 RPC_STATD_NO_NOTIFY=
253 If set to a positive integer, has the same effect as
254 --no-notify.
255
257 /var/lib/nfs/statd/sm directory containing monitor list
258 /var/lib/nfs/statd/sm.bak
260 directory containing notify list
261 /var/lib/nfs/statd/state NSM state number for this host
263 /var/run/run.statd.pid pid file
265 /etc/netconfig network transport capability database
267
269 sm-notify(8), nfs(5), rpc.nfsd(8), rpcbind(8), tcpd(8), iptables(8),
270 netconfig(5)
271 RFC 1094 - "NFS: Network File System Protocol Specification"
273 RFC 1813 - "NFS Version 3 Protocol Specification"
274 OpenGroup Protocols for Interworking: XNFS, Version 3W - Chapter 11
275
277 Jeff Uphoff <juphoff@users.sourceforge.net>
278 Olaf Kirch <okir@monad.swb.de>
279 H.J. Lu <hjl@gnu.org>
280 Lon Hohberger <hohberger@missioncriticallinux.com>
281 Paul Clements <paul.clements@steeleye.com>
282 Chuck Lever <chuck.lever@oracle.com>
283 1 November 2009 RPC.STATD(8)