3 .\" Copyright (C) 1999 Olaf Kirch <okir@monad.swb.de>
4 .\" Modified by Jeffrey A. Uphoff, 1999, 2002, 2005.
5 .\" Modified by Lon Hohberger, 2000.
6 .\" Modified by Paul Clements, 2004.
8 .\" Rewritten by Chuck Lever <chuck.lever@oracle.com>, 2009.
9 .\" Copyright 2009 Oracle. All rights reserved.
11 .TH RPC.STATD 8 "1 November 2009
13 rpc.statd \- NSM service daemon
15 .BI "rpc.statd [-dh?FLNvV] [-H " prog "] [-n " my-name "] [-o " outgoing-port "] [-p " listener-port "] [-P " path " ]
17 File locks are not part of persistent file system state.
18 Lock state is thus lost when a host reboots.
20 Network file systems must also detect when lock state is lost
21 because a remote host has rebooted.
22 After an NFS client reboots, an NFS server must release all file locks
23 held by applications that were running on that client.
24 After a server reboots, a client must remind the
25 server of file locks held by applications running on that client.
27 For NFS version 2 [RFC1094] and NFS version 3 [RFC1813], the
28 .I Network Status Monitor
29 protocol (or NSM for short)
30 is used to notify NFS peers of reboots.
31 On Linux, two separate user-space components constitute the NSM service:
34 A daemon that listens for reboot notifications from other hosts, and
35 manages the list of hosts to be notified when the local system reboots
38 A helper program that notifies NFS peers after the local system reboots
40 The local NFS lock manager alerts its local
42 of each remote peer that should be monitored.
43 When the local system reboots, the
45 command notifies the NSM service on monitored peers of the reboot.
46 When a remote reboots, that peer notifies the local
48 which in turn passes the reboot notification
49 back to the local NFS lock manager.
50 .SH NSM OPERATION IN DETAIL
51 The first file locking interaction between an NFS client and server causes
52 the NFS lock managers on both peers to contact their local NSM service to
53 store information about the opposite peer.
54 On Linux, the local lock manager contacts
58 records information about each monitored NFS peer on persistent storage.
59 This information describes how to contact a remote peer
60 in case the local system reboots,
61 how to recognize which monitored peer is reporting a reboot,
62 and how to notify the local lock manager when a monitored peer
63 indicates it has rebooted.
65 An NFS client sends a hostname, known as the client's
67 in each file lock request.
68 An NFS server can use this hostname to send asynchronous GRANT
69 calls to a client, or to notify the client it has rebooted.
71 The Linux NFS server can provide the client's
73 or the client's network address to
75 For the purposes of the NSM protocol,
76 this name or address is known as the monitored peer's
78 In addition, the local lock manager tells
80 what it thinks its own hostname is.
81 For the purposes of the NSM protocol,
82 this hostname is known as
85 There is no equivalent interaction between an NFS server and a client
86 to inform the client of the server's
88 Therefore NFS clients do not actually know what
90 an NFS server might use in an SM_NOTIFY request.
91 The Linux NFS client uses the server hostname from the mount command
92 to identify rebooting NFS servers.
93 .SS Reboot notification
94 When the local system reboots, the
96 command reads the list of monitored peers from persistent storage and
97 sends an SM_NOTIFY request to the NSM service on each listed remote peer.
100 string as the destination.
101 To identify which host has rebooted, the
105 string recorded when that remote was monitored.
108 matches incoming SM_NOTIFY requests using this string,
109 or the caller's network address,
110 to one or more peers on its own monitor list.
114 does not find a peer on its monitor list that matches
115 an incoming SM_NOTIFY request,
116 the notification is not forwarded to the local lock manager.
117 In addition, each peer has its own
118 .IR "NSM state number" ,
119 a 32-bit integer that is bumped after each reboot by the
123 uses this number to distinguish between actual reboots
124 and replayed notifications.
126 Part of NFS lock recovery is rediscovering
127 which peers need to be monitored again.
130 command clears the monitor list on persistent storage after each reboot.
133 .BR -d , " --no-syslog
136 to write log messages on
138 instead of to the system log,
141 option was also specified.
143 .BR -F , " --foreground
146 attached to its controlling terminal so that NSM
147 operation can be monitored directly or run under a debugger.
148 If this option is not specified,
150 backgrounds itself soon after it starts.
152 .BR -h , " -?" , " --help
155 to display usage information on
159 .BI "\-H," "" " \-\-ha-callout " prog
160 Specifies a high availability callout program.
161 If this option is not specified, no callouts are performed.
163 .B High-availability callouts
164 section below for details.
166 .BR -L , " --no-notify
171 command when it starts up,
172 preserving the existing NSM state number and monitor list.
176 command contains a check to ensure it runs only once after each system reboot.
177 This prevents spurious reboot notification if
183 .BI "\-n, " "" "\-\-name " ipaddr " | " hostname
184 Specifies the bind address used for RPC listener sockets.
187 form can be expressed as either an IPv4 or an IPv6 presentation address.
188 If this option is not specified,
190 uses a wildcard address as the transport bind address.
192 This string is also passed to the
194 command to be used as the source address from which
195 to send reboot notification requests.
205 command, and then exit.
208 command can also be run directly, this option is deprecated.
210 .BI "\-o," "" " \-\-outgoing\-port " port
211 Specifies the source port number the
213 command should use when sending reboot notifications.
218 .BI "\-p," "" " \-\-port " port
219 Specifies the port number used for RPC listener sockets.
220 If this option is not specified,
222 chooses a random ephemeral port for each listener socket.
224 This option can be used to fix the port value of its listeners when
225 SM_NOTIFY requests must traverse a firewall between clients and servers.
227 .BI "\-P, " "" \-\-state\-directory\-path " pathname
228 Specifies the pathname of the parent directory
229 where NSM state information resides.
230 If this option is not specified,
238 attempts to set its effective UID and GID to the owner
239 and group of this directory.
241 .BR -v ", " -V ", " --version
244 to display version information on
250 daemon must be started as root to acquire privileges needed
251 to create sockets with privileged source ports, and to access the
252 state information database.
255 maintains a long-running network service, however, it drops root privileges
256 as soon as it starts up to reduce the risk of a privilege escalation attack.
258 During normal operation,
259 the effective user ID it chooses is the owner of the state directory.
260 This allows it to continue to access files in that directory after it
261 has dropped its root privileges.
262 To control which user ID
269 You can also protect your
277 library, add the hostnames of peers that should be allowed access to
278 .IR /etc/hosts.allow .
283 binary has a different filename.
285 For further information see the
291 Lock recovery after a reboot is critical to maintaining data integrity
292 and preventing unnecessary application hangs.
295 match SM_NOTIFY requests to NLM requests, a number of best practices
296 should be observed, including:
298 The UTS nodename of your systems should match the DNS names that NFS
299 peers use to contact them
301 The UTS nodenames of your systems should always be fully qualified domain names
303 The forward and reverse DNS mapping of the UTS nodenames should be
306 The hostname the client uses to mount the server should match the server's
308 in SM_NOTIFY requests it sends
310 Unmounting an NFS file system does not necessarily stop
311 either the NFS client or server from monitoring each other.
312 Both may continue monitoring each other for a time in case subsequent
313 NFS traffic between the two results in fresh mounts and additional
318 kernel module is unloaded during normal operation,
319 all remote NFS peers are unmonitored.
320 This can happen on an NFS client, for example,
321 if an automounter removes all NFS mount
322 points due to inactivity.
323 .SS High-availability callouts
325 can exec a special callout program during processing of
326 successful SM_MON, SM_UNMON, and SM_UNMON_ALL requests.
327 Such a program may be used in High Availability NFS (HA-NFS)
328 environments to track lock state that may need to be migrated after
331 The name of the callout program is specified with the
334 The program is run with 3 arguments:
339 depending on the reason for the callout.
342 of the monitored peer.
345 of the requesting lock manager.
346 .SS IPv6 and TI-RPC support
347 TI-RPC is a pre-requisite for supporting NFS on IPv6.
348 If TI-RPC support is built into
350 it attempts to start listeners on network transports marked
353 As long as at least one network transport listener starts successfully,
359 directory containing monitor list
361 .I /var/lib/nfs/sm.bak
362 directory containing notify list
364 .I /var/lib/nfs/state
365 NSM state number for this host
367 .I /var/run/run.statd.pid
371 network transport capability database
378 .BR hosts_access (5),
382 RFC 1094 - "NFS: Network File System Protocol Specification"
384 RFC 1813 - "NFS Version 3 Protocol Specification"
386 OpenGroup Protocols for Interworking: XNFS, Version 3W - Chapter 11
388 Jeff Uphoff <juphoff@users.sourceforge.net>
390 Olaf Kirch <okir@monad.swb.de>
392 H.J. Lu <hjl@gnu.org>
394 Lon Hohberger <hohberger@missioncriticallinux.com>
396 Paul Clements <paul.clements@steeleye.com>
398 Chuck Lever <chuck.lever@oracle.com>