xref: /illumos-gate/usr/src/man/man8/share_nfs.8 (revision 1a2d662a91cee3bf82f41cd47c7ae6f3825d9db2)
1.\"
2.\" CDDL HEADER START
3.\"
4.\" The contents of this file are subject to the terms of the
5.\" Common Development and Distribution License (the "License").
6.\" You may not use this file except in compliance with the License.
7.\"
8.\" You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9.\" or http://www.opensolaris.org/os/licensing.
10.\" See the License for the specific language governing permissions
11.\" and limitations under the License.
12.\"
13.\" When distributing Covered Code, include this CDDL HEADER in each
14.\" file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15.\" If applicable, add the following below this CDDL HEADER, with the
16.\" fields enclosed by brackets "[]" replaced with your own identifying
17.\" information: Portions Copyright [yyyy] [name of copyright owner]
18.\"
19.\" CDDL HEADER END
20.\"
21.\"
22.\" Copyright (C) 2008, Sun Microsystems, Inc. All Rights Reserved
23.\" Copyright 2014 Nexenta Systems, Inc.  All rights reserved.
24.\" Copyright 2016 Jason King.
25.\"
26.Dd November 22, 2021
27.Dt SHARE_NFS 8
28.Os
29.Sh NAME
30.Nm share_nfs
31.Nd make local NFS file systems available for mounting by remote systems
32.Sh SYNOPSIS
33.Nm share
34.Op Fl d Ar description
35.Op Fl F Sy nfs
36.Op Fl o Ar specific_options
37.Ar pathname
38.Sh DESCRIPTION
39The
40.Nm share
41utility makes local file systems available for mounting by remote systems.
42It starts the
43.Xr nfsd 8
44and
45.Xr mountd 8
46daemons if they are not already running.
47.Pp
48If no argument is specified, then
49.Nm share
50displays all file systems currently shared, including NFS file systems and file
51systems shared through other distributed file system packages.
52.Sh OPTIONS
53The following options are supported:
54.Bl -tag -width "indented"
55.It Fl d Ar description
56Provide a comment that describes the file system to be shared.
57.It Fl F Sy nfs
58Share NFS file system type.
59.It Fl o Ar specific_options
60Specify
61.Ar specific_options
62in a comma-separated list of keywords and attribute-value-assertions for
63interpretation by the file-system-type-specific command.
64If
65.Ar specific_options
66is not specified, then by default sharing is read-write to all clients.
67.Ar specific_options
68can be any combination of the following:
69.Bl -tag -width "indented"
70.It Sy aclok
71Allows the NFS server to do access control for NFS Version 2 clients (running
72SunOS 2.4 or earlier).
73When
74.Sy aclok
75is set on the server, maximal access is given to all clients.
76For example, with
77.Sy aclok
78set, if anyone has read permissions, then everyone does.
79If
80.Sy aclok
81is not set, minimal access is given to all clients.
82.It Sy anon Ns = Ns Ar uid
83Set
84.Ar uid
85to be the effective user ID of unknown users.
86By default, unknown users are given the effective user ID UID_NOBODY.
87If uid is set to -1, access is denied.
88.It Ar charset Ns = Ns Ar access_list
89Where
90.Ar charset
91is one of: euc-cn, euc-jp, euc-jpms, euc-kr, euc-tw, iso8859-1, iso8859-2,
92iso8859-5, iso8859-6, iso8859-7, iso8859-8, iso8859-9, iso8859-13, iso8859-15,
93koi8-r.
94.Pp
95Clients that match the
96.Ar access_list
97for one of these properties will be assumed to be using that character set and
98file and path names will be converted to UTF-8 for the server.
99.It Sy gidmap Ns = Ns Ar mapping Ns Oo ~ Ns Ar mapping Oc Ns ...
100Where
101.Ar mapping
102is:
103.Oo Ar clnt Oc : Ns Oo Ar srv Oc : Ns Ar access_list
104.Pp
105Allows remapping the group ID (gid) in the incoming request to some other gid.
106This effectively changes the identity of the user in the request to that of
107some other local user.
108.Pp
109For clients where the gid in the incoming request is
110.Ar clnt
111and the client matches the
112.Ar access_list ,
113change the group ID to
114.Ar srv .
115If
116.Ar clnt
117is asterisk (*), all groups are mapped by this rule.
118If
119.Ar clnt
120is omitted, all unknown groups are mapped by this rule.
121If
122.Ar srv
123is set to -1, access is denied.
124If
125.Ar srv
126is omitted, the gid is mapped to UID_NOBODY.
127.Pp
128The particular
129.Ar mapping Ns s
130are separated in the
131.Sy gidmap Ns =
132option by tilde (~) and are evaluated in the specified order until a match is
133found.
134Both
135.Sy root Ns =
136and
137.Sy root_mapping Ns =
138options (if specified) are evaluated before the
139.Sy gidmap Ns =
140option.
141The
142.Sy gidmap Ns =
143option is skipped in the case where the client matches the
144.Sy root Ns =
145option.
146.Pp
147The
148.Sy gidmap Ns =
149option is evaluated before the
150.Sy anon Ns =
151option.
152.Pp
153This option is supported only for AUTH_SYS.
154.It Sy index Ns = Ns Ar file
155Load
156.Ar file
157rather than a listing of the directory containing this file when the
158directory is referenced by an NFS URL.
159.It Sy log Ns Oo = Ns Ar tag Oc
160Enables NFS server logging for the specified file system.
161The optional
162.Ar tag
163determines the location of the related log files.
164The
165.Ar tag
166is defined in
167.Pa /etc/nfs/nfslog.conf .
168If no
169.Ar tag
170is specified, the default values associated with the global tag in
171.Pa /etc/nfs/nfslog.conf
172are used.
173Support of NFS server logging is only available for NFS Version 2 and
174Version 3 requests.
175.It Sy nohide
176By default, if server exports two filesystems, one of which is mounted as a
177child of the other, NFS Version 2 and Version 3 clients must mount both
178filesystems explicitly in order to access them.
179If a client only mounts the parent, it will see an empty directory at the
180location where the other filesystem is mounted.
181.Pp
182Setting the
183.Sy nohide
184option on a filesystem causes it to no longer be hidden in this manner, and the
185client will be able to move from the parent filesystem to this one without
186noticing the change.
187However, some NFS clients or applications may not function correctly when
188this option is used.
189In particular, files on different underlying filesystems may appear to have
190the same inode numbers.
191The
192.Sy nohide
193option only applies to NFS Version 2 and Version 3 requests.
194.It Sy noaclfab
195By default, the NFS server will fabricate POSIX-draft style ACLs in response
196to ACL requests from NFS Version 2 or Version 3 clients accessing shared
197file systems that do not support POSIX-draft ACLs (such as ZFS).
198Specifying
199.Sy noaclfab
200disables this behavior.
201.It Sy none Ns = Ns Ar access_list
202Access is not allowed to any client that matches the access list.
203The exception is when the access list is an asterisk (*), in which case
204.Sy ro
205or
206.Sy rw
207can override
208.Sy none .
209.It Sy nosub
210Prevents clients from mounting subdirectories of shared directories.
211For example, if
212.Pa /export
213is shared with the
214.Sy nosub
215option on server
216.Qq fooey
217then a NFS client cannot do:
218.Bd -literal -offset indent
219mount -F nfs fooey:/export/home/mnt
220.Ed
221.Pp
222NFS Version 4 does not use the MOUNT protocol.
223The
224.Sy nosub
225option only applies to NFS Version 2 and Version 3 requests.
226.It Sy nosuid
227By default, clients are allowed to create files on the shared file system with
228the setuid or setgid mode enabled.
229Specifying
230.Sy nosuid
231causes the server file system to silently ignore any attempt to enable the
232setuid or setgid mode bits.
233.It Sy public
234Moves the location of the public file handle from root
235.Pa ( / )
236to the exported directory for WebNFS-enabled browsers and clients.
237This option does not enable WebNFS service; WebNFS is always on.
238Only one file system per server may use this option.
239Any other option, including the
240.Sy ro Ns = Ns Ar list
241and
242.Sy rw Ns = Ns Ar list
243options can be included with the
244.Sy public
245option.
246.It Sy ro
247Sharing is read-only to all clients.
248.It Sy ro Ns = Ns Ar access_list
249Sharing is read-only to the clients listed in
250.Ar access_list ;
251overrides the
252.Sy rw
253suboption for the clients specified.
254See
255.Sx access_list
256below.
257.It Sy root Ns = Ns Ar access_list
258Only root users from the hosts specified in
259.Ar access_list
260have root access.
261See
262.Sx access_list
263below.
264By default, no host has root access, so root users are mapped to an anonymous
265user ID (see the
266.Sy anon Ns = Ns Ar uid
267option described above).
268Netgroups can be used if the file system shared is using UNIX authentication
269(AUTH_SYS).
270.It Sy root_mapping Ns = Ns Ar uid
271For a client that is allowed root access, map the root UID to the specified
272user id.
273.It Sy rw
274Sharing is read-write to all clients.
275.It Sy rw Ns = Ns Ar access_list
276Sharing is read-write to the clients listed in
277.Ar access_list ;
278overrides the
279.Sy ro
280suboption for the clients specified.
281See
282.Sx access_list
283below.
284.It Sy sec Ns = Ns Ar mode Ns Oo : Ns Ar mode Oc Ns ...
285Sharing uses one or more of the specified security modes.
286The
287.Ar mode
288in the
289.Sy sec Ns = Ns Ar mode
290option must be a mode name supported on the client.
291If the
292.Sy sec Ns =
293option is not specified, the default security mode used is AUTH_SYS.
294Multiple
295.Sy sec Ns =
296options can be specified on the command line, although each mode can appear
297only once.
298The security modes are defined in
299.Xr nfssec 7 .
300.Pp
301Each
302.Sy sec Ns =
303option specifies modes that apply to any subsequent
304.Sy window Ns = ,
305.Sy rw ,
306.Sy ro ,
307.Sy rw Ns = ,
308.Sy ro Ns = ,
309and
310.Sy root Ns =
311options that are provided before another
312.Sy sec Ns =
313option.
314Each additional
315.Sy sec Ns =
316resets the security mode context, so that more
317.Sy window Ns = ,
318.Sy rw ,
319.Sy ro ,
320.Sy rw Ns = ,
321.Sy ro Ns = ,
322and
323.Sy root Ns =
324options can be supplied for additional modes.
325.It Sy sec Ns = Ns Sy none
326If the option
327.Sy sec Ns = Ns Sy none
328is specified when the client uses AUTH_NONE, or if the client uses a security
329mode that is not one that the file system is shared with, then the credential
330of each NFS request is treated as unauthenticated.
331See the
332.Sy anon Ns = Ns Ar uid
333option for a description of how unauthenticated requests are handled.
334.It Sy secure
335This option has been deprecated in favor of the
336.Sy sec Ns = Ns Sy dh
337option.
338.It Sy uidmap Ns = Ns Ar mapping Ns Oo ~ Ns Ar mapping Oc Ns ...
339Where
340.Ar mapping
341is:
342.Oo Ar clnt Oc : Ns Oo Ar srv Oc : Ns Ar access_list
343.Pp
344Allows remapping the user ID (uid) in the incoming request to some other uid.
345This effectively changes the identity of the user in the request to that of
346some other local user.
347.Pp
348For clients where the uid in the incoming request is
349.Ar clnt
350and the client matches the
351.Ar access_list ,
352change the user ID to
353.Ar srv .
354If
355.Ar clnt
356is asterisk (*), all users are mapped by this rule.
357If
358.Ar clnt
359is omitted, all unknown users are mapped by this rule.
360If
361.Ar srv
362is set to -1, access is denied.
363If
364.Ar srv
365is omitted, the uid is mapped to UID_NOBODY.
366.Pp
367The particular
368.Ar mapping Ns s
369are separated in the
370.Sy uidmap Ns =
371option by tilde (~) and are evaluated in the specified order until a match is
372found.
373Both
374.Sy root Ns =
375and
376.Sy root_mapping Ns =
377options (if specified) are evaluated before the
378.Sy uidmap Ns =
379option.
380The
381.Sy uidmap Ns =
382option is skipped in the case where the client matches the
383.Sy root Ns =
384option.
385.Pp
386The
387.Sy uidmap Ns =
388option is evaluated before the
389.Sy anon Ns =
390option.
391.Pp
392This option is supported only for AUTH_SYS.
393.It Sy window Ns = Ns Ar value
394When sharing with
395.Sy sec Ns = Ns Sy dh ,
396set the maximum life time (in seconds) of the RPC request's credential (in the
397authentication header) that the NFS server allows.
398If a credential arrives with a life time larger than what is allowed, the NFS
399server rejects the request.
400The default value is 30000 seconds (8.3 hours).
401.El
402.El
403.Ss access_list
404The
405.Ar access_list
406argument is a colon-separated list whose components may be any number of the
407following:
408.Bl -tag -width "indented"
409.It Sy hostname
410The name of a host.
411With a server configured for DNS or LDAP naming in the nsswitch
412.Sy hosts
413entry, any hostname must be represented as a fully qualified DNS or LDAP name.
414.It Sy netgroup
415A netgroup contains a number of hostnames.
416With a server configured for DNS or LDAP naming in the nsswitch
417.Sy hosts
418entry, any hostname in a netgroup must be represented as a fully qualified DNS
419or LDAP name.
420.It Sy domain name suffix
421To use domain membership the server must use DNS or LDAP to resolve hostnames to
422IP addresses; that is, the
423.Sy hosts
424entry in the
425.Pa /etc/nsswitch.conf
426must specify
427.Sy dns
428or
429.Sy ldap
430ahead of
431.Sy nis
432since only DNS and LDAP return the full domain name of the host.
433Other name services like NIS cannot be used to resolve hostnames on the server
434because when mapping an IP address to a hostname they do not return domain
435information.
436For example,
437.Bd -literal -offset indent
438NIS   172.16.45.9 --> "myhost"
439.Ed
440.Pp
441and
442.Bd -literal -offset indent
443DNS or LDAP   172.16.45.9 --> "myhost.mydomain.example.com"
444.Ed
445.Pp
446The domain name suffix is distinguished from hostnames and netgroups by a
447prefixed dot.
448For example,
449.Bd -literal -offset indent
450rw=.mydomain.example.com
451.Ed
452.Pp
453A single dot can be used to match a hostname with no suffix.
454For example,
455.Bd -literal -offset indent
456rw=.
457.Ed
458.Pp
459matches
460.Qq mydomain
461but not
462.Qq mydomain.example.com .
463This feature can be used to match hosts resolved through NIS rather
464than DNS and LDAP.
465.It Sy network
466The network or subnet component is preceded by an at-sign (@).
467It can be either a name or a dotted address.
468If a name, it is converted to a dotted address by
469.Xr getnetbyname 3SOCKET .
470For example,
471.Bd -literal -offset indent
472=@mynet
473.Ed
474.Pp
475would be equivalent to:
476.Bd -literal -offset indent
477=@172.16 or =@172.16.0.0
478.Ed
479.Pp
480The network prefix assumes an octet-aligned netmask determined from the zeroth
481octet in the low-order part of the address up to and including the high-order
482octet, if you want to specify a single IP address (see below).
483In the case where network prefixes are not byte-aligned, the syntax allows a
484mask length to be specified explicitly following a slash (/) delimiter.
485For example,
486.Bd -literal -offset indent
487=@theothernet/17 or =@172.16.132/22
488.Ed
489.Pp
490where the mask is the number of leftmost contiguous significant bits in the
491corresponding IP address.
492.Pp
493When specifying individual IP addresses, use the same @ notation described
494above, without a netmask specification.
495For example:
496.Bd -literal -offset indent
497=@172.16.132.14
498.Ed
499.Pp
500Multiple, individual IP addresses would be specified, for example, as:
501.Bd -literal -offset indent
502root=@172.16.132.20:@172.16.134.20
503.Ed
504.El
505.Pp
506A prefixed minus sign (-) denies access to that component of
507.Ar access_list .
508The list is searched sequentially until a match is found that either grants or
509denies access, or until the end of the list is reached.
510For example, if host
511.Qq terra
512is in the
513.Qq engineering
514netgroup, then
515.Bd -literal -offset indent
516rw=-terra:engineering
517.Ed
518.Pp
519denies access to
520.Qq terra
521but
522.Bd -literal -offset indent
523rw=engineering:-terra
524.Ed
525.Pp
526grants access to
527.Qq terra .
528.Sh OPERANDS
529The following operands are supported:
530.Bl -tag -width "pathname"
531.It Sy pathname
532The pathname of the file system to be shared.
533.El
534.Sh FILES
535.Bl -tag -width "/etc/nfs/nfslog.conf"
536.It Pa /etc/dfs/fstypes
537list of system types, NFS by default
538.It Pa /etc/dfs/sharetab
539system record of shared file systems
540.It Pa /etc/nfs/nfslogtab
541system record of logged file systems
542.It Pa /etc/nfs/nfslog.conf
543logging configuration file
544.El
545.Sh EXIT STATUS
546.Ex -std
547.Sh EXAMPLES
548.Ss Example 1 Sharing A File System With Logging Enabled
549The following example shows the
550.Pa /export
551file system shared with logging enabled:
552.Bd -literal -offset indent
553share -o log /export
554.Ed
555.Pp
556The default global logging parameters are used since no tag identifier is
557specified.
558The location of the log file, as well as the necessary logging work
559files, is specified by the global entry in
560.Pa /etc/nfs/nfslog.conf .
561The
562.Xr nfslogd 8
563daemon runs only if at least one file system entry in
564.Pa /etc/dfs/dfstab
565is shared with logging enabled upon starting or rebooting the system.
566Simply sharing a file system with logging enabled from the command line does not
567start the
568.Xr nfslogd 8 .
569.Ss Example 2 Remap A User Coming From The Particular NFS Client
570The following example remaps the user with uid
571.Sy 100
572at client
573.Sy 10.0.0.1
574to user
575.Sy joe :
576.Bd -literal -offset indent
577share -o uidmap=100:joe:@10.0.0.1 /export
578.Ed
579.Sh SEE ALSO
580.Xr getnetbyname 3SOCKET ,
581.Xr netgroup 5 ,
582.Xr nfslog.conf 5 ,
583.Xr acl 7 ,
584.Xr attributes 7 ,
585.Xr nfssec 7 ,
586.Xr mount 8 ,
587.Xr mountd 8 ,
588.Xr nfsd 8 ,
589.Xr nfslogd 8 ,
590.Xr share 8 ,
591.Xr unshare 8
592.Sh NOTES
593If the
594.Sy sec Ns =
595option is presented at least once, all uses of the
596.Sy window Ns = ,
597.Sy rw ,
598.Sy ro ,
599.Sy rw Ns = ,
600.Sy ro Ns = ,
601and
602.Sy root Ns =
603options must come after the first
604.Sy sec Ns =
605option.
606If the
607.Sy sec Ns =
608option is not presented, then
609.Sy sec Ns = Ns Sy sys
610is implied.
611.Pp
612If one or more explicit
613.Sy sec Ns =
614options are presented,
615.Sy sys
616must appear in one of the options mode lists for accessing using the AUTH_SYS
617security mode to be allowed.
618For example:
619.Bd -literal -offset indent
620share -F nfs /var
621share -F nfs -o sec=sys /var
622.Ed
623.Pp
624grants read-write access to any host using AUTH_SYS, but
625.Bd -literal -offset indent
626share -F nfs -o sec=dh /var
627.Ed
628.Pp
629grants no access to clients that use AUTH_SYS.
630.Pp
631Unlike previous implementations of
632.Nm ,
633access checking for the
634.Sy window Ns = ,
635.Sy rw ,
636.Sy ro ,
637.Sy rw Ns = ,
638and
639.Sy ro Ns =
640options is done per NFS request, instead of per mount request.
641.Pp
642Combining multiple security modes can be a security hole in situations where
643the
644.Sy ro Ns =
645and
646.Sy rw Ns =
647options are used to control access to weaker security modes.
648In this example,
649.Bd -literal -offset indent
650share -F nfs -o sec=dh,rw,sec=sys,rw=hosta /var
651.Ed
652.Pp
653an intruder can forge the IP address for
654.Qq hosta
655(albeit on each NFS request) to side-step the stronger controls of AUTH_DES.
656Something like:
657.Bd -literal -offset indent
658share -F nfs -o sec=dh,rw,sec=sys,ro /var
659.Ed
660.Pp
661is safer, because any client (intruder or legitimate) that avoids AUTH_DES only
662gets read-only access.
663In general, multiple security modes per share command should only be used in
664situations where the clients using more secure modes get stronger access than
665clients using less secure modes.
666.Pp
667If
668.Sy rw Ns =
669and
670.Sy ro Ns =
671options are specified in the same
672.Sy sec Ns =
673clause, and a client is in both lists, the order of the two options determines
674the access the client gets.
675If client
676.Qq hosta
677is in two netgroups,
678.Qq group1
679and
680.Qq group2 ,
681in this example, the client would get read-only access:
682.Bd -literal -offset indent
683share -F nfs -o ro=group1,rw=group2 /var
684.Ed
685.Pp
686In this example
687.Qq hosta
688would get read-write access:
689.Bd -literal -offset indent
690share -F nfs -o rw=group2,ro=group1 /var
691.Ed
692.Pp
693If within a
694.Sy sec Ns =
695clause, both the
696.Sy ro
697and
698.Sy rw Ns =
699options are specified, for compatibility, the order of the options rule is not
700enforced.
701All hosts would get read-only access, with the exception to those in the
702read-write list.
703Likewise, if the
704.Sy ro Ns =
705and
706.Sy rw
707options are specified, all hosts get read-write access with the exceptions of
708those in the read-only list.
709.Pp
710The
711.Sy ro Ns =
712and
713.Sy rw Ns =
714options are guaranteed to work over UDP and TCP but may not work over other
715transport providers.
716.Pp
717The
718.Sy root Ns =
719option with AUTH_SYS is guaranteed to work over UDP and TCP but may not work
720over other transport providers.
721.Pp
722The
723.Sy root Ns =
724option with AUTH_DES is guaranteed to work over any transport provider.
725.Pp
726There are no interactions between the
727.Sy root Ns =
728option and the
729.Sy rw ,
730.Sy ro ,
731.Sy rw Ns = ,
732and
733.Sy ro Ns =
734options.
735Putting a host in the root list does not override the semantics of the other
736options.
737The access the host gets is the same as when the
738.Sy root Ns =
739option is absent.
740For example, the following share command denies access to
741.Qq hostb :
742.Bd -literal -offset indent
743share -F nfs -o ro=hosta,root=hostb /var
744.Ed
745.Pp
746The following gives read-only permissions to
747.Qq hostb :
748.Bd -literal -offset indent
749share -F nfs -o ro=hostb,root=hostb /var
750.Ed
751.Pp
752The following gives read-write permissions to
753.Qq hostb :
754.Bd -literal -offset indent
755share -F nfs -o ro=hosta,rw=hostb,root=hostb /var
756.Ed
757.Pp
758If the file system being shared is a symbolic link to a valid pathname, the
759canonical path (the path which the symbolic link follows) is shared.
760For example, if
761.Pa /export/foo
762is a symbolic link to
763.Pa /export/bar ,
764the following share command results in
765.Pa /export/bar
766as the shared pathname (and not
767.Pa /export/foo ) :
768.Bd -literal -offset indent
769share -F nfs /export/foo
770.Ed
771.Pp
772An NFS mount of
773.Lk server:/export/foo
774results in
775.Lk server:/export/bar
776really being mounted.
777.Pp
778This line in the
779.Pa /etc/dfs/dfstab
780file shares the
781.Pa /disk
782file system read-only at boot time:
783.Bd -literal -offset indent
784share -F nfs -o ro /disk
785.Ed
786.Pp
787The
788.Xr mountd 8
789process allows the processing of a path name that contains a symbolic link.
790This allows the processing of paths that are not themselves explicitly shared
791with
792.Nm .
793For example,
794.Pa /export/foo
795might be a symbolic link that refers to
796.Pa /export/bar
797which has been specifically shared.
798When the client mounts
799.Pa /export/foo
800the mountd processing follows the symbolic link and responds with the
801.Pa /export/bar .
802The NFS Version 4 protocol does not use the mountd processing and the client's
803use of
804.Pa /export/foo
805does not work as it does with NFS Version 2 and Version 3 and the client
806receives an error when attempting to mount
807.Pa /export/foo .
808.Pp
809The
810.Sy nohide
811option violates RFC 1094,
812.%T "Network File System Protocol Specification"
813and RFC 1813,
814.%T "NFS: Network File System Version 3 Protocol Specification"
815.Pp
816The
817.Sy nohide
818option is provided for compatibility with Linux NFS.
819