1 .\" Copyright (c) 2018 Rick Macklem
3 .\" Redistribution and use in source and binary forms, with or without
4 .\" modification, are permitted provided that the following conditions
6 .\" 1. Redistributions of source code must retain the above copyright
7 .\" notice, this list of conditions and the following disclaimer.
8 .\" 2. Redistributions in binary form must reproduce the above copyright
9 .\" notice, this list of conditions and the following disclaimer in the
10 .\" documentation and/or other materials provided with the distribution.
12 .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
13 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
14 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
15 .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
16 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
17 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
18 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
19 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
20 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
21 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 .Nd NFS Version 4.1 and 4.2 Parallel NFS Protocol Server
35 servers may be configured to provide a
40 system needs to be configured as a MetaData Server (MDS) and
41 at least one additional
43 system needs to be configured as one or
44 more Data Servers (DS)s.
48 systems are configured to be NFSv4.1 and NFSv4.2
53 if you are not familiar with configuring a NFSv4.n server.
54 All DS(s) and the MDS should support NFSv4.2 as well as NFSv4.1.
55 Mixing an MDS that supports NFSv4.2 with any DS(s) that do not support
56 NFSv4.2 will not work correctly.
57 As such, all DS(s) must be upgraded from
61 before upgrading the MDS.
62 .Sh DS server configuration
63 The DS(s) need to be configured as NFSv4.1 and NFSv4.2 server(s),
64 with a top level exported
65 directory used for storage of data files.
66 This directory must be owned by
68 and would normally have a mode of
70 Within this directory there needs to be additional directories named
71 ds0,...,dsN (where N is 19 by default) also owned by
75 These are the directories where the data files are stored.
76 The following command can be run by root when in the top level exported
77 directory to create these subdirectories.
78 .Bd -literal -offset indent
79 jot -w ds 20 0 | xargs mkdir -m 700
84 is the default and can be set to a larger value on the MDS as shown below.
86 The top level exported directory used for storage of data files must be
87 exported to the MDS with the
88 .Dq maproot=root sec=sys
89 export options so that the MDS can create entries in these subdirectories.
90 It must also be exported to all pNFS aware clients, but these clients do
93 export option and this directory should be exported to them with the same
94 options as used by the MDS to export file system(s) to the clients.
96 It is possible to have multiple DSs on the same
99 of these DSs must have a separate top level exported directory used for storage
100 of data files and each
101 of these DSs must be mountable via a separate IP address.
102 Alias addresses can be set on the DS server system for a network
105 to create these different IP addresses.
106 Multiple DSs on the same server may be useful when data for different file systems
107 on the MDS are being stored on different file system volumes on the
110 .Sh MDS server configuration
111 The MDS must be a separate
117 It is configured as a NFSv4.1 and NFSv4.2 server with
118 file system(s) exported to clients.
121 command line argument for
123 is used to indicate that it is running as the MDS for a pNFS server.
125 The DS(s) must all be mounted on the MDS using the following mount options:
126 .Bd -literal -offset indent
127 nfsv4,minorversion=2,soft,retrans=2
130 so that they can be defined as DSs in the
133 Normally these mounts would be entered in the
136 For example, if there are four DSs named nfsv4-data[0-3], the
138 lines might look like:
140 nfsv4-data0:/ /data0 nfs rw,nfsv4,minorversion=2,soft,retrans=2 0 0
141 nfsv4-data1:/ /data1 nfs rw,nfsv4,minorversion=2,soft,retrans=2 0 0
142 nfsv4-data2:/ /data2 nfs rw,nfsv4,minorversion=2,soft,retrans=2 0 0
143 nfsv4-data3:/ /data3 nfs rw,nfsv4,minorversion=2,soft,retrans=2 0 0
150 indicates that the NFS server is a pNFS MDS and specifies what
157 nfs_server_flags line in your
161 nfs_server_flags="-u -t -n 128 -p nfsv4-data0:/data0,nfsv4-data1:/data1,nfsv4-data2:/data2,nfsv4-data3:/data3"
164 This example specifies that the data files should be distributed over the
165 four DSs and File layouts will be issued to pNFS enabled clients.
166 If issuing Flexible File layouts is desired for this case, setting the sysctl
167 .Dq vfs.nfsd.default_flexfile
174 Alternately, this variant of
176 will specify that two way mirroring is to be done, via the
180 nfs_server_flags="-u -t -n 128 -p nfsv4-data0:/data0,nfsv4-data1:/data1,nfsv4-data2:/data2,nfsv4-data3:/data3 -m 2"
183 With two way mirroring, the data file for each exported file on the MDS
184 will be stored on two of the DSs.
185 When mirroring is enabled, the server will always issue Flexible File layouts.
187 It is also possible to specify which DSs are to be used to store data files for
188 specific exported file systems on the MDS.
189 For example, if the MDS has exported two file systems
193 to clients, the following variant of
195 will specify that data files for
197 will be stored on nfsv4-data0 and nfsv4-data1, whereas the data files for
199 will be store on nfsv4-data2 and nfsv4-data3.
201 nfs_server_flags="-u -t -n 128 -p nfsv4-data0:/data0#/export1,nfsv4-data1:/data1#/export1,nfsv4-data2:/data2#/export2,nfsv4-data3:/data3#/export2"
204 This can be used by system administrators to control where data files are
205 stored and might be useful for control of storage use.
206 For this case, it may be convenient to co-locate more than one of the DSs
209 server, using separate file systems on the DS system
210 for storage of the respective DS's data files.
211 If mirroring is desired for this case, the
213 option also needs to be specified.
214 There must be enough DSs assigned to each exported file system on the MDS
215 to support the level of mirroring.
216 The above example would be fine for two way mirroring, but four way mirroring
217 would not work, since there are only two DSs assigned to each exported file
220 The number of subdirectories in each DS is defined by the
221 .Dq vfs.nfs.dsdirsize
223 This value can be increased from the default of 20, but only when the
225 is not running and after the additional ds20,... subdirectories have been
226 created on all the DSs.
227 For a service that will store a large number of files this sysctl should be
228 set much larger, to avoid the number of entries in a subdirectory from
231 Once operational, NFSv4.1 or NFSv4.2
236 option should do I/O directly on the DSs.
237 The clients mounting the MDS must be running the
239 daemon for pNFS to work.
241 .Bd -literal -offset indent
248 Non-pNFS aware clients or NFSv3 mounts will do all I/O RPCs on the MDS,
249 which acts as a proxy for the appropriate DS(s).
250 .Sh Backing up a pNFS service
251 Since the data is separated from the metadata, the simple way to back up
252 a pNFS service is to do so from an NFS client that has the service mounted
254 If you back up the MDS exported file system(s) on the MDS, you must do it
255 in such a way that the
257 namespace extended attributes get backed up.
258 .Sh Handling of failed mirrored DSs
259 When a mirrored DS fails, it can be disabled one of three ways:
261 1 - The MDS detects a problem when trying to do proxy
262 operations on the DS.
263 This can take a couple of minutes
264 after the DS failure or network partitioning occurs.
266 2 - A pNFS client can report an I/O error that occurred for a DS to the MDS in
267 the arguments for a LayoutReturn operation.
269 3 - The system administrator can perform the pnfsdskill(8) command on the MDS
271 If the system administrator does a pnfsdskill(8) and it fails with ENXIO
272 (Device not configured) that normally means the DS was already
273 disabled via #1 or #2.
274 Since doing this is harmless, once a system administrator knows that
275 there is a problem with a mirrored DS, doing the command is recommended.
277 Once a system administrator knows that a mirrored DS has malfunctioned
278 or has been network partitioned, they should do the following as root/su
280 .Bd -literal -offset indent
281 # pnfsdskill <mounted-on-path-of-DS>
282 # umount -N <mounted-on-path-of-DS>
285 Note that the <mounted-on-path-of-DS> must be the exact mounted-on path
286 string used when the DS was mounted on the MDS.
288 Once the mirrored DS has been disabled, the pNFS service should continue to
289 function, but file updates will only happen on the DS(s) that have not been disabled.
290 Assuming two way mirroring, that implies the one DS of the pair stored in the
292 extended attribute for the file on the MDS, for files stored on the disabled DS.
294 The next step is to clear the IP address in the
296 extended attribute on all files on the MDS for the failed DS.
297 This is done so that, when the disabled DS is repaired and brought back online,
298 the data files on this DS will not be used, since they may be out of date.
299 The command that clears the IP address is
306 # pnfsdsfile -r nfsv4-data3 yyy.c
307 yyy.c: nfsv4-data2.home.rick ds0/207508569ff983350c000000ec7c0200e4c57b2e0000000000000000 0.0.0.0 ds0/207508569ff983350c000000ec7c0200e4c57b2e0000000000000000
310 replaces nfsv4-data3 with an IPv4 address of 0.0.0.0, so that nfsv4-data3
313 Normally this will be called within a
315 command for all regular
316 files in the exported directory tree and must be done on the MDS.
319 you will probably also want the
321 option so that it won't spit out the results for every file.
322 If the disabled/repaired DS is nfsv4-data3, the commands done on the MDS
325 # cd <top-level-exported-dir>
326 # find . -type f -exec pnfsdsfile -q -r nfsv4-data3 {} \;
329 There is a problem with the above command if the file found by
331 is renamed or unlinked before the
333 command is done on it.
334 This should normally generate an error message.
335 A simple unlink is harmless
336 but a link/unlink or rename might result in the file not having been processed
338 To check that all files have their IP addresses set to 0.0.0.0 these
339 commands can be used (assuming the
343 # cd <top-level-exported-dir>
344 # find . -type f -exec pnfsdsfile {} \; | sed "/nfsv4-data3/!d"
347 Any line(s) printed require the
352 Once this is done, the replaced/repaired DS can be brought back online.
353 It should have empty ds0,...,dsN directories under the top level exported
354 directory for storage of data files just like it did when first set up.
355 Mount it on the MDS exactly as you did before disabling it.
356 For the nfsv4-data3 example, the command would be:
358 # mount -t nfs -o nfsv4,minorversion=2,soft,retrans=2 nfsv4-data3:/ /data3
361 Then restart the nfsd to re-enable the DS.
363 # /etc/rc.d/nfsd restart
366 Now, new files can be stored on nfsv4-data3,
367 but files with the IP address zeroed out on the MDS will not yet use the
368 repaired DS (nfsv4-data3).
369 The next step is to go through the exported file tree on the MDS and,
371 files with an IPv4 address of 0.0.0.0 in its extended attribute, copy the file
372 data to the repaired DS and re-enable use of this mirror for it.
373 This command for copying the file data for one MDS file is
375 and it will also normally be used in a
377 For the example case, the commands on the MDS would be:
379 # cd <top-level-exported-dir>
380 # find . -type f -exec pnfsdscopymr -r /data3 {} \;
383 When this completes, the recovery should be complete or at least nearly so.
384 As noted above, if a link/unlink or rename occurs on a file name while the
387 is in progress, it may not get copied.
388 To check for any file(s) not yet copied, the commands are:
390 # cd <top-level-exported-dir>
391 # find . -type f -exec pnfsdsfile {} \; | sed "/0\.0\.0\.0/!d"
394 If this command prints out any file name(s), these files must
397 command done on them to complete the recovery.
399 # pnfsdscopymr -r /data3 <file-path-reported>
402 If this command fails with the error
404 .Dq pnfsdscopymr: Copymr failed for file <path>: Device not configured
406 repeatedly, this may be caused by a Read/Write layout that has not
408 The only way to get rid of such a layout is to restart the
411 All of these commands are designed to be
412 done while the pNFS service is running and can be re-run safely.
414 For a more detailed discussion of the setup and management of a pNFS service
416 .Bd -literal -offset indent
417 https://people.freebsd.org/~rmacklem/pnfs-planb-setup.txt
436 service first appeared in
439 Since the MDS cannot be mirrored, it is a single point of failure just
443 For non-mirrored configurations, all
445 systems used in the service
446 are single points of failure.