1 <?xml version="1.0" encoding="utf-8"?>
2 <chapter xmlns="http://docbook.org/ns/docbook"
3 xmlns:xl="http://www.w3.org/1999/xlink" version="5.0" xml:lang="en-US"
5 xmlns:xi="http://www.w3.org/2001/XInclude">
6 <title xml:id="userutilities.title">User Utilities</title>
7 <para>This chapter describes user utilities.</para>
8 <section xml:id="dbdoclet.50438206_94597">
11 <primary>lfs</primary>
13 <literal>lfs</literal>
16 <literal>lfs</literal> utility can be used for user configuration routines
17 and monitoring.</para>
19 <title>Synopsis</title>
22 lfs changelog [--follow] <replaceable>mdt_name</replaceable> [startrec [endrec]]
23 lfs changelog_clear <replaceable>mdt_name id endrec</replaceable>
24 lfs check <replaceable>mds|osts|servers</replaceable>
25 lfs data_version [-nrw] <replaceable>filename</replaceable>
26 lfs df [-i] [-h] [--pool]-p <replaceable>fsname</replaceable>[.<replaceable>pool</replaceable>] [<replaceable>path</replaceable>] [--lazy]
27 lfs find [[!] --atime|-A [-+]N] [[!] --mtime|-M [-+]N]
28 [[!] --ctime|-C [-+]N] [--maxdepth|-D N] [--name|-n <replaceable>pattern</replaceable>]
29 [--print|-p] [--print0|-P] [[!] --obd|-O <replaceable>ost_name</replaceable>[,<replaceable>ost_name...</replaceable>]]
30 [[!] --size|-S [+-]N[kMGTPE]] --type |-t {bcdflpsD}]
31 [[!] --gid|-g|--group|-G <replaceable>gname|gid</replaceable>]
32 [[!] --uid|-u|--user|-U <replaceable>uname|uid</replaceable>]
33 <replaceable>dirname|filename</replaceable>
34 lfs getname [-h]|[path...]
35 lfs getstripe [--obd|-O <replaceable>ost_name</replaceable>] [--quiet|-q] [--verbose|-v]
36 [--stripe-count|-c] [--stripe-index|-i]
37 [--stripe-size|-s] [--pool|-p] [--directory|-d]
38 [--mdt-index|-M] [--recursive|-r] [--raw|-R]
40 <replaceable>dirname|filename</replaceable> ...
41 lfs setstripe [--size|-s stripe_size] [--stripe-count|-c <replaceable>stripe_count</replaceable>]
42 [--overstripe-count|-C <replaceable>stripe_count</replaceable>]
43 [--stripe-index|-i <replaceable>start_ost_index</replaceable>]
44 [--ost-list|-o <replaceable>ost_indicies</replaceable>]
45 [--pool|-p <replaceable>pool</replaceable>]
46 <replaceable>dirname|filename</replaceable>
47 lfs setstripe -d <replaceable>dir</replaceable>
49 lfs pool_list <replaceable>filesystem</replaceable>[.<replaceable>pool</replaceable>]| <replaceable>pathname</replaceable>
50 lfs quota [-q] [-v] [-h] [-o <replaceable>obd_uuid</replaceable>|-I <replaceable>ost_idx</replaceable>|-i <replaceable>mdt_idx</replaceable>]
51 [-u <replaceable>username|uid|-g</replaceable> <replaceable>group|gid</replaceable>|-p <replaceable>projid</replaceable>] <replaceable>/mount_point</replaceable>
52 lfs quota -t -u|-g|-p <replaceable>/mount_point</replaceable>
53 lfs setquota {-u|--user|-g|--group|-p|--project} <replaceable>uname|uid|gname|gid|projid</replaceable>
54 [--block-softlimit <replaceable>block_softlimit</replaceable>]
55 [--block-hardlimit <replaceable>block_hardlimit</replaceable>]
56 [--inode-softlimit <replaceable>inode_softlimit</replaceable>]
57 [--inode-hardlimit <replaceable>inode_hardlimit</replaceable>]
58 <replaceable>/mount_point</replaceable>
59 lfs setquota -u|--user|-g|--group|-p|--project <replaceable>uname|uid|gname|gid|projid</replaceable>
60 [-b <replaceable>block_softlimit</replaceable>] [-B <replaceable>block_hardlimit</replaceable>]
61 [-i <replaceable>inode-softlimit</replaceable>] [-I <replaceable>inode_hardlimit</replaceable>]
62 <replaceable>/mount_point</replaceable>
63 lfs setquota -t -u|-g|-p [--block-grace <replaceable>block_grace</replaceable>]
64 [--inode-grace <replaceable>inode_grace</replaceable>]
65 <replaceable>/mount_point</replaceable>
66 lfs setquota -t -u|-g|-p [-b <replaceable>block_grace</replaceable>] [-i <replaceable>inode_grace</replaceable>]
67 <replaceable>/mount_point</replaceable>
71 <para>In the above example, the
73 <replaceable>/mount_point</replaceable>
74 </literal> parameter refers to the mount point of the Lustre file
78 <para>The old lfs quota output was very detailed and contained
79 cluster-wide quota statistics (including cluster-wide limits for a
80 user/group and cluster-wide usage for a user/group), as well as
81 statistics for each MDS/OST. Now,
82 <literal>lfs quota</literal> has been updated to provide only
83 cluster-wide statistics, by default. To obtain the full report of
84 cluster-wide limits, usage and statistics, use the
85 <literal>-v</literal> option with
86 <literal>lfs quota</literal>.</para>
88 <para condition="l24">
89 The <literal>quotacheck</literal>, <literal>quotaon</literal> and <literal>quotaoff</literal>
90 sub-commands were deprecated in the Lustre 2.4 release, and removed completely in
91 the Lustre 2.8 release. See <xref linkend="enabling_disk_quotas"/> for details on
92 configuring and checking quotas.
96 <title>Description</title>
98 <literal>lfs</literal> utility is used to create a new file with a
99 specific striping pattern, determine the default striping pattern, gather
100 the extended attributes (object numbers and location) for a specific
101 file, find files with specific attributes, list OST information or set
102 quota limits. It can be invoked interactively without any arguments or in
103 a non-interactive mode with one of the supported arguments.</para>
106 <title>Options</title>
108 <literal>lfs</literal> options are listed and described below. For a
109 complete list of available options, type help at the
110 <literal>lfs</literal> prompt.</para>
111 <informaltable frame="all">
113 <colspec colname="c1" colwidth="20*" />
114 <colspec colname="c2" colwidth="30*" />
115 <colspec colname="c3" colwidth="50*" />
118 <entry nameend="c2" namest="c1">
120 <emphasis role="bold">Option</emphasis>
125 <emphasis role="bold">Description</emphasis>
132 <entry nameend="c2" namest="c1">
134 <literal>changelog</literal>
138 <para>Shows the metadata changes on an MDT. Start and end
139 points are optional. The
140 <literal>--follow</literal> option blocks on new changes; this
141 option is only valid when run directly on the MDT node.</para>
145 <entry nameend="c2" namest="c1">
147 <literal>changelog_clear</literal>
151 <para>Indicates that changelog records previous to
153 <replaceable>endrec</replaceable>
154 </literal> are no longer of interest to a particular consumer
156 <replaceable>id</replaceable>
157 </literal>, potentially allowing the MDT to free up disk space.
160 <replaceable>endrec</replaceable>
161 </literal> of 0 indicates the current last record. Changelog
162 consumers must be registered on the MDT node using
163 <literal>lctl</literal>.</para>
167 <entry nameend="c2" namest="c1">
168 <literal>check</literal>
171 <para>Displays the status of MDS or OSTs (as specified in the
172 command) or all servers (MDS and OSTs).</para>
176 <entry nameend="c2" namest="c1">
178 <literal>data_version [-nrw]
179 <replaceable>filename</replaceable></literal>
183 <para>Displays the current version of file data. If
184 <literal>-n</literal> is specified, the data version is read
185 without taking a lock. As a consequence, the data version could
186 be outdated if there are dirty caches on filesystem clients, but
187 this option will not force data flushes and has less of an
188 impact on the filesystem. If <literal>-r</literal> is specified,
189 the data version is read after dirty pages on clients are
190 flushed. If <literal>-w</literal> is specified, the data version
191 is read after all caching pages on clients are flushed.
194 Even with <literal>-r</literal> or <literal>-w</literal>, race
195 conditions are possible and the data version should be checked
196 before and after an operation to be confident the data did not
200 The data version is the sum of the last committed transaction
201 numbers of all data objects of a file. It is used by HSM policy
202 engines for verifying that file data has not been changed during
203 an archive operation or before a release operation, and by OST
204 migration, primarily for verifying that file data has not been
205 changed during a data copy, when done in non-blocking mode.
210 <entry nameend="c2" namest="c1">
211 <literal>df [-i] [-h] [--pool|-p
212 <replaceable>fsname</replaceable>[.
213 <replaceable>pool</replaceable>] [
214 <replaceable>path</replaceable>] [--lazy]</literal>
218 <literal>-i</literal> to report file system disk space usage or
219 inode usage of each MDT or OST or, if a pool is specified with
221 <literal>-p</literal> option, a subset of OSTs.</para>
222 <para>By default, the usage of all mounted Lustre file systems
224 <literal>path</literal> option is included, only the usage for
225 the specified file system is reported. If the
226 <literal>-h</literal> option is included, the output is printed
227 in human-readable format, using SI base-2 suffixes for
228 <emphasis role="bold">M</emphasis>ega-,
229 <emphasis role="bold">G</emphasis>iga-,
230 <emphasis role="bold">T</emphasis>era-,
231 <emphasis role="bold">P</emphasis>eta-, or
232 <emphasis role="bold">E</emphasis>xabytes.</para>
234 <literal>--lazy</literal> option is specified, any OST that is
235 currently disconnected from the client will be skipped. Using
237 <literal>--lazy</literal> option prevents the
238 <literal>df</literal> output from being blocked when an OST is
239 offline. Only the space on the OSTs that can currently be
240 accessed are returned. The
241 <literal>llite.*.lazystatfs</literal> tunable can be enabled to
242 make this the default behaviour for all
243 <literal>statfs()</literal> operations.</para>
247 <entry nameend="c2" namest="c1">
249 <literal>find</literal>
253 <para>Searches the directory tree rooted at the given
254 directory/filename for files that match the given
257 <literal>!</literal> before an option negates its meaning (files
258 NOT matching the parameter). Using
259 <literal>+</literal> before a numeric value means files with the
260 parameter OR MORE. Using
261 <literal>-</literal> before a numeric value means files with the
262 parameter OR LESS.</para>
268 <literal>--atime</literal>
271 <para>File was last accessed N*24 hours ago. (There is no
273 <literal>atime</literal> is kept coherent across the
275 <para>OSTs store a transient
276 <literal>atime</literal> that is updated when clients do read
278 <literal>atime</literal> is written to the MDS when the file is
279 closed. However, on-disk atime is only updated if it is more
280 than 60 seconds old (
281 <literal>/proc/fs/lustre/mds/*/max_atime_diff</literal>). The
282 Lustre software considers the latest
283 <literal>atime</literal> from all OSTs. If a
284 <literal>setattr</literal> is set by user, then it is updated on
285 both the MDS and OST, allowing the
286 <literal>atime</literal> to go backward.</para>
295 <literal>--ctime</literal>
299 <para>File status was last changed N*24 hours ago.</para>
308 <literal>--mtime</literal>
312 <para>File data was last modified N*24 hours ago.</para>
321 <literal>--obd</literal>
325 <para>File has an object on a specific OST(s).</para>
334 <literal>--size</literal>
338 <para>File has a size in bytes, or kilo-, Mega-, Giga-, Tera-,
339 Peta- or Exabytes if a suffix is given.</para>
348 <literal>--type</literal>
352 <para>File has the type - block, character, directory, pipe,
353 file, symlink, socket or door (used in Solaris operating
363 <literal>--uid</literal>
367 <para>File has a specific numeric user ID.</para>
376 <literal>--user</literal>
380 <para>File owned by a specific user (numeric user ID
390 <literal>--gid</literal>
394 <para>File has a specific group ID.</para>
403 <literal>--group</literal>
407 <para>File belongs to a specific group (numeric group ID
417 <literal>-maxdepth</literal></para>
420 <para>Limits find to descend at most N levels of the directory
430 <literal>--print</literal>/
431 <literal>--print0</literal></para>
434 <para>Prints the full filename, followed by a new line or NULL
435 character correspondingly.</para>
439 <entry nameend="c2" namest="c1">
441 <literal>osts [path]</literal>
445 <para>Lists all OSTs for the file system. If a path located on
446 a mounted Lustre file system is specified, then only OSTs
447 belonging to this file system are displayed.</para>
451 <entry nameend="c2" namest="c1">
453 <literal>getname [path...]</literal>
457 <para>List each Lustre file system instance associated with
458 each Lustre mount point. If no path is specified, all Lustre
459 mount points are interrogated. If a list of paths is provided,
460 the instance of each path is provided. If the path is not a
461 Lustre instance 'No such device' is returned.</para>
465 <entry nameend="c2" namest="c1">
467 <literal>getstripe</literal>
471 <para>Lists striping information for a given filename or
472 directory. By default, the stripe count, stripe size and offset
474 <para>If you only want specific striping information, then the
476 <literal>--stripe-count</literal>,
477 <literal>--stripe-size</literal>,
478 <literal>--stripe-index</literal>,
479 <literal>--layout</literal>, or
480 <literal>--pool</literal> plus various combinations of these
481 options can be used to retrieve specific information.</para>
483 <literal>--raw</literal> option is specified, the stripe
484 information is printed without substituting the file system
485 default values for unspecified fields. If the striping EA is
486 not set, 0, 0, and -1 will be printed for the stripe count,
487 size, and offset respectively.</para>
488 <para>The <literal>--mdt-index</literal> prints the index of
489 the MDT for a given directory. See
490 <xref linkend="lustremaint.rmremotedir" />.</para>
500 <replaceable>ost_name</replaceable></literal>
504 <para>Lists files that have an object on a specific OST.</para>
513 <literal>--quiet</literal>
517 <para>Lists details about the file's object ID
527 <literal>--verbose</literal>
531 <para>Prints additional striping information.</para>
540 <literal>--stripe-count</literal>
544 <para>Lists the stripe count (how many OSTs to use).</para>
553 <literal>--index</literal>
557 <para>Lists the index for each OST in the file system.</para>
566 <literal>--offset</literal>
570 <para>Lists the OST index on which file striping starts.</para>
579 <literal>--pool</literal>
583 <para>Lists the pools to which a file belongs.</para>
592 <literal>--size</literal>
596 <para>Lists the stripe size (how much data to write to one OST
597 before moving to the next OST).</para>
606 <literal>--directory</literal>
610 <para>Lists entries about a specified directory instead of its
611 contents (in the same manner as
612 <literal>ls -d</literal>).</para>
621 <literal>--recursive</literal>
625 <para>Recurses into all sub-directories.</para>
629 <entry nameend="c2" namest="c1">
631 <literal>setstripe</literal>
635 <para>Create new files with a specific file layout (stripe
636 pattern) configuration.
638 <para>The file cannot exist prior to using
639 <literal>setstripe</literal>. A directory must exist prior to
641 <literal>setstripe</literal>.</para>
651 <literal>--stripe-count stripe_cnt</literal>
655 <para>Number of OSTs over which to stripe a file. A
656 <literal>stripe_cnt</literal> of 0 uses the file system-wide
657 default stripe count (default is 1). A
658 <literal>stripe_cnt</literal> of -1 stripes over all available
668 <literal>--overstripe-count stripe_cnt</literal>
672 <para>The same as --stripe-count, but allows overstriping,
673 which will place more than one stripe per OST if
674 <literal>stripe_cnt</literal> is greater than the number of
675 OSTs. Overstriping is useful for matching the number of stripes
676 to the number of processes, or with very fast OSTs, where one
677 stripe per OST is not enough to get full performance.</para>
686 <literal>--size stripe_size</literal>
688 <para>The default stripe-size is 0. The default start-ost is
689 -1. Do NOT confuse them! If you set start-ost to 0, all new
690 file creations occur on OST 0 (seldom a good idea).</para>
691 </footnote> </para>
694 <para>Number of bytes to store on an OST before moving to the
695 next OST. A stripe_size of 0 uses the file system's default
696 stripe size, (default is 1 MB). Can be specified with
697 <emphasis role="bold">k</emphasis>(KB),
698 <emphasis role="bold">m</emphasis>(MB), or
699 <emphasis role="bold">g</emphasis>(GB), respectively.</para>
708 <literal>--stripe-index start_ost_index</literal>
712 <para>The OST index (base 10, starting at 0) on which to start
713 striping for this file. A start_ost_index value of -1 allows
714 the MDS to choose the starting index. This is the default
715 value, and it means that the MDS selects the starting OST as it
716 wants. We strongly recommend selecting this default, as it
717 allows space and load balancing to be done by the MDS as
719 <literal>start_ost_index</literal> value has no relevance on
720 whether the MDS will use round-robin or QoS weighted allocation
721 for the remaining stripes in the file.</para>
730 <literal>--ost-index ost_indices</literal>
734 <para>This option is used to specify the exact stripe
735 layout on the the file system. <literal>ost_indices</literal>
736 is a list of OSTs referenced by their indices and index ranges
737 separated by commas, e.g. <literal>1,2-4,7</literal>.</para>
747 <replaceable>pool</replaceable></literal>
751 <para>Name of the pre-defined pool of OSTs (see
752 <xref linkend="dbdoclet.50438219_38274" />) that will be used
754 <literal>stripe_cnt</literal>,
755 <literal>stripe_size</literal> and
756 <literal>start_ost</literal> values are used as well. The
757 start-ost value must be part of the pool or an error is
762 <entry nameend="c2" namest="c1">
764 <literal>setstripe -d</literal>
768 <para>Deletes default striping on the specified
773 <entry nameend="c2" namest="c1">
775 <literal>pool_list {filesystem}[.poolname]|{pathname}</literal>
779 <para>Lists pools in the file system or pathname, or OSTs in
780 the file system's pool.</para>
784 <entry nameend="c2" namest="c1">
786 <literal>quota [-q] [-v] [-o
787 <replaceable>obd_uuid</replaceable>|-i
788 <replaceable>mdt_idx</replaceable>|-I
789 <replaceable>ost_idx</replaceable>] [-u|-g|-p
790 <replaceable>uname|uid|gname|gid|projid]</replaceable>
791 <replaceable>/mount_point</replaceable></literal>
796 <para>Displays disk usage and limits, either for the full file
797 system or for objects on a specific OBD. A user or group name
798 or an usr, group and project ID can be specified. If all user,
799 group project ID are omitted, quotas for the current UID/GID
800 are shown. The <literal>-q</literal> option disables printing
801 of additional descriptions (including column titles). It fills
802 in blank spaces in the
803 <literal>grace</literal> column with zeros (when there is no
804 grace period set), to ensure that the number of columns is
806 <literal>-v</literal> option provides more verbose (per-OBD
807 statistics) output.</para>
811 <entry nameend="c2" namest="c1">
814 <replaceable>-u|-g|-p</replaceable>
815 <replaceable>/mount_point</replaceable></literal>
819 <para>Displays block and inode grace times for user (
820 <literal>-u</literal>) or group (
821 <literal>-g</literal>) or project (
822 <literal>-p</literal>) quotas.</para>
826 <entry nameend="c2" namest="c1">
828 <literal>setquota {-u|-g|-p
829 <replaceable>uname|uid|gname|gid|projid}</replaceable>
831 <replaceable>block_softlimit</replaceable>]
833 <replaceable>block_hardlimit</replaceable>]
835 <replaceable>inode_softlimit</replaceable>]
837 <replaceable>inode_hardlimit</replaceable>]
838 <replaceable>/mount_point</replaceable></literal>
842 <para>Sets file system quotas for users, groups or one project.
843 Limits can be specified with
844 <literal>--{block|inode}-{softlimit|hardlimit}</literal> or
845 their short equivalents
846 <literal>-b</literal>,
847 <literal>-B</literal>,
848 <literal>-i</literal>,
849 <literal>-I</literal>. Users can set 1, 2, 3 or 4 limits.
852 <literal>setquota</literal> interface is supported, but it may
853 be removed in a future Lustre software release.</para>
854 </footnote>Also, limits can be specified with special suffixes,
855 -b, -k, -m, -g, -t, and -p to indicate units of 1, 2^10, 2^20,
856 2^30, 2^40 and 2^50, respectively. By default, the block limits
857 unit is 1 kilobyte (1,024), and block limits are always
858 kilobyte-grained (even if specified in bytes). See
859 <xref linkend="dbdoclet.50438206_11903" />.</para>
863 <entry nameend="c2" namest="c1">
865 <literal>setquota -t -u|-g|-p [--block-grace
866 <replaceable>block_grace</replaceable>] [--inode-grace
867 <replaceable>inode_grace</replaceable>]
868 <replaceable>/mount_point</replaceable></literal>
872 <para>Sets the file system quota grace times for users or
873 groups. Grace time is specified in '
874 <literal>XXwXXdXXhXXmXXs</literal>' format or as an integer
876 <xref linkend="dbdoclet.50438206_11903" />.</para>
880 <entry nameend="c2" namest="c1">
882 <literal>help</literal>
886 <para>Provides brief help on various
887 <literal>lfs</literal> arguments.</para>
891 <entry nameend="c2" namest="c1">
893 <literal>exit/quit</literal>
897 <para>Quits the interactive
898 <literal>lfs</literal> session.</para>
905 <section remap="h5" xml:id="dbdoclet.50438206_11903">
906 <title>Examples</title>
907 <para>Creates a file striped on two OSTs with 128 KB on each
910 $ lfs setstripe -s 128k -c 2 /mnt/lustre/file1
912 <para>Deletes a default stripe pattern on a given directory. New files
913 use the default striping pattern.</para>
915 $ lfs setstripe -d /mnt/lustre/dir
917 <para>Lists the detailed object allocation of a given file.</para>
919 $ lfs getstripe -v /mnt/lustre/file1
921 <para>List all the mounted Lustre file systems and corresponding Lustre
926 <para>Efficiently lists all files in a given directory and its
927 subdirectories.</para>
929 $ lfs find /mnt/lustre
931 <para>Recursively lists all regular files in a given directory more than
934 $ lfs find /mnt/lustre -mtime +30 -type f -print
936 <para>Recursively lists all files in a given directory that have objects
937 on OST2-UUID. The lfs check servers command checks the status of all
938 servers (MDT and OSTs).</para>
940 $ lfs find --obd OST2-UUID /mnt/lustre/
942 <para>Lists all OSTs in the file system.</para>
946 <para>Lists space usage per OST and MDT in human-readable format.</para>
950 <para>Lists inode usage per OST and MDT.</para>
954 <para>List space or inode usage for a specific OST pool.</para>
957 <replaceable>filesystem</replaceable>[.
958 <replaceable>pool</replaceable>] |
959 <replaceable>pathname</replaceable>
961 <para>List quotas of user 'bob'.</para>
963 $ lfs quota -u bob /mnt/lustre
965 <para>List quotas of project ID '1'.</para>
967 $ lfs quota -p 1 /mnt/lustre
969 <para>Show grace times for user quotas on
970 <literal>/mnt/lustre</literal>.</para>
972 $ lfs quota -t -u /mnt/lustre
974 <para>Sets quotas of user 'bob', with a 1 GB block quota hardlimit and a
975 2 GB block quota softlimit.</para>
977 $ lfs setquota -u bob --block-softlimit 2000000 --block-hardlimit 1000000
980 <para>Sets grace times for user quotas: 1000 seconds for block quotas, 1
981 week and 4 days for inode quotas.</para>
983 $ lfs setquota -t -u --block-grace 1000 --inode-grace 1w4d /mnt/lustre
985 <para>Checks the status of all servers (MDT, OST)</para>
989 <para>Creates a file striped on two OSTs from the pool
990 <literal>my_pool</literal></para>
992 $ lfs setstripe --pool my_pool -c 2 /mnt/lustre/file
994 <para>Lists the pools defined for the mounted Lustre file system
995 <literal>/mnt/lustre</literal></para>
997 $ lfs pool_list /mnt/lustre/
999 <para>Lists the OSTs which are members of the pool
1000 <literal>my_pool</literal> in file system
1001 <literal>my_fs</literal></para>
1003 $ lfs pool_list my_fs.my_pool
1005 <para>Finds all directories/files associated with
1006 <literal>poolA</literal>.</para>
1008 $ lfs find /mnt/lustre --pool poolA
1010 <para>Finds all directories/files not associated with a pool.</para>
1012 $ lfs find /mnt//lustre --pool ""
1014 <para>Finds all directories/files associated with pool.</para>
1016 $ lfs find /mnt/lustre ! --pool ""
1018 <para>Associates a directory with the pool
1019 <literal>my_pool</literal>, so all new files and directories are created
1022 $ lfs setstripe --pool my_pool /mnt/lustre/dir
1025 <section remap="h5">
1026 <title>See Also</title>
1028 <xref linkend="dbdoclet.50438219_38274" />
1032 <section xml:id="dbdoclet.lfs_migrate">
1035 <primary>lfs_migrate</primary>
1037 <literal>lfs_migrate</literal>
1040 <literal>lfs_migrate</literal> utility is a simple to migrate file
1041 <emphasis>data</emphasis> between OSTs.</para>
1042 <section remap="h5">
1043 <title>Synopsis</title>
1045 lfs_migrate [<replaceable>lfs_setstripe_options</replaceable>]
1046 [-h] [-n] [-q] [-R] [-s] [-y] [-0] [file|directory ...]
1049 <section remap="h5">
1050 <title>Description</title>
1052 <literal>lfs_migrate</literal> utility is a tool to assist migration
1053 of file data between Lustre OSTs. The utility copies each specified
1054 file to a temporary file using supplied <literal>lfs setstripe</literal>
1055 options, if any, optionally verifies the file contents have not changed,
1056 and then swaps the layout (OST objects) from the temporary file and the
1057 original file (for Lustre 2.5 and later), or renames the temporary file
1058 to the original filename. This allows the user/administrator to balance
1059 space usage between OSTs, or move files off OSTs that are starting to show
1060 hardware problems (though are still functional) or will be removed.</para>
1062 <para>For versions of Lustre before 2.5,
1063 <literal>lfs_migrate</literal> was not integrated with the MDS at all.
1064 That made it UNSAFE for use on files that were being modified by other
1065 applications, since the file was migrated through a copy and rename of
1066 the file. With Lustre 2.5 and later, the new file layout is swapped
1067 with the existing file layout, which ensures that the user-visible
1068 inode number is kept, and open file handles and locks on the file are
1071 <para>Files to be migrated can be specified as command-line arguments. If
1072 a directory is specified on the command-line then all files within the
1073 directory are migrated. If no files are specified on the command-line,
1074 then a list of files is read from the standard input, making
1075 <literal>lfs_migrate</literal> suitable for use with
1076 <literal>lfs find</literal> to locate files on specific OSTs and/or
1077 matching other file attributes, and other tools that generate a list
1078 of files on standard output.</para>
1079 <para>Unless otherwise specified through command-line options, the
1080 file allocation policies on the MDS dictate where the new files
1081 are placed, taking into account whether specific OSTs have been
1082 disabled on the MDS via <literal>lctl</literal> (preventing new
1083 files from being allocated there), whether some OSTs are overly full
1084 (reducing the number of files placed on those OSTs), or if there is
1085 a specific default file striping for the parent directory (potentially
1086 changing the stripe count, stripe size, OST pool, or OST index of a
1090 <literal>lfs_migrate</literal> utility can also be used in some cases to
1093 <primary>fragmentation</primary>
1094 </indexterm>fragmentation. File fragmentation will typically reduce
1095 Lustre file system performance. File fragmentation may be observed on
1096 an aged file system and will commonly occur if the file was written by
1097 many threads. Provided there is sufficient free space (or if it was
1098 written when the file system was nearly full) that is less fragmented
1099 than the file being copied, re-writing a file with
1100 <literal>lfs_migrate</literal> will result in a migrated file with
1101 reduced fragmentation. The tool
1102 <literal>filefrag</literal> can be used to report file fragmentation.
1104 <xref linkend="dbdoclet.50438206_75125" /></para>
1107 <para>As long as a file has extent lengths of tens of megabytes (
1108 <replaceable>read_bandwidth * seek_time</replaceable>) or more, the
1109 read performance for the file will not be significantly impacted by
1110 fragmentation, since the read pipeline can be filled by large reads
1111 from disk even with an occasional disk seek.</para>
1114 <section remap="h5">
1115 <title>Options</title>
1116 <para>Options supporting
1117 <literal>lfs_migrate</literal> are described below.</para>
1118 <informaltable frame="all">
1120 <colspec colname="c1" colwidth="50*" />
1121 <colspec colname="c2" colwidth="50*" />
1126 <emphasis role="bold">Option</emphasis>
1131 <emphasis role="bold">Description</emphasis>
1141 <replaceable>stripecount</replaceable></literal>
1145 <para>Restripe file using the specified stripe count. This
1146 option may not be specified at the same time as the
1147 <literal>-R</literal> option.</para>
1153 <literal>-h</literal>
1157 <para>Display help information.</para>
1162 <literal>-l</literal>
1165 <para>Migrate files with hard links (skips, by default). Files
1166 with multiple hard links are split into multiple separate files
1168 <literal>lfs_migrate</literal>, so they are skipped, by
1169 default, to avoid breaking the hard links.</para>
1174 <literal>-n</literal>
1177 <para>Only print the names of files to be migrated.</para>
1183 <literal>-q</literal>
1187 <para>Run quietly (does not print filenames or status).</para>
1192 <literal>-R</literal>
1194 <entry>Restripe file using default directory striping instead of
1195 keeping striping. This option may not be specified at the same
1197 <literal>-c</literal> option.</entry>
1201 <literal>-s</literal>
1203 <entry>Skip file data comparison after migrate. Default is to
1204 compare migrated file against original to verify
1205 correctness.</entry>
1210 <literal>-y</literal>
1215 <literal>y</literal>' to usage warning without prompting (for
1216 scripts, use with caution).</para>
1221 <literal>-0</literal>
1223 <entry>Expect NUL-terminated filenames on standard input, as
1224 generated by <literal>lfs find -print0</literal> or
1225 <literal>find -print0</literal>. This allows filenames with
1226 embedded newlines to be handled correctly.
1233 <section remap="h5">
1234 <title>Examples</title>
1235 <para>Rebalance all files in
1236 <literal>/mnt/lustre/dir</literal>:</para>
1238 $ lfs_migrate /mnt/lustre/dir
1240 <para>Migrate files in /test filesystem on OST0004 larger than 4 GB in
1241 size and older than a day old:</para>
1243 $ lfs find /test -obd test-OST0004 -size +4G -mtime +1 | lfs_migrate -y
1246 <section remap="h5">
1247 <title>See Also</title>
1249 <xref linkend="dbdoclet.50438206_94597" />
1253 <section xml:id="dbdoclet.50438206_75125">
1256 <primary>filefrag</primary>
1258 <literal>filefrag</literal>
1261 <literal>e2fsprogs</literal> package contains the
1262 <literal>filefrag</literal> tool which reports the extent of file
1263 fragmentation.</para>
1264 <section remap="h5">
1265 <title>Synopsis</title>
1267 filefrag [ -belsv ] [ files... ]
1270 <section remap="h5">
1271 <title>Description</title>
1273 <literal>filefrag</literal> utility reports the extent of fragmentation in
1275 <literal>filefrag</literal> utility obtains the extent information from
1276 Lustre files using the
1277 <literal>FIEMAP ioctl</literal>, which is efficient and fast, even for
1278 very large files.</para>
1279 <para>In default mode
1281 <para>The default mode is faster than the verbose/extent mode since it
1282 only counts the number of extents.</para>
1284 <literal>filefrag</literal> prints the number of physically discontiguous
1285 extents in the file. In extent or verbose mode, each extent is printed
1286 with details such as the blocks allocated on each OST. For a Lustre file
1287 system, the extents are printed in device offset order (i.e. all of the
1288 extents for one OST first, then the next OST, etc.), not file logical
1289 offset order. If the file logical offset order was used, the Lustre
1290 striping would make the output very verbose and difficult to see if there
1291 was file fragmentation or not.</para>
1293 <para>Note that as long as a file has extent lengths of tens of
1294 megabytes or more (i.e.
1295 <replaceable>read_bandwidth * seek_time >
1296 extent_length</replaceable>), the read performance for the file will
1297 not be significantly impacted by fragmentation, since file readahead
1298 can fully utilize the disk disk bandwidth even with occasional
1301 <para>In default mode
1303 <para>The default mode is faster than the verbose/extent mode.</para>
1305 <literal>filefrag</literal> returns the number of physically discontiguous
1306 extents in the file. In extent or verbose mode, each extent is printed
1307 with details. For a Lustre file system, the extents are printed in device
1308 offset order, not logical offset order.</para>
1310 <section remap="h5">
1311 <title>Options</title>
1312 <para>The options and descriptions for the
1313 <literal>filefrag</literal> utility are listed below.</para>
1314 <informaltable frame="all">
1316 <colspec colname="c1" colwidth="50*" />
1317 <colspec colname="c2" colwidth="50*" />
1322 <emphasis role="bold">Option</emphasis>
1327 <emphasis role="bold">Description</emphasis>
1336 <literal>-b</literal>
1340 <para>Uses the 1024-byte blocksize for the output. By default,
1341 this blocksize is used by the Lustre file system, since OSTs
1342 may use different block sizes.</para>
1348 <literal>-e</literal>
1352 <para>Uses the extent mode when printing the output. This is
1353 the default for Lustre files in verbose mode.</para>
1359 <literal>-l</literal>
1363 <para>Displays extents in LUN offset order. This is the only
1364 available mode for Lustre.</para>
1370 <literal>-s</literal>
1374 <para>Synchronizes any unwritten file data to disk before
1375 requesting the mapping.</para>
1381 <literal>-v</literal>
1385 <para>Prints the file's layout in verbose mode when checking
1386 file fragmentation, including the logical to physical mapping
1387 for each extent in the file and the OST index.</para>
1394 <section remap="h5">
1395 <title>Examples</title>
1396 <para>Lists default output.</para>
1398 $ filefrag /mnt/lustre/foo
1399 /mnt/lustre/foo: 13 extents found
1401 <para>Lists verbose output in extent format.</para>
1403 $ filefrag -v /mnt/lustre/foo
1404 Filesystem type is: bd00bd0
1405 File size of /mnt/lustre/foo is 1468297786 (1433888 blocks of 1024 bytes)
1406 ext: device_logical: physical_offset: length: dev: flags:
1407 0: 0.. 122879: 2804679680..2804802559: 122880: 0002: network
1408 1: 122880.. 245759: 2804817920..2804940799: 122880: 0002: network
1409 2: 245760.. 278527: 2804948992..2804981759: 32768: 0002: network
1410 3: 278528.. 360447: 2804982784..2805064703: 81920: 0002: network
1411 4: 360448.. 483327: 2805080064..2805202943: 122880: 0002: network
1412 5: 483328.. 606207: 2805211136..2805334015: 122880: 0002: network
1413 6: 606208.. 729087: 2805342208..2805465087: 122880: 0002: network
1414 7: 729088.. 851967: 2805473280..2805596159: 122880: 0002: network
1415 8: 851968.. 974847: 2805604352..2805727231: 122880: 0002: network
1416 9: 974848.. 1097727: 2805735424..2805858303: 122880: 0002: network
1417 10: 1097728.. 1220607: 2805866496..2805989375: 122880: 0002: network
1418 11: 1220608.. 1343487: 2805997568..2806120447: 122880: 0002: network
1419 12: 1343488.. 1433599: 2806128640..2806218751: 90112: 0002: network
1420 /mnt/lustre/foo: 13 extents found
1424 <section xml:id="dbdoclet.50438206_86244">
1427 <primary>mount</primary>
1429 <literal>mount</literal>
1432 <literal>mount(8)</literal> Linux command is used to mount a Lustre file
1433 system. When mounting a Lustre file system, mount(8) executes the
1434 <literal>/sbin/mount.lustre</literal> command to complete the mount. The
1435 mount command supports these options specific to a Lustre file
1437 <informaltable frame="all">
1439 <colspec colname="c1" colwidth="50*" />
1440 <colspec colname="c2" colwidth="50*" />
1445 <emphasis role="bold">Server options</emphasis>
1450 <emphasis role="bold">Description</emphasis>
1459 <literal>abort_recov</literal>
1463 <para>Aborts recovery when starting a target</para>
1469 <literal>nosvc</literal>
1473 <para>Starts only MGS/MGC servers</para>
1479 <literal>nomgs</literal>
1483 <para>Start a MDT with a co-located MGS without starting the
1490 <literal>exclude</literal>
1494 <para>Starts with a dead OST</para>
1500 <literal>md_stripe_cache_size</literal>
1504 <para>Sets the stripe cache size for server side disk with a
1505 striped raid configuration</para>
1511 <informaltable frame="all">
1513 <colspec colname="c1" colwidth="50*" />
1514 <colspec colname="c2" colwidth="50*" />
1519 <emphasis role="bold">Client options</emphasis>
1524 <emphasis role="bold">Description</emphasis>
1533 <literal>flock/noflock/localflock</literal>
1537 <para>Enables/disables global flock or local flock support</para>
1543 <literal>user_xattr/nouser_xattr</literal>
1547 <para>Enables/disables user-extended attributes</para>
1553 <literal>user_fid2path/nouser_fid2path</literal>
1557 <para>Enables/disables FID to path translation by
1558 regular users</para>
1564 <literal>retry=</literal>
1568 <para>Number of times a client will retry to mount the file
1576 <section xml:id="dbdoclet.50438206_56217">
1577 <title>Handling Timeouts</title>
1578 <para>Timeouts are the most common cause of hung applications. After a
1579 timeout involving an MDS or failover OST, applications attempting to access
1580 the disconnected resource wait until the connection gets
1582 <para>When a client performs any remote operation, it gives the server a
1583 reasonable amount of time to respond. If a server does not reply either due
1584 to a down network, hung server, or any other reason, a timeout occurs which
1585 requires a recovery.</para>
1586 <para>If a timeout occurs, a message (similar to this one), appears on the
1587 console of the client, and in
1588 <literal>/var/log/messages</literal>:</para>
1590 LustreError: 26597:(client.c:810:ptlrpc_expire_one_request()) @@@ timeout
1592 req@a2d45200 x5886/t0 o38->mds_svc_UUID@NID_mds_UUID:12 lens 168/64 ref 1 fl