1 <?xml version='1.0' encoding='UTF-8'?><chapter xmlns="http://docbook.org/ns/docbook" xmlns:xl="http://www.w3.org/1999/xlink" version="5.0" xml:lang="en-US" xml:id="managingstripingfreespace">
2 <title xml:id="managingstripingfreespace.title">Managing File Layout (Striping) and Free
4 <para>This chapter describes file layout (striping) and I/O options, and includes the following
8 <para><xref linkend="dbdoclet.50438209_79324"/></para>
11 <para><xref linkend="dbdoclet.50438209_48033"/></para>
14 <para><xref linkend="dbdoclet.50438209_78664"/></para>
17 <para><xref linkend="dbdoclet.50438209_44776"/></para>
20 <para><xref linkend="dbdoclet.50438209_10424"/></para>
23 <para><xref xmlns:xlink="http://www.w3.org/1999/xlink" linkend="section_syy_gcl_qk"/></para>
26 <section xml:id="dbdoclet.50438209_79324">
29 <primary>space</primary>
32 <primary>striping</primary>
33 <secondary>how it works</secondary>
36 <primary>striping</primary>
40 <primary>space</primary>
41 <secondary>striping</secondary>
42 </indexterm>How Lustre File System Striping Works</title>
43 <para>In a Lustre file system, the MDS allocates objects to OSTs using either a round-robin
44 algorithm or a weighted algorithm. When the amount of free space is well balanced (i.e., by
45 default, when the free space across OSTs differs by less than 17%), the round-robin algorithm
46 is used to select the next OST to which a stripe is to be written. Periodically, the MDS
47 adjusts the striping layout to eliminate some degenerated cases in which applications that
48 create very regular file layouts (striping patterns) preferentially use a particular OST in
50 <para> Normally the usage of OSTs is well balanced. However, if users create a small number of
51 exceptionally large files or incorrectly specify striping parameters, imbalanced OST usage may
52 result. When the free space across OSTs differs by more than a specific amount (17% by
53 default), the MDS then uses weighted random allocations with a preference for allocating
54 objects on OSTs with more free space. (This can reduce I/O performance until space usage is
55 rebalanced again.) For a more detailed description of how striping is allocated, see <xref
56 linkend="dbdoclet.50438209_10424"/>.</para>
57 <para condition="l22">Files can only be striped over a finite number of OSTs. Prior to Lustre
58 software release 2.2, the maximum number of OSTs that a file could be striped across was
59 limited to 160. As of Lustre software release 2.2, the maximum number of OSTs is 2000. For
60 more information, see <xref xmlns:xlink="http://www.w3.org/1999/xlink"
61 linkend="section_syy_gcl_qk"/>.</para>
63 <section xml:id="dbdoclet.50438209_48033">
65 <primary>file layout</primary>
66 <secondary>See striping</secondary>
67 </indexterm><indexterm>
68 <primary>striping</primary>
69 <secondary>considerations</secondary>
72 <primary>space</primary>
73 <secondary>considerations</secondary>
74 </indexterm> Lustre File Layout (Striping) Considerations</title>
75 <para>Whether you should set up file striping and what parameter values you select depends on
76 your needs. A good rule of thumb is to stripe over as few objects as will meet those needs and
78 <para>Some reasons for using striping include:</para>
81 <para><emphasis role="bold">Providing high-bandwidth access.</emphasis> Many applications
82 require high-bandwidth access to a single file, which may be more bandwidth than can be
83 provided by a single OSS. Examples are a scientific application that writes to a single
84 file from hundreds of nodes, or a binary executable that is loaded by many nodes when an
85 application starts.</para>
86 <para>In cases like these, a file can be striped over as many OSSs as it takes to achieve
87 the required peak aggregate bandwidth for that file. Striping across a larger number of
88 OSSs should only be used when the file size is very large and/or is accessed by many nodes
89 at a time. Currently, Lustre files can be striped across up to 2000 OSTs, the maximum
90 stripe count for an <literal>ldiskfs</literal> file system.</para>
93 <para><emphasis role="bold">Improving performance when OSS bandwidth is exceeded.</emphasis>
94 Striping across many OSSs can improve performance if the aggregate client bandwidth
95 exceeds the server bandwidth and the application reads and writes data fast enough to take
96 advantage of the additional OSS bandwidth. The largest useful stripe count is bounded by
97 the I/O rate of the clients/jobs divided by the performance per OSS.</para>
100 <para><emphasis role="bold">Providing space for very large files.</emphasis> Striping is
101 useful when a single OST does not have enough free space to hold the entire file.</para>
104 <para>Some reasons to minimize or avoid striping:</para>
107 <para><emphasis role="bold">Increased overhead.</emphasis> Striping results in more locks
108 and extra network operations during common operations such as <literal>stat</literal> and
109 <literal>unlink</literal>. Even when these operations are performed in parallel, one
110 network operation takes less time than 100 operations.</para>
111 <para>Increased overhead also results from server contention. Consider a cluster with 100
112 clients and 100 OSSs, each with one OST. If each file has exactly one object and the load
113 is distributed evenly, there is no contention and the disks on each server can manage
114 sequential I/O. If each file has 100 objects, then the clients all compete with one
115 another for the attention of the servers, and the disks on each node seek in 100 different
116 directions resulting in needless contention.</para>
119 <para><emphasis role="bold">Increased risk.</emphasis> When files are striped across all
120 servers and one of the servers breaks down, a small part of each striped file is lost. By
121 comparison, if each file has exactly one stripe, fewer files are lost, but they are lost
122 in their entirety. Many users would prefer to lose some of their files entirely than all
123 of their files partially.</para>
127 <title><indexterm><primary>striping</primary><secondary>size</secondary></indexterm>
128 Choosing a Stripe Size</title>
129 <para>Choosing a stripe size is a balancing act, but reasonable defaults are described below.
130 The stripe size has no effect on a single-stripe file.</para>
133 <para><emphasis role="bold">The stripe size must be a multiple of the page
134 size.</emphasis> Lustre software tools enforce a multiple of 64 KB (the maximum page
135 size on ia64 and PPC64 nodes) so that users on platforms with smaller pages do not
136 accidentally create files that might cause problems for ia64 clients.</para>
139 <para><emphasis role="bold">The smallest recommended stripe size is 512 KB.</emphasis>
140 Although you can create files with a stripe size of 64 KB, the smallest practical stripe
141 size is 512 KB because the Lustre file system sends 1MB chunks over the network.
142 Choosing a smaller stripe size may result in inefficient I/O to the disks and reduced
146 <para><emphasis role="bold">A good stripe size for sequential I/O using high-speed
147 networks is between 1 MB and 4 MB.</emphasis> In most situations, stripe sizes larger
148 than 4 MB may result in longer lock hold times and contention during shared file
152 <para><emphasis role="bold">The maximum stripe size is 4 GB.</emphasis> Using a large
153 stripe size can improve performance when accessing very large files. It allows each
154 client to have exclusive access to its own part of a file. However, a large stripe size
155 can be counterproductive in cases where it does not match your I/O pattern.</para>
158 <para><emphasis role="bold">Choose a stripe pattern that takes into account the write
159 patterns of your application.</emphasis> Writes that cross an object boundary are
160 slightly less efficient than writes that go entirely to one server. If the file is
161 written in a consistent and aligned way, make the stripe size a multiple of the
162 <literal>write()</literal> size.</para>
167 <section xml:id="dbdoclet.50438209_78664">
169 <primary>striping</primary>
170 <secondary>configuration</secondary>
171 </indexterm>Setting the File Layout/Striping Configuration (<literal>lfs
172 setstripe</literal>)</title>
173 <para>Use the <literal>lfs setstripe</literal> command to create new files with a specific file layout (stripe pattern) configuration.</para>
174 <screen>lfs setstripe [--size|-s stripe_size] [--count|-c stripe_count] \
175 [--index|-i start_ost] [--pool|-p pool_name] <replaceable>filename|dirname</replaceable> </screen>
176 <para><emphasis role="bold">
177 <literal>stripe_size</literal>
180 <para>The <literal>stripe_size</literal> indicates how much data to write to one OST before
181 moving to the next OST. The default <literal>stripe_size</literal> is 1 MB. Passing a
182 <literal>stripe_size</literal> of 0 causes the default stripe size to be used. Otherwise,
183 the <literal>stripe_size</literal> value must be a multiple of 64 KB.</para>
184 <para><emphasis role="bold">
185 <literal>stripe_count</literal>
188 <para>The <literal>stripe_count</literal> indicates how many OSTs to use. The default <literal>stripe_count</literal> value is 1. Setting <literal>stripe_count</literal> to 0 causes the default stripe count to be used. Setting <literal>stripe_count</literal> to -1 means stripe over all available OSTs (full OSTs are skipped).</para>
189 <para><emphasis role="bold">
190 <literal>start_ost</literal>
193 <para>The start OST is the first OST to which files are written. The default value for
194 <literal>start_ost</literal> is -1, which allows the MDS to choose the starting index. This
195 setting is strongly recommended, as it allows space and load balancing to be done by the MDS
196 as needed. If the value of <literal>start_ost</literal> is set to a value other than -1, the
197 file starts on the specified OST index. OST index numbering starts at 0.</para>
199 <para>If the specified OST is inactive or in a degraded mode, the MDS will silently choose
200 another target.</para>
203 <para>If you pass a <literal>start_ost</literal> value of 0 and a
204 <literal>stripe_count</literal> value of <emphasis>1</emphasis>, all files are written to
205 OST 0, until space is exhausted. <emphasis role="italic">This is probably not what you meant
206 to do.</emphasis> If you only want to adjust the stripe count and keep the other
207 parameters at their default settings, do not specify any of the other parameters:</para>
208 <para><screen>client# lfs setstripe -c <replaceable>stripe_count</replaceable> <replaceable>filename</replaceable></screen></para>
210 <para><emphasis role="bold">
211 <literal>pool_name</literal>
214 <para>The <literal>pool_name</literal> specifies the OST pool to which the file will be written.
215 This allows limiting the OSTs used to a subset of all OSTs in the file system. For more
216 details about using OST pools, see <link xl:href="ManagingFileSystemIO.html#50438211_75549"
217 >Creating and Managing OST Pools</link>.</para>
219 <title>Specifying a File Layout (Striping Pattern) for a Single File</title>
220 <para>It is possible to specify the file layout when a new file is created using the command <literal>lfs setstripe</literal>. This allows users to override the file system default parameters to tune the file layout more optimally for their application. Execution of an <literal>lfs setstripe</literal> command fails if the file already exists.</para>
221 <section xml:id="dbdoclet.50438209_60155">
222 <title>Setting the Stripe Size</title>
223 <para>The command to create a new file with a specified stripe size is similar to:</para>
224 <screen>[client]# lfs setstripe -s 4M /mnt/lustre/new_file</screen>
225 <para>This example command creates the new file <literal>/mnt/lustre/new_file</literal> with a stripe size of 4 MB.</para>
226 <para>Now, when the file is created, the new stripe setting creates the file on a single OST with a stripe size of 4M:</para>
227 <screen> [client]# lfs getstripe /mnt/lustre/new_file
230 lmm_stripe_size: 4194304
234 obdidx objid objid group
235 1 690550 0xa8976 0 </screen>
236 <para>In this example, the stripe size is 4 MB.</para>
239 <title><indexterm><primary>striping</primary><secondary>count</secondary></indexterm>
240 Setting the Stripe Count</title>
241 <para>The command below creates a new file with a stripe count of <literal>-1</literal> to
242 specify striping over all available OSTs:</para>
243 <screen>[client]# lfs setstripe -c -1 /mnt/lustre/full_stripe</screen>
244 <para>The example below indicates that the file <literal>full_stripe</literal> is striped
245 over all six active OSTs in the configuration:</para>
246 <screen>[client]# lfs getstripe /mnt/lustre/full_stripe
247 /mnt/lustre/full_stripe
248 obdidx objid objid group
255 <para> This is in contrast to the output in <xref linkend="dbdoclet.50438209_60155"/>, which
256 shows only a single object for the file.</para>
261 <primary>striping</primary>
262 <secondary>per directory</secondary>
263 </indexterm>Setting the Striping Layout for a Directory</title>
264 <para>In a directory, the <literal>lfs setstripe</literal> command sets a default striping
265 configuration for files created in the directory. The usage is the same as <literal>lfs
266 setstripe</literal> for a regular file, except that the directory must exist prior to
267 setting the default striping configuration. If a file is created in a directory with a
268 default stripe configuration (without otherwise specifying striping), the Lustre file system
269 uses those striping parameters instead of the file system default for the new file.</para>
270 <para>To change the striping pattern for a sub-directory, create a directory with desired file
271 layout as described above. Sub-directories inherit the file layout of the root/parent
276 <primary>striping</primary>
277 <secondary>per file system</secondary>
278 </indexterm>Setting the Striping Layout for a File System</title>
279 <para>Setting the striping specification on the <literal>root</literal> directory determines
280 the striping for all new files created in the file system unless an overriding striping
281 specification takes precedence (such as a striping layout specified by the application, or
282 set using <literal>lfs setstripe</literal>, or specified for the parent directory).</para>
284 <para>The striping settings for a <literal>root</literal> directory are, by default, applied
285 to any new child directories created in the root directory, unless striping settings have
286 been specified for the child directory.</para>
291 <primary>striping</primary>
292 <secondary>on specific OST</secondary>
293 </indexterm>Creating a File on a Specific OST</title>
294 <para>You can use <literal>lfs setstripe</literal> to create a file on a specific OST. In the
295 following example, the file <literal>file1</literal> is created on the first OST (OST index
297 <screen>$ lfs setstripe --count 1 --index 0 file1
298 $ dd if=/dev/zero of=file1 count=1 bs=100M
302 $ lfs getstripe file1
305 lmm_stripe_size: 1048576
309 obdidx objid objid group
310 0 37364 0x91f4 0</screen>
313 <section xml:id="dbdoclet.50438209_44776">
314 <title><indexterm><primary>striping</primary><secondary>getting information</secondary></indexterm>Retrieving File Layout/Striping Information (<literal>getstripe</literal>)</title>
315 <para>The <literal>lfs getstripe</literal> command is used to display information that shows
316 over which OSTs a file is distributed. For each OST, the index and UUID is displayed, along
317 with the OST index and object ID for each stripe in the file. For directories, the default
318 settings for files created in that directory are displayed.</para>
320 <title>Displaying the Current Stripe Size</title>
321 <para>To see the current stripe size for a Lustre file or directory, use the <literal>lfs
322 getstripe</literal> command. For example, to view information for a directory, enter a
323 command similar to:</para>
324 <screen>[client]# lfs getstripe /mnt/lustre </screen>
325 <para>This command produces output similar to:</para>
327 (Default) stripe_count: 1 stripe_size: 1M stripe_offset: -1</screen>
328 <para>In this example, the default stripe count is <literal>1</literal> (data blocks are
329 striped over a single OST), the default stripe size is 1 MB, and the objects are created
330 over all available OSTs.</para>
331 <para>To view information for a file, enter a command similar to:</para>
332 <screen>$ lfs getstripe /mnt/lustre/foo
335 lmm_stripe_size: 1048576
339 obdidx objid objid group
340 2 835487 m0xcbf9f 0 </screen>
341 <para>In this example, the file is located on <literal>obdidx 2</literal>, which corresponds
342 to the OST <literal>lustre-OST0002</literal>. To see which node is serving that OST, run:
343 <screen>$ lctl get_param osc.lustre-OST0002-osc.ost_conn_uuid
344 osc.lustre-OST0002-osc.ost_conn_uuid=192.168.20.1@tcp</screen></para>
347 <title>Inspecting the File Tree</title>
348 <para>To inspect an entire tree of files, use the <literal>lfs find</literal> command:</para>
349 <screen>lfs find [--recursive | -r] <replaceable>file|directory</replaceable> ...</screen>
353 <primary>striping</primary>
354 <secondary>remote directories</secondary>
355 </indexterm>Locating the MDT for a remote directory</title>
356 <para condition="l24">Lustre software release 2.4 can be configured with
357 multiple MDTs in the same file system. Each sub-directory can have a
358 different MDT. To identify on which MDT a given subdirectory is
359 located, pass the <literal>getstripe [--mdt-index|-M]</literal>
360 parameters to <literal>lfs</literal>. An example of this command is
361 provided in the section <xref linkend="dbdoclet.rmremotedir"/>.</para>
364 <section xml:id="dbdoclet.50438209_10424">
366 <primary>space</primary>
367 <secondary>free space</secondary>
368 </indexterm><indexterm>
369 <primary>striping</primary>
370 <secondary>round-robin algorithm</secondary>
371 </indexterm><indexterm>
372 <primary>striping</primary>
373 <secondary>weighted algorithm</secondary>
374 </indexterm><indexterm>
375 <primary>round-robin algorithm</primary>
376 </indexterm><indexterm>
377 <primary>weighted algorithm</primary>
378 </indexterm>Managing Free Space</title>
379 <para>To optimize file system performance, the MDT assigns file stripes to OSTs based on two
380 allocation algorithms. The <emphasis role="italic">round-robin</emphasis> allocator gives
381 preference to location (spreading out stripes across OSSs to increase network bandwidth
382 utilization) and the weighted allocator gives preference to available space (balancing loads
383 across OSTs). Threshold and weighting factors for these two algorithms can be adjusted by the
384 user. The MDT reserves 0.1 percent of total OST space and 32 inodes for each OST. The MDT
385 stops object allocation for the OST if available space is less than reserved or the OST has
386 fewer than 32 free inodes. The MDT starts object allocation when available space is twice
387 as big as the reserved space and the OST has more than 64 free inodes. Note, clients
388 could append existing files no matter what object allocation state is.</para>
389 <para condition="l29"> The reserved space for each OST can be adjusted by the user. Use the
390 <literal>lctl set_param</literal> command, for example the next command reserve 1GB space
392 <screen>lctl set_param -P osp.*.reserved_mb_low=1024</screen></para>
393 <para>This section describes how to check available free space on disks and how free space is
394 allocated. It then describes how to set the threshold and weighting factors for the allocation
396 <section xml:id="dbdoclet.50438209_35838">
397 <title>Checking File System Free Space</title>
398 <para>Free space is an important consideration in assigning file stripes. The <literal>lfs
399 df</literal> command can be used to show available disk space on the mounted Lustre file
400 system and space consumption per OST. If multiple Lustre file systems are mounted, a path
401 may be specified, but is not required. Options to the <literal>lfs df</literal> command are
403 <informaltable frame="all">
405 <colspec colname="c1" colwidth="50*"/>
406 <colspec colname="c2" colwidth="50*"/>
410 <para><emphasis role="bold">Option</emphasis></para>
413 <para><emphasis role="bold">Description</emphasis></para>
420 <para> <literal>-h</literal></para>
423 <para> Displays sizes in human readable format (for example: 1K, 234M, 5G).</para>
428 <para> <literal role="bold">-i, --inodes</literal></para>
431 <para> Lists inodes instead of block usage.</para>
438 <para>The <literal>df -i</literal> and <literal>lfs df -i</literal> commands show the
439 <emphasis role="italic">minimum</emphasis> number of inodes that can be created in the
440 file system at the current time. If the total number of objects available across all of
441 the OSTs is smaller than those available on the MDT(s), taking into account the default
442 file striping, then <literal>df -i</literal> will also report a smaller number of inodes
443 than could be created. Running <literal>lfs df -i</literal> will report the actual number
444 of inodes that are free on each target.</para>
445 <para>For ZFS file systems, the number of inodes that can be created is dynamic and depends
446 on the free space in the file system. The Free and Total inode counts reported for a ZFS
447 file system are only an estimate based on the current usage for each target. The Used
448 inode count is the actual number of inodes used by the file system.</para>
450 <para><emphasis role="bold">Examples</emphasis></para>
451 <screen>[client1] $ lfs df
452 UUID 1K-blockS Used Available Use% Mounted on
453 mds-lustre-0_UUID 9174328 1020024 8154304 11% /mnt/lustre[MDT:0]
454 ost-lustre-0_UUID 94181368 56330708 37850660 59% /mnt/lustre[OST:0]
455 ost-lustre-1_UUID 94181368 56385748 37795620 59% /mnt/lustre[OST:1]
456 ost-lustre-2_UUID 94181368 54352012 39829356 57% /mnt/lustre[OST:2]
457 filesystem summary: 282544104 167068468 39829356 57% /mnt/lustre
459 [client1] $ lfs df -h
460 UUID bytes Used Available Use% Mounted on
461 mds-lustre-0_UUID 8.7G 996.1M 7.8G 11% /mnt/lustre[MDT:0]
462 ost-lustre-0_UUID 89.8G 53.7G 36.1G 59% /mnt/lustre[OST:0]
463 ost-lustre-1_UUID 89.8G 53.8G 36.0G 59% /mnt/lustre[OST:1]
464 ost-lustre-2_UUID 89.8G 51.8G 38.0G 57% /mnt/lustre[OST:2]
465 filesystem summary: 269.5G 159.3G 110.1G 59% /mnt/lustre
467 [client1] $ lfs df -i
468 UUID Inodes IUsed IFree IUse% Mounted on
469 mds-lustre-0_UUID 2211572 41924 2169648 1% /mnt/lustre[MDT:0]
470 ost-lustre-0_UUID 737280 12183 725097 1% /mnt/lustre[OST:0]
471 ost-lustre-1_UUID 737280 12232 725048 1% /mnt/lustre[OST:1]
472 ost-lustre-2_UUID 737280 12214 725066 1% /mnt/lustre[OST:2]
473 filesystem summary: 2211572 41924 2169648 1% /mnt/lustre[OST:2]</screen>
477 <primary>striping</primary>
478 <secondary>allocations</secondary>
479 </indexterm> Stripe Allocation Methods</title>
480 <para>Two stripe allocation methods are provided:</para>
483 <para><emphasis role="bold">Round-robin allocator</emphasis> - When the OSTs have
484 approximately the same amount of free space, the round-robin allocator alternates
485 stripes between OSTs on different OSSs, so the OST used for stripe 0 of each file is
486 evenly distributed among OSTs, regardless of the stripe count. In a simple example with
487 eight OSTs numbered 0-7, objects would be allocated like this:</para>
489 <screen>File 1: OST1, OST2, OST3, OST4
490 File 2: OST5, OST6, OST7
491 File 3: OST0, OST1, OST2, OST3, OST4, OST5
492 File 4: OST6, OST7, OST0</screen>
494 <para>Here are several more sample round-robin stripe orders (each letter represents a
495 different OST on a single OSS):</para>
496 <informaltable frame="none">
498 <colspec colname="c1" colwidth="50*"/>
499 <colspec colname="c2" colwidth="50*"/>
506 <para> One 3-OST OSS</para>
511 <para> 3x3: ABABAB</para>
514 <para> Two 3-OST OSSs</para>
519 <para> 3x4: BBABABA</para>
522 <para> One 3-OST OSS (A) and one 4-OST OSS (B)</para>
527 <para> 3x5: BBABBABA</para>
530 <para> One 3-OST OSS (A) and one 5-OST OSS (B)</para>
535 <para> 3x3x3: ABCABCABC</para>
538 <para> Three 3-OST OSSs</para>
546 <para><emphasis role="bold">Weighted allocator</emphasis> - When the free space difference
547 between the OSTs becomes significant, the weighting algorithm is used to influence OST
548 ordering based on size (amount of free space available on each OST) and location
549 (stripes evenly distributed across OSTs). The weighted allocator fills the emptier OSTs
550 faster, but uses a weighted random algorithm, so the OST with the most free space is not
551 necessarily chosen each time.</para>
554 <para>The allocation method is determined by the amount of free-space imbalance on the OSTs.
555 When free space is relatively balanced across OSTs, the faster round-robin allocator is
556 used, which maximizes network balancing. The weighted allocator is used when any two OSTs
557 are out of balance by more than the specified threshold (17% by default). The threshold
558 between the two allocation methods is defined in the file
559 <literal>/proc/fs/<replaceable>fsname</replaceable>/lov/<replaceable>fsname</replaceable>-mdtlov/qos_threshold_rr</literal>. </para>
560 <para>To set the <literal>qos_threshold_r</literal> to <literal>25</literal>, enter this
562 MGS:<screen>lctl set_param lov.<replaceable>fsname</replaceable>-mdtlov.qos_threshold_rr=25</screen></para>
566 <primary>space</primary>
567 <secondary>location weighting</secondary>
568 </indexterm>Adjusting the Weighting Between Free Space and Location</title>
569 <para>The weighting priority used by the weighted allocator is set in the file
570 <literal>/proc/fs/<replaceable>fsname</replaceable>/lov/<replaceable>fsname</replaceable>-mdtlov/qos_prio_free</literal>.
571 Increasing the value of <literal>qos_prio_free</literal> puts more weighting on the amount
572 of free space available on each OST and less on how stripes are distributed across OSTs. The
573 default value is <literal>91</literal> (percent). When the free space priority is set to
574 <literal>100</literal> (percent), weighting is based entirely on free space and location
575 is no longer used by the striping algorithm. </para>
576 <para>To change the allocator weighting to <literal>100</literal>, enter this command on the
578 <screen>lctl conf_param <replaceable>fsname</replaceable>-MDT0000.lov.qos_prio_free=100</screen>
581 <para>When <literal>qos_prio_free</literal> is set to <literal>100</literal>, a weighted
582 random algorithm is still used to assign stripes, so, for example, if OST2 has twice as
583 much free space as OST1, OST2 is twice as likely to be used, but it is not guaranteed to
588 <section xml:id="section_syy_gcl_qk">
590 <primary>striping</primary>
591 <secondary>wide striping</secondary>
592 </indexterm><indexterm>
593 <primary>wide striping</primary>
594 </indexterm>Lustre Striping Internals</title>
595 <para>For Lustre releases prior to Lustre software release 2.2, files can be striped across a
596 maximum of 160 OSTs. Lustre inodes use an extended attribute to record the location of each
597 object (the object ID and the number of the OST on which it is stored). The size of the
598 extended attribute limits the maximum stripe count to 160 objects.</para>
599 <para condition="l22">In Lustre software release 2.2 and subsequent releases, the maximum number
600 of OSTs over which files can be striped has been raised to 2000 by allocating a new block on
601 which to store the extended attribute that holds the object information. This feature, known
602 as "wide striping," only allocates the additional extended attribute data block if the file is
603 striped with a stripe count greater than 160. The file layout (object ID, OST number) is
604 stored on the new data block with a pointer to this block stored in the original Lustre inode
605 for the file. For files smaller than 160 objects, the Lustre inode is used to store the file