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"
4 xml:id="upgradinglustre">
5 <title xml:id="upgradinglustre.title">Upgrading a Lustre File System</title>
6 <para>This chapter describes interoperability between Lustre software
7 releases. It also provides procedures for upgrading from Lustre software
8 release 1.8 to Lustre software release 2.x , from a Lustre software release
9 2.x to a more recent Lustre software release 2.x (major release upgrade), and
10 from a a Lustre software release 2.x.y to a more recent Lustre software
11 release 2.x.y (minor release upgrade). It includes the following
16 <xref linkend="dbdoclet.50438205_82079" />
21 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
22 linkend="Upgrading_2.x" />
27 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
28 linkend="Upgrading_2.x.x" />
32 <section xml:id="dbdoclet.50438205_82079">
35 <primary>Lustre</primary>
36 <secondary>upgrading</secondary>
40 <primary>upgrading</primary>
41 </indexterm>Release Interoperability and Upgrade Requirements</title>
43 <emphasis role="italic">
44 <emphasis role="bold">Lustre software release 2.x (major)
49 <para>All servers must be upgraded at the same time, while some or
50 all clients may be upgraded independently of the servers.</para>
53 <para>All servers must be be upgraded to a Linux kernel supported by
54 the Lustre software. See the Lustre Release Notes for your Lustre
55 version for a list of tested Linux distributions.</para>
58 <para>Clients to be upgraded must be running a compatible Linux
59 distribution as described in the Release Notes.</para>
64 <emphasis role="italic">
65 <emphasis role="bold">Lustre software release 2.x.y release (minor)
71 <para>All servers must be upgraded at the same time, while some or all
72 clients may be upgraded.</para>
75 <para>Rolling upgrades are supported for minor releases allowing
76 individual servers and clients to be upgraded without stopping the
77 Lustre file system.</para>
81 <section xml:id="Upgrading_2.x">
84 <primary>upgrading</primary>
85 <secondary>major release (2.x to 2.x)</secondary>
88 <primary>wide striping</primary>
91 <primary>MDT</primary>
92 <secondary>multiple MDSs</secondary>
95 <primary>large_xattr</primary>
96 <secondary>ea_inode</secondary>
99 <primary>wide striping</primary>
100 <secondary>large_xattr</secondary>
101 <tertiary>ea_inode</tertiary>
102 </indexterm>Upgrading to Lustre Software Release 2.x (Major
104 <para>The procedure for upgrading from a Lustre software release 2.x to a
105 more recent 2.x release of the Lustre software is described in this
108 <para>This procedure can also be used to upgrade Lustre software release
109 1.8.6-wc1 or later to any Lustre software release 2.x. To upgrade other
110 versions of Lustre software release 1.8.x, contact your support
114 <para condition="l22">In Lustre software release 2.2, a feature has been
115 added that allows striping across up to 2000 OSTs. By default, this "wide
116 striping" feature is disabled. It is activated by setting the
117 <literal>large_xattr</literal> or
118 <literal>ea_inode</literal> option on the MDT using either
119 <literal>mkfs.lustre</literal> or
120 <literal>tune2fs</literal>. For example after upgrading an existing file
121 system to Lustre software release 2.2 or later, wide striping can be
122 enabled by running the following command on the MDT device before
124 <screen>tune2fs -O large_xattr</screen>
125 Once the wide striping feature is enabled and in use on the MDT, it is
126 not possible to directly downgrade the MDT file system to an earlier
127 version of the Lustre software that does not support wide striping. To
128 disable wide striping:
131 <para>Delete all wide-striped files.</para>
134 <literal>lfs_migrate</literal> with the option
135 <literal>-c</literal>
136 <replaceable>stripe_count</replaceable>(set
137 <replaceable>stripe_count</replaceable>to 160) to move the files to
138 another location.</para>
141 <para>Unmount the MDT.</para>
144 <para>Run the following command to turn off the
145 <literal>large_xattr</literal> option:
146 <screen>tune2fs -O ^large_xattr</screen></para>
148 </orderedlist>Using either
149 <literal>mkfs.lustre</literal> or
150 <literal>tune2fs</literal> with
151 <literal>large_xattr</literal> or
152 <literal>ea_inode</literal> option reseults in
153 <literal>ea_inode</literal> in the file system feature list.</para>
155 <note condition="l23">
156 <para>To generate a list of all files with more than 160 stripes use
157 <literal>lfs find</literal> with the
158 <literal>--stripe-count</literal> option:
159 <screen>lfs find ${mountpoint} --stripe-count=+160</screen></para>
161 <note condition="l24">
162 <para>In Lustre software release 2.4, a new feature allows using multiple
163 MDTs, which can each serve one or more remote sub-directories in the file
165 <literal>root</literal> directory is always located on MDT0.</para>
166 <para>Note that clients running a release prior to the Lustre software
167 release 2.4 can only see the namespace hosted by MDT0 and will return an
168 IO error if an attempt is made to access a directory on another
171 <para>To upgrade a Lustre software release 2.x to a more recent major
172 release, complete these steps:</para>
175 <para>Create a complete, restorable file system backup.</para>
177 <para>Before installing the Lustre software, back up ALL data. The
178 Lustre software contains kernel modifications that interact with
179 storage devices and may introduce security issues and data loss if
180 not installed, configured, or administered properly. If a full backup
181 of the file system is not practical, a device-level backup of the MDT
182 file system is recommended. See
183 <xref linkend="backupandrestore" /> for a procedure.</para>
187 <para>Shut down the file system by unmounting all clients and servers
188 in the order shown below (unmounting a block device causes the Lustre
189 software to be shut down on that node):</para>
190 <orderedlist numeration="loweralpha">
192 <para>Unmount the clients. On each client node, run:</para>
193 <screen>umount -a -t lustre</screen>
196 <para>Unmount the MDT. On the MDS node, run:</para>
197 <screen>umount -a -t lustre</screen>
200 <para>Unmount all the OSTs. On each OSS node, run:</para>
201 <screen>umount -a -t lustre</screen>
206 <para>Upgrade the Linux operating system on all servers to a compatible
207 (tested) Linux distribution and reboot.</para>
210 <para>Upgrade the Linux operating system on all clients to Red Hat
211 Enterprise Linux 6 or other compatible (tested) distribution and
215 <para>Download the Lustre server RPMs for your platform from the
216 <link xl:href="https://wiki.hpdd.intel.com/display/PUB/Lustre+Releases">
217 Lustre Releases</link>repository. See
218 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
219 linkend="table_cnh_5m3_gk" />for a list of required packages.</para>
222 <para>Install the Lustre server packages on all Lustre servers (MGS,
223 MDSs, and OSSs).</para>
224 <orderedlist numeration="loweralpha">
226 <para>Log onto a Lustre server as the
227 <literal>root</literal> user</para>
231 <literal>yum</literal> command to install the packages:</para>
233 <screen># yum --nogpgcheck install pkg1.rpm pkg2.rpm ... </screen>
237 <para>Verify the packages are installed correctly:</para>
239 <screen>rpm -qa|egrep "lustre|wc"</screen>
243 <para>Repeat these steps on each Lustre server.</para>
248 <para>Download the Lustre client RPMs for your platform from the
249 <link xl:href="https://wiki.hpdd.intel.com/display/PUB/Lustre+Releases">
250 Lustre Releases</link>repository. See
251 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
252 linkend="table_j3r_ym3_gk" />for a list of required packages.</para>
254 <para>The version of the kernel running on a Lustre client must be
255 the same as the version of the
256 <literal>lustre-client-modules-</literal>
257 <replaceable>ver</replaceable>package being installed. If not, a
258 compatible kernel must be installed on the client before the Lustre
259 client packages are installed.</para>
263 <para>Install the Lustre client packages on each of the Lustre clients
264 to be upgraded.</para>
265 <orderedlist numeration="loweralpha">
267 <para>Log onto a Lustre client as the
268 <literal>root</literal> user.</para>
272 <literal>yum</literal> command to install the packages:</para>
274 <screen># yum --nogpgcheck install pkg1.rpm pkg2.rpm ... </screen>
278 <para>Verify the packages were installed correctly:</para>
280 <screen># rpm -qa|egrep "lustre|kernel"</screen>
284 <para>Repeat these steps on each Lustre client.</para>
289 <para>(Optional) For upgrades to Lustre software release 2.2 or higher,
290 to enable wide striping on an existing MDT, run the following command
292 <screen>mdt# tune2fs -O large_xattr <replaceable>device</replaceable></screen></para>
293 <para>For more information about wide striping, see
294 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
295 linkend="section_syy_gcl_qk" />.</para>
298 <para>(Optional) For upgrades to Lustre software release 2.4 or higher,
299 to format an additional MDT, complete these steps:
300 <orderedlist numeration="loweralpha">
302 <para>Determine the index used for the first MDT (each MDT must
303 have unique index). Enter:
304 <screen>client$ lctl dl | grep mdc
305 36 UP mdc lustre-MDT0000-mdc-ffff88004edf3c00
306 4c8be054-144f-9359-b063-8477566eb84e 5</screen></para>
307 <para>In this example, the next available index is 1.</para>
310 <para>Add the new block device as a new MDT at the next available
311 index by entering (on one line):
312 <screen>mds# mkfs.lustre --reformat --fsname=<replaceable>filesystem_name</replaceable> --mdt \
313 --mgsnode=<replaceable>mgsnode</replaceable> --index <replaceable>1</replaceable>
314 <replaceable>/dev/mdt1_device</replaceable></screen></para>
316 </orderedlist></para>
319 <para>(Optional) If you are upgrading to Lustre software release 2.3 or
320 higher from Lustre software release 2.2 or earlier and want to enable
321 the quota feature, complete these steps:
322 <orderedlist numeration="loweralpha">
324 <para>Before setting up the file system, enter on both the MDS and
326 <screen>tunefs.lustre --quota</screen></para>
329 <para>When setting up the file system, enter:
330 <screen>conf_param $FSNAME.quota.mdt=$QUOTA_TYPE
331 conf_param $FSNAME.quota.ost=$QUOTA_TYPE</screen></para>
333 </orderedlist></para>
336 <para>(Optional) If you are upgrading from Lustre software release 1.8,
337 you must manually enable the FID-in-dirent feature. On the MDS, enter:
338 <screen>tune2fs –O dirdata /dev/<replaceable>mdtdev</replaceable></screen></para>
340 <para>This step is not reversible. Do not complete this step until
341 you are sure you will not be downgrading the Lustre software.</para>
343 <para condition='l24'>This step only enables FID-in-dirent for newly
344 created files. If you are upgrading to Lustre software release 2.4,
345 you can use LFSCK to enable FID-in-dirent for existing files. For
346 more information about FID-in-dirent and related functionalities in
347 LFSCK, see <xref xmlns:xlink="http://www.w3.org/1999/xlink"
348 linkend="understandinglustre.storageio" />.</para>
351 <para>Start the Lustre file system by starting the components in the
352 order shown in the following steps:</para>
353 <orderedlist numeration="loweralpha">
355 <para>Mount the MGT. On the MGS, run
356 <screen>mgs# mount -a -t lustre</screen></para>
359 <para>Mount the MDT(s). On each MDT, run:
360 <screen>mds# mount -a -t lustre</screen></para>
363 <para>Mount all the OSTs. On each OSS node, run:</para>
364 <screen>oss# mount -a -t lustre</screen>
366 <para>This command assumes that all the OSTs are listed in the
367 <literal>/etc/fstab</literal> file. OSTs that are not listed in
369 <literal>/etc/fstab</literal> file, must be mounted individually
370 by running the mount command:</para>
371 <screen>mount -t lustre <replaceable>/dev/block_device</replaceable><replaceable>/mount_point</replaceable></screen>
375 <para>Mount the file system on the clients. On each client node,
377 <screen>client# mount -a -t lustre</screen>
383 <para>The mounting order described in the steps above must be followed
384 for the initial mount and registration of a Lustre file system after an
385 upgrade. For a normal start of a Lustre file system, the mounting order
386 is MGT, OSTs, MDT(s), clients.</para>
388 <para>If you have a problem upgrading a Lustre file system, see
389 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
390 linkend="dbdoclet.50438198_30989" />for some ways to get help.</para>
392 <section xml:id="Upgrading_2.x.x">
395 <primary>upgrading</primary>
396 <secondary>2.X.y to 2.X.y (minor release)</secondary>
397 </indexterm>Upgrading to Lustre Software Release 2.x.y (Minor
399 <para>Rolling upgrades are supported for upgrading from any Lustre software
400 release 2.x.y to a more recent Lustre software release 2.X.y. This allows
401 the Lustre file system to continue to run while individual servers (or
402 their failover partners) and clients are upgraded one at a time. The
403 procedure for upgrading a Lustre software release 2.x.y to a more recent
404 minor release is described in this section.</para>
405 <para>To upgrade Lustre software release 2.x.y to a more recent minor
406 release, complete these steps:</para>
409 <para>Create a complete, restorable file system backup.</para>
411 <para>Before installing the Lustre software, back up ALL data. The
412 Lustre software contains kernel modifications that interact with
413 storage devices and may introduce security issues and data loss if
414 not installed, configured, or administered properly. If a full backup
415 of the file system is not practical, a device-level backup of the MDT
416 file system is recommended. See
417 <xref linkend="backupandrestore" /> for a procedure.</para>
421 <para>Download the Lustre server RPMs for your platform from the
422 <link xl:href="https://wiki.hpdd.intel.com/display/PUB/Lustre+Releases">
423 Lustre Releases</link>repository. See
424 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
425 linkend="table_cnh_5m3_gk" />for a list of required packages.</para>
428 <para>For a rolling upgrade, complete any procedures required to keep
429 the Lustre file system running while the server to be upgraded is
430 offline, such as failing over a primary server to its secondary
434 <para>Unmount the Lustre server to be upgraded (MGS, MDS, or
438 <para>Install the Lustre server packages on the Lustre server.</para>
439 <orderedlist numeration="loweralpha">
441 <para>Log onto the Lustre server as the
442 <literal>root</literal> user</para>
446 <literal>yum</literal> command to install the packages:</para>
448 <screen># yum --nogpgcheck install pkg1.rpm pkg2.rpm ... </screen>
452 <para>Verify the packages are installed correctly:</para>
454 <screen>rpm -qa|egrep "lustre|wc"</screen>
458 <para>Mount the Lustre server to restart the Lustre software on the
460 <screen>server# mount -a -t lustre</screen></para>
463 <para>Repeat these steps on each Lustre server.</para>
468 <para>Download the Lustre client RPMs for your platform from the
469 <link xl:href="https://wiki.hpdd.intel.com/display/PUB/Lustre+Releases">
470 Lustre Releases</link>repository. See
471 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
472 linkend="table_j3r_ym3_gk" />for a list of required packages.</para>
475 <para>Install the Lustre client packages on each of the Lustre clients
476 to be upgraded.</para>
477 <orderedlist numeration="loweralpha">
479 <para>Log onto a Lustre client as the
480 <literal>root</literal> user.</para>
484 <literal>yum</literal> command to install the packages:</para>
486 <screen># yum --nogpgcheck install pkg1.rpm pkg2.rpm ... </screen>
490 <para>Verify the packages were installed correctly:</para>
492 <screen># rpm -qa|egrep "lustre|kernel"</screen>
496 <para>Mount the Lustre client to restart the Lustre software on the
498 <screen>client# mount -a -t lustre</screen></para>
501 <para>Repeat these steps on each Lustre client.</para>
506 <para>If you have a problem upgrading a Lustre file system, see
507 <xref xmlns:xlink="http://www.w3.org/1999/xlink"
508 linkend="dbdoclet.50438198_30989" />for some suggestions for how to get