-<?xml version="1.0" encoding="UTF-8"?>
-<article version="5.0" xml:lang="en-US" xmlns="http://docbook.org/ns/docbook" xmlns:xl="http://www.w3.org/1999/xlink">
- <info>
- <title>Benchmarking Lustre Performance (Lustre I/O Kit)</title>
- </info>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289895" xreflabel=""/>This chapter describes the Lustre I/O kit, a collection of I/O benchmarking tools for a Lustre cluster, and PIOS, a parallel I/O simulator for Linux and Solaris. It includes:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289899" xreflabel=""/><link xl:href="BenchmarkingTests.html#50438212_44437">Using Lustre I/O Kit Tools</link></para>
+<?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="benchmarkingtests">
+ <title xml:id="benchmarkingtests.title">Benchmarking Lustre File System Performance (Lustre I/O
+ Kit)</title>
+ <para>This chapter describes the Lustre I/O kit, a collection of I/O benchmarking tools for a
+ Lustre cluster, and PIOS, a parallel I/O simulator for Linux and
+ Solaris<superscript>*</superscript> operating systems. It includes:</para>
+ <itemizedlist>
+ <listitem>
+ <para><xref linkend="dbdoclet.50438212_44437"/></para>
</listitem>
-<listitem>
- <para> </para>
+ <listitem>
+ <para><xref linkend="dbdoclet.50438212_51053"/></para>
</listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303084" xreflabel=""/><link xl:href="BenchmarkingTests.html#50438212_51053">Testing I/O Performance of Raw Hardware (sgpdd_survey)</link></para>
+ <listitem>
+ <para><xref linkend="dbdoclet.50438212_26516"/></para>
</listitem>
-<listitem>
- <para> </para>
+ <listitem>
+ <para><xref linkend="dbdoclet.50438212_85136"/></para>
</listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303088" xreflabel=""/><link xl:href="BenchmarkingTests.html#50438212_26516">Testing OST Performance (obdfilter_survey)</link></para>
+ <listitem>
+ <para><xref linkend="mds_survey_ref"/></para>
</listitem>
-<listitem>
- <para> </para>
+ <listitem>
+ <para><xref linkend="dbdoclet.50438212_58201"/></para>
</listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303097" xreflabel=""/><link xl:href="BenchmarkingTests.html#50438212_85136">Testing OST I/O Performance (ost_survey)</link></para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303101" xreflabel=""/><link xl:href="BenchmarkingTests.html#50438212_58201">Collecting Application Profiling Information (stats-collect)</link></para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <section remap="h2">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1289909" xreflabel=""/></title>
- <section remap="h2">
- <title>24.1 <anchor xml:id="dbdoclet.50438212_44437" xreflabel=""/>Using Lustre I/O Kit Tools</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1302857" xreflabel=""/>The tools in the Lustre I/O Kit are used to benchmark Lustre hardware and validate that it is working as expected before you install the Lustre software. It can also be used to to validate the performance of the various hardware and software layers in the cluster and also to find and troubleshoot I/O issues.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300612" xreflabel=""/>Typically, performance is measured starting with single raw devices and then proceeding to groups of devices. Once raw performance has been established, other software layers are then added incrementally and tested.</para>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1289911" xreflabel=""/>24.1.1 Contents of the Lustre I/O Kit</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303002" xreflabel=""/>The I/O kit contains three tests, each of which tests a progressively higher layer in the Lustre stack:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300548" xreflabel=""/>sgpdd_survey - Measure basic “bare metal†performance of devices while bypassing the kernel block device layers, buffer cache, and file system.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300572" xreflabel=""/>obdfilter_survey - Measure the performance of one or more OSTs directly on the OSS node or alternately over the network from a Lustre client.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300557" xreflabel=""/>ost_survey - Performs I/O against OSTs individually to allow performance comparisons to detect if an OST is performing suboptimally due to hardware issues.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300595" xreflabel=""/>Typically with these tests, Lustre should deliver 85-90% of the raw device performance.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300756" xreflabel=""/>A utility stats-collect is also provided to collect application profiling information from Lustre clients and servers. See <link xl:href="BenchmarkingTests.html#50438212_58201">Collecting Application Profiling Information (stats-collect)</link> for more information.</para>
- </section>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1289914" xreflabel=""/>24.1.2 Preparing to Use the Lustre <anchor xml:id="dbdoclet.50438212_marker-1289913" xreflabel=""/>I/O Kit</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300699" xreflabel=""/>The following prerequisites must be met to use the tests in the Lustre I/O kit:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300700" xreflabel=""/> Password-free remote access to nodes in the system (provided by ssh or rsh).</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303852" xreflabel=""/> LNET self-test completed to test that Lustre Networking has been properly installed and configured. See <link xl:href="LNETSelfTest.html#50438223_71556">Chapter 23</link>: <link xl:href="LNETSelfTest.html#50438223_21832">Testing Lustre Network Performance (LNET Self-Test)</link>.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300701" xreflabel=""/> Lustre file system software installed.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300702" xreflabel=""/>sg3_utils package providing the sgp_dd tool (sg3_utils is a separate RPM package available online using YUM).</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289915" xreflabel=""/>Download the Lustre I/O kit (lustre-iokit)from:</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300015" xreflabel=""/><link xl:href="http://downloads.lustre.org/public/tools/lustre-iokit/">http://downloads.lustre.org/public/tools/lustre-iokit/</link></para>
- </section>
- </section>
- <section remap="h2">
- <title>24.2 Testing I/O Performance of Raw Hardware (sgpdd_survey<anchor xml:id="dbdoclet.50438212_51053" xreflabel=""/><anchor xml:id="dbdoclet.50438212_marker-1302844" xreflabel=""/>)</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289936" xreflabel=""/>The sgpdd_survey tool is used to test bare metal I/O performance of the raw hardware, while bypassing as much of the kernel as possible. This survey may be used to characterize the performance of a SCSI device by simulating an OST serving multiple stripe files. The data gathered by this survey can help set expectations for the performance of a Lustre OST using this device.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300866" xreflabel=""/>The script uses sgp_dd to carry out raw sequential disk I/O. It runs with variable numbers of sgp_dd threads to show how performance varies with different request queue depths.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300867" xreflabel=""/>The script spawns variable numbers of sgp_dd instances, each reading or writing a separate area of the disk to demonstrate performance variance within a number of concurrent stripe files.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305925" xreflabel=""/>Several tips and insights for disk performance measurement are described below. Some of this information is specific to RAID arrays and/or the Linux RAID implementation.</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305926" xreflabel=""/><emphasis>Performance is limited by the slowest disk.</emphasis></para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305927" xreflabel=""/>Before creating a RAID array, benchmark all disks individually. We have frequently encountered situations where drive performance was not consistent for all devices in the array. Replace any disks that are significantly slower than the rest.</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305928" xreflabel=""/><emphasis>Disks and arrays are very sensitive to request size.</emphasis></para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305929" xreflabel=""/>To identify the optimal request size for a given disk, benchmark the disk with different record sizes ranging from 4 KB to 1 to 2 MB.</para>
- <informaltable frame="none">
- <tgroup cols="2">
- <colspec colname="c1" colwidth="5*"/>
- <colspec colname="c2" colwidth="95*"/>
-
+ </itemizedlist>
+ <section xml:id="dbdoclet.50438212_44437">
+ <title>
+ <indexterm><primary>benchmarking</primary><secondary>with Lustre I/O Kit</secondary></indexterm>
+ <indexterm><primary>profiling</primary><see>benchmarking</see></indexterm>
+ <indexterm><primary>tuning</primary><see>benchmarking</see></indexterm>
+ <indexterm><primary>performance</primary><see>benchmarking</see></indexterm>
- <tbody>
- <row>
- <entry><para><inlinemediaobject><imageobject role="html">
- <imagedata fileref="./shared/caution.gif" scalefit="1"/>
- </imageobject>
-<imageobject role="fo">
- <imagedata contentdepth="100%" contentwidth="" depth="" fileref="./shared/caution.gif" scalefit="1" width="100%"/>
- </imageobject>
-</inlinemediaobject></para></entry>
- <entry><para><emphasis role="bold">Caution -</emphasis><anchor xml:id="dbdoclet.50438212_pgfId-1300876" xreflabel=""/>The sgpdd_survey script overwrites the device being tested, which results in the <emphasis>LOSS OF ALL DATA</emphasis> on that device. Exercise caution when selecting the device to be tested.</para></entry>
- </row>
- </tbody>
- </tgroup>
- </informaltable>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
- <tbody>
- <row>
- <entry><para><emphasis role="bold">Note -</emphasis> Array performance with all LUNs loaded does not always match the performance of a single LUN when tested in isolation.</para></entry>
- </row>
- </tbody>
- </tgroup>
- </informaltable>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303815" xreflabel=""/></para>
- <para><emphasis role="bold">Prequisites:</emphasis></para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300842" xreflabel=""/>sgp_dd tool in the sg3_utils package</para>
+ Using Lustre I/O Kit Tools</title>
+ <para>The tools in the Lustre I/O Kit are used to benchmark Lustre file system hardware and
+ validate that it is working as expected before you install the Lustre software. It can also be
+ used to to validate the performance of the various hardware and software layers in the cluster
+ and also to find and troubleshoot I/O issues.</para>
+ <para>Typically, performance is measured starting with single raw devices and then proceeding to groups of devices. Once raw performance has been established, other software layers are then added incrementally and tested.</para>
+ <section remap="h3">
+ <title>Contents of the Lustre I/O Kit</title>
+ <para>The I/O kit contains three tests, each of which tests a progressively higher layer in
+ the Lustre software stack:</para>
+ <itemizedlist>
+ <listitem>
+ <para><literal>sgpdd-survey</literal> - Measure basic 'bare metal' performance
+ of devices while bypassing the kernel block device layers, buffer cache, and file
+ system.</para>
</listitem>
-<listitem>
- <para> </para>
+ <listitem>
+ <para><literal>obdfilter-survey</literal> - Measure the performance of one or more OSTs
+ directly on the OSS node or alternately over the network from a Lustre client.</para>
</listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300845" xreflabel=""/> Lustre software is <emphasis>NOT</emphasis> required</para>
+ <listitem>
+ <para><literal>ost-survey</literal> - Performs I/O against OSTs individually to allow
+ performance comparisons to detect if an OST is performing suboptimally due to hardware
+ issues.</para>
</listitem>
-<listitem>
- <para> </para>
+ </itemizedlist>
+ <para>Typically with these tests, a Lustre file system should deliver 85-90% of the raw device
+ performance.</para>
+ <para>A utility <literal>stats-collect</literal> is also provided to collect application profiling information from Lustre clients and servers. See <xref linkend="dbdoclet.50438212_58201"/> for more information.</para>
+ </section>
+ <section remap="h3">
+ <title>Preparing to Use the Lustre I/O Kit</title>
+ <para>The following prerequisites must be met to use the tests in the Lustre I/O kit:</para>
+ <itemizedlist>
+ <listitem>
+ <para>Password-free remote access to nodes in the system (provided by <literal>ssh</literal> or <literal>rsh</literal>).</para>
</listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289939" xreflabel=""/>The device(s) being tested must meet one of these two requirements:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300909" xreflabel=""/> If the device is a SCSI device, it must appear in the output of sg_map (make sure the kernel module sg is loaded).</para>
+ <listitem>
+ <para>LNET self-test completed to test that Lustre networking has been properly installed
+ and configured. See <xref linkend="lnetselftest"/>.</para>
</listitem>
-<listitem>
- <para> </para>
+ <listitem>
+ <para>Lustre file system software installed.</para>
</listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300920" xreflabel=""/> If the device is a raw device, it must appear in the output of raw -qa.</para>
+ <listitem>
+ <para><literal>sg3_utils</literal> package providing the <literal>sgp_dd</literal> tool (<literal>sg3_utils</literal> is a separate RPM package available online using YUM).</para>
</listitem>
-<listitem>
- <para> </para>
+ </itemizedlist>
+ <para>Download the Lustre I/O kit (<literal>lustre-iokit</literal>)from:</para>
+ <para><link xl:href="http://downloads.hpdd.intel.com/">http://downloads.hpdd.intel.com/</link></para>
+ </section>
+ </section>
+ <section xml:id="dbdoclet.50438212_51053">
+ <title><indexterm>
+ <primary>benchmarking</primary>
+ <secondary>raw hardware with sgpdd-survey</secondary>
+ </indexterm>Testing I/O Performance of Raw Hardware (<literal>sgpdd-survey</literal>)</title>
+ <para>The <literal>sgpdd-survey</literal> tool is used to test bare metal I/O performance of the
+ raw hardware, while bypassing as much of the kernel as possible. This survey may be used to
+ characterize the performance of a SCSI device by simulating an OST serving multiple stripe
+ files. The data gathered by this survey can help set expectations for the performance of a
+ Lustre OST using this device.</para>
+ <para>The script uses <literal>sgp_dd</literal> to carry out raw sequential disk I/O. It runs with variable numbers of <literal>sgp_dd</literal> threads to show how performance varies with different request queue depths.</para>
+ <para>The script spawns variable numbers of <literal>sgp_dd</literal> instances, each reading or writing a separate area of the disk to demonstrate performance variance within a number of concurrent stripe files.</para>
+ <para>Several tips and insights for disk performance measurement are described below. Some of this information is specific to RAID arrays and/or the Linux RAID implementation.</para>
+ <itemizedlist>
+ <listitem>
+ <para><emphasis>Performance is limited by the slowest disk.</emphasis></para>
+ <para>Before creating a RAID array, benchmark all disks individually. We have frequently encountered situations where drive performance was not consistent for all devices in the array. Replace any disks that are significantly slower than the rest.</para>
+ </listitem>
+ <listitem>
+ <para><emphasis>Disks and arrays are very sensitive to request size.</emphasis></para>
+ <para>To identify the optimal request size for a given disk, benchmark the disk with different record sizes ranging from 4 KB to 1 to 2 MB.</para>
+ </listitem>
+ </itemizedlist>
+ <caution>
+ <para>The <literal>sgpdd-survey</literal> script overwrites the device being tested, which
+ results in the <emphasis>
+ <emphasis role="bold">LOSS OF ALL DATA</emphasis>
+ </emphasis> on that device. Exercise caution when selecting the device to be tested.</para>
+ </caution>
+ <note>
+ <para>Array performance with all LUNs loaded does not always match the performance of a single LUN when tested in isolation.</para>
+ </note>
+ <para><emphasis role="bold">Prerequisites:</emphasis></para>
+ <itemizedlist>
+ <listitem>
+ <para><literal>sgp_dd</literal> tool in the <literal>sg3_utils</literal> package</para>
+ </listitem>
+ <listitem>
+ <para>Lustre software is <emphasis>NOT</emphasis> required</para>
+ </listitem>
+ </itemizedlist>
+ <para>The device(s) being tested must meet one of these two requirements:</para>
+ <itemizedlist>
+ <listitem>
+ <para>If the device is a SCSI device, it must appear in the output of <literal>sg_map</literal> (make sure the kernel module <literal>sg</literal> is loaded).</para>
+ </listitem>
+ <listitem>
+ <para>If the device is a raw device, it must appear in the output of <literal>raw -qa</literal>.</para>
+ </listitem>
+ </itemizedlist>
+ <para>Raw and SCSI devices cannot be mixed in the test specification.</para>
+ <note>
+ <para>If you need to create raw devices to use the <literal>sgpdd-survey</literal> tool, note
+ that raw device 0 cannot be used due to a bug in certain versions of the "raw"
+ utility (including the version shipped with Red Hat Enterprise Linux 4U4.)</para>
+ </note>
+ <section remap="h3">
+ <title><indexterm><primary>benchmarking</primary><secondary>tuning storage</secondary></indexterm>Tuning Linux Storage Devices</title>
+ <para>To get large I/O transfers (1 MB) to disk, it may be necessary to tune several kernel parameters as specified:</para>
+ <screen>/sys/block/sdN/queue/max_sectors_kb = 4096
+/sys/block/sdN/queue/max_phys_segments = 256
+/proc/scsi/sg/allow_dio = 1
+/sys/module/ib_srp/parameters/srp_sg_tablesize = 255
+/sys/block/sdN/queue/scheduler</screen>
+ <note>
+ <para>Recommended schedulers are <emphasis role="bold">deadline</emphasis> and <emphasis
+ role="bold">noop</emphasis>. The scheduler is set by default to <emphasis role="bold"
+ >deadline</emphasis>, unless it has already been set to <emphasis role="bold"
+ >noop</emphasis>.</para>
+ </note>
+ </section>
+ <section remap="h3">
+ <title>Running sgpdd-survey</title>
+ <para>The <literal>sgpdd-survey</literal> script must be customized for the particular device
+ being tested and for the location where the script saves its working and result files (by
+ specifying the <literal>${rslt}</literal> variable). Customization variables are described
+ at the beginning of the script.</para>
+ <para>When the <literal>sgpdd-survey</literal> script runs, it creates a number of working
+ files and a pair of result files. The names of all the files created start with the prefix
+ defined in the variable <literal>${rslt}</literal>. (The default value is
+ <literal>/tmp</literal>.) The files include:</para>
+ <itemizedlist>
+ <listitem>
+ <para>File containing standard output data (same as <literal>stdout</literal>)</para>
+ <screen><replaceable>rslt_date_time</replaceable>.summary</screen>
</listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1300930" xreflabel=""/>Raw and SCSI devices cannot be mixed in the test specification.</para>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
- <tbody>
- <row>
- <entry><para><emphasis role="bold">Note -</emphasis><anchor xml:id="dbdoclet.50438212_pgfId-1300900" xreflabel=""/>If you need to create raw devices to use the sgpdd_survey tool, note that raw device 0 cannot be used due to a bug in certain versions of the "raw" utility (including that shipped with RHEL4U4.)</para></entry>
- </row>
- </tbody>
- </tgroup>
- </informaltable>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1289945" xreflabel=""/>24.2.1 Tuning Linux Storage Devices</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303368" xreflabel=""/>To get large I/O transfers (1 MB) to disk, it may be necessary to tune several kernel parameters as specified:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1303369" xreflabel=""/>/sys/block/sdN/queue/max_sectors_kb = 4096
-<anchor xml:id="dbdoclet.50438212_pgfId-1303370" xreflabel=""/>/sys/block/sdN/queue/max_phys_segments = 256
-<anchor xml:id="dbdoclet.50438212_pgfId-1303371" xreflabel=""/>/proc/scsi/sg/allow_dio = 1
-<anchor xml:id="dbdoclet.50438212_pgfId-1303372" xreflabel=""/>/sys/module/ib_srp/parameters/srp_sg_tablesize = 255
-<anchor xml:id="dbdoclet.50438212_pgfId-1304119" xreflabel=""/>/sys/block/sdN/queue/scheduler</screen>
- </section>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1303366" xreflabel=""/>24.2.2 Running sgpdd_survey</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304151" xreflabel=""/>The sgpdd_survey script must be customized for the particular device being tested and for the location where the script saves its working and result files (by specifying the ${rslt} variable). Customization variables are described at the beginning of the script.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301030" xreflabel=""/>When the sgpdd_survey script runs, it creates a number of working files and a pair of result files. The names of all the files created start with the prefixdefined in the variable ${rslt}. (The default value is /tmp.) The files include:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301035" xreflabel=""/> File containing standard output data (same as stdout)</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1301031" xreflabel=""/>${rslt}_<emphasis><date/time></emphasis>.summary<emphasis/></screen>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301053" xreflabel=""/> Temporary (tmp) files</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1301032" xreflabel=""/>${rslt}_<emphasis><date/time></emphasis>_*
+ <listitem>
+ <para>Temporary (tmp) files</para>
+ <screen><replaceable>rslt_date_time</replaceable>_*
</screen>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301063" xreflabel=""/> Collected tmp files for post-mortem</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1289951" xreflabel=""/>${rslt}_<emphasis><date/time></emphasis>.detail
+ </listitem>
+ <listitem>
+ <para>Collected tmp files for post-mortem</para>
+ <screen><replaceable>rslt_date_time</replaceable>.detail
</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289952" xreflabel=""/>The stdout and the .summary file will contain lines like this:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1289953" xreflabel=""/>total_size 8388608K rsz 1024 thr 1 crg 1 180.45 MB/s 1 x 180.50 \=/ 180.50 \
-MB/s
+ </listitem>
+ </itemizedlist>
+ <para>The <literal>stdout</literal> and the <literal>.summary</literal> file will contain lines like this:</para>
+ <screen>total_size 8388608K rsz 1024 thr 1 crg 1 180.45 MB/s 1 x 180.50 \
+ = 180.50 MB/s
</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289954" xreflabel=""/>Each line corresponds to a run of the test. Each test run will have a different number of threads, record size, or number of regions.</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303307" xreflabel=""/>total_size - Size of file being tested in KBs (8 GB in above example).</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303308" xreflabel=""/>rsz - Record size in KBs (1 MB in above example).</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303311" xreflabel=""/>thr - Number of threads generating I/O (1 thread in above example).</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303312" xreflabel=""/> crg - Current regions, the number of disjount areas on the disk to which I/O is being sent (1 region in above example, indicating that no seeking is done).</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303313" xreflabel=""/>MB/s - Aggregate bandwidth measured by dividing the total amount of data by the elapsed time (180.45 MB/s in the above example).</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303314" xreflabel=""/>MB/s - The remaining numbers show the number of regions X performance of the slowest disk as a sanity check on the aggregate bandwidth.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289955" xreflabel=""/>If there are so many threads that the sgp_dd script is unlikely to be able to allocate I/O buffers, then ENOMEM is printed in place of the aggregate bandwidth result.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289956" xreflabel=""/>If one or more sgp_dd instances do not successfully report a bandwidth number, then FAILED is printed in place of the aggregate bandwidth result.</para>
- </section>
+ <para>Each line corresponds to a run of the test. Each test run will have a different number of threads, record size, or number of regions.</para>
+ <itemizedlist>
+ <listitem>
+ <para><literal>total_size</literal> - Size of file being tested in KBs (8 GB in above example).</para>
+ </listitem>
+ <listitem>
+ <para><literal>rsz</literal> - Record size in KBs (1 MB in above example).</para>
+ </listitem>
+ <listitem>
+ <para><literal>thr</literal> - Number of threads generating I/O (1 thread in above example).</para>
+ </listitem>
+ <listitem>
+ <para><literal>crg</literal> - Current regions, the number of disjoint areas on the disk to which I/O is being sent (1 region in above example, indicating that no seeking is done).</para>
+ </listitem>
+ <listitem>
+ <para><literal>MB/s</literal> - Aggregate bandwidth measured by dividing the total amount of data by the elapsed time (180.45 MB/s in the above example).</para>
+ </listitem>
+ <listitem>
+ <para><literal>MB/s</literal> - The remaining numbers show the number of regions X performance of the slowest disk as a sanity check on the aggregate bandwidth.</para>
+ </listitem>
+ </itemizedlist>
+ <para>If there are so many threads that the <literal>sgp_dd</literal> script is unlikely to be able to allocate I/O buffers, then <literal>ENOMEM</literal> is printed in place of the aggregate bandwidth result.</para>
+ <para>If one or more <literal>sgp_dd</literal> instances do not successfully report a bandwidth number, then <literal>FAILED</literal> is printed in place of the aggregate bandwidth result.</para>
+ </section>
+ </section>
+ <section xml:id="dbdoclet.50438212_26516">
+ <title><indexterm>
+ <primary>benchmarking</primary>
+ <secondary>OST performance</secondary>
+ </indexterm>Testing OST Performance (<literal>obdfilter-survey</literal>)</title>
+ <para>The <literal>obdfilter-survey</literal> script generates sequential I/O from varying
+ numbers of threads and objects (files) to simulate the I/O patterns of a Lustre client.</para>
+ <para>The <literal>obdfilter-survey</literal> script can be run directly on the OSS node to
+ measure the OST storage performance without any intervening network, or it can be run remotely
+ on a Lustre client to measure the OST performance including network overhead.</para>
+ <para>The <literal>obdfilter-survey</literal> is used to characterize the performance of the
+ following:</para>
+ <itemizedlist>
+ <listitem>
+ <para><emphasis role="bold">Local file system</emphasis> - In this mode, the
+ <literal>obdfilter-survey</literal> script exercises one or more instances of the
+ obdfilter directly. The script may run on one or more OSS nodes, for example, when the
+ OSSs are all attached to the same multi-ported disk subsystem.</para>
+ <para>Run the script using the <literal>case=disk</literal> parameter to run the test against all the local OSTs. The script automatically detects all local OSTs and includes them in the survey.</para>
+ <para>To run the test against only specific OSTs, run the script using the <literal>target=parameter</literal> to list the OSTs to be tested explicitly. If some OSTs are on remote nodes, specify their hostnames in addition to the OST name (for example, <literal>oss2:lustre-OST0004</literal>).</para>
+ <para>All <literal>obdfilter</literal> instances are driven directly. The script automatically loads the <literal>obdecho</literal> module (if required) and creates one instance of <literal>echo_client</literal> for each <literal>obdfilter</literal> instance in order to generate I/O requests directly to the OST.</para>
+ <para>For more details, see <xref linkend="dbdoclet.50438212_59319"/>.</para>
+ </listitem>
+ <listitem>
+ <para><emphasis role="bold">Network</emphasis> - In this mode, the Lustre client generates I/O requests over the network but these requests are not sent to the OST file system. The OSS node runs the obdecho server to receive the requests but discards them before they are sent to the disk.</para>
+ <para>Pass the parameters <literal>case=network</literal> and <literal>target=<replaceable>hostname|IP_of_server</replaceable></literal> to the script. For each network case, the script does the required setup.</para>
+ <para>For more details, see <xref linkend="dbdoclet.50438212_36037"/></para>
+ </listitem>
+ <listitem>
+ <para><emphasis role="bold">Remote file system over the network</emphasis> - In this mode
+ the <literal>obdfilter-survey</literal> script generates I/O from a Lustre client to a
+ remote OSS to write the data to the file system.</para>
+ <para>To run the test against all the local OSCs, pass the parameter <literal>case=netdisk</literal> to the script. Alternately you can pass the target= parameter with one or more OSC devices (e.g., <literal>lustre-OST0000-osc-ffff88007754bc00</literal>) against which the tests are to be run.</para>
+ <para>For more details, see <xref linkend="dbdoclet.50438212_62662"/>.</para>
+ </listitem>
+ </itemizedlist>
+ <caution>
+ <para>The <literal>obdfilter-survey</literal> script is potentially destructive and there is a
+ small risk data may be lost. To reduce this risk, <literal>obdfilter-survey</literal> should
+ not be run on devices that contain data that needs to be preserved. Thus, the best time to
+ run <literal>obdfilter-survey</literal> is before the Lustre file system is put into
+ production. The reason <literal>obdfilter-survey</literal> may be safe to run on a
+ production file system is because it creates objects with object sequence 2. Normal file
+ system objects are typically created with object sequence 0.</para>
+ </caution>
+ <note>
+ <para>If the <literal>obdfilter-survey</literal> test is terminated before it completes, some
+ small amount of space is leaked. you can either ignore it or reformat the file
+ system.</para>
+ </note>
+ <note>
+ <para>The <literal>obdfilter-survey</literal> script is <emphasis>NOT</emphasis> scalable
+ beyond tens of OSTs since it is only intended to measure the I/O performance of individual
+ storage subsystems, not the scalability of the entire system.</para>
+ </note>
+ <note>
+ <para>The <literal>obdfilter-survey</literal> script must be customized, depending on the
+ components under test and where the script's working files should be kept.
+ Customization variables are described at the beginning of the
+ <literal>obdfilter-survey</literal> script. In particular, pay attention to the listed
+ maximum values listed for each parameter in the script.</para>
+ </note>
+ <section xml:id="dbdoclet.50438212_59319">
+ <title><indexterm><primary>benchmarking</primary><secondary>local disk</secondary></indexterm>Testing Local Disk Performance</title>
+ <para>The <literal>obdfilter-survey</literal> script can be run automatically or manually
+ against a local disk. This script profiles the overall throughput of storage hardware,
+ including the file system and RAID layers managing the storage, by sending workloads to the
+ OSTs that vary in thread count, object count, and I/O size.</para>
+ <para>When the <literal>obdfilter-survey</literal> script is run, it provides information
+ about the performance abilities of the storage hardware and shows the saturation
+ points.</para>
+ <para>The <literal>plot-obdfilter</literal> script generates from the output of the
+ <literal>obdfilter-survey</literal> a CSV file and parameters for importing into a
+ spreadsheet or gnuplot to visualize the data.</para>
+ <para>To run the <literal>obdfilter-survey</literal> script, create a standard Lustre file
+ system configuration; no special setup is needed.</para>
+ <para><emphasis role="bold">To perform an automatic run:</emphasis></para>
+ <orderedlist>
+ <listitem>
+ <para>Start the Lustre OSTs.</para>
+ <para>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
+ </listitem>
+ <listitem>
+ <para>Verify that the obdecho module is loaded. Run:</para>
+ <screen>modprobe obdecho</screen>
+ </listitem>
+ <listitem>
+ <para>Run the <literal>obdfilter-survey</literal> script with the parameter
+ <literal>case=disk</literal>.</para>
+ <para>For example, to run a local test with up to two objects (nobjhi), up to two threads (thrhi), and 1024 MB transfer size (size):</para>
+ <screen>$ nobjhi=2 thrhi=2 size=1024 case=disk sh obdfilter-survey</screen>
+ </listitem>
+ </orderedlist>
+ <para><emphasis role="italic">To perform a manual run:</emphasis></para>
+ <orderedlist>
+ <listitem>
+ <para>Start the Lustre OSTs.</para>
+ <para>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
+ </listitem>
+ <listitem>
+ <para>Verify that the <literal>obdecho</literal> module is loaded. Run:</para>
+ <screen>modprobe obdecho</screen>
+ </listitem>
+ <listitem>
+ <para>Determine the OST names.</para>
+ <para>On the OSS nodes to be tested, run the <literal>lctl dl</literal> command. The OST device names are listed in the fourth column of the output. For example:</para>
+ <screen>$ lctl dl |grep obdfilter
+0 UP obdfilter lustre-OST0001 lustre-OST0001_UUID 1159
+2 UP obdfilter lustre-OST0002 lustre-OST0002_UUID 1159
+...</screen>
+ </listitem>
+ <listitem>
+ <para>List all OSTs you want to test.</para>
+ <para>Use the <literal>target=parameter</literal> to list the OSTs separated by spaces. List the individual OSTs by name using the format
+ <literal><replaceable>fsname</replaceable>-<replaceable>OSTnumber</replaceable></literal>
+ (for example, <literal>lustre-OST0001</literal>). You do not have to specify an MDS or LOV.</para>
+ </listitem>
+ <listitem>
+ <para>Run the <literal>obdfilter-survey</literal> script with the
+ <literal>target=parameter</literal>.</para>
+ <para>For example, to run a local test with up to two objects (<literal>nobjhi</literal>), up to two threads (<literal>thrhi</literal>), and 1024 Mb (size) transfer size:</para>
+ <screen>$ nobjhi=2 thrhi=2 size=1024 targets="lustre-OST0001 \
+ lustre-OST0002" sh obdfilter-survey</screen>
+ </listitem>
+ </orderedlist>
+ </section>
+ <section xml:id="dbdoclet.50438212_36037">
+ <title><indexterm><primary>benchmarking</primary><secondary>network</secondary></indexterm>Testing Network Performance</title>
+ <para>The <literal>obdfilter-survey</literal> script can only be run automatically against a
+ network; no manual test is provided.</para>
+ <para>To run the network test, a specific Lustre file system setup is needed. Make sure that
+ these configuration requirements have been met.</para>
+ <para><emphasis role="bold">To perform an automatic run:</emphasis></para>
+ <orderedlist>
+ <listitem>
+ <para>Start the Lustre OSTs.</para>
+ <para>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
+ </listitem>
+ <listitem>
+ <para>Verify that the <literal>obdecho</literal> module is loaded. Run:</para>
+ <screen>modprobe obdecho</screen>
+ </listitem>
+ <listitem>
+ <para>Start <literal>lctl</literal> and check the device list, which must be empty. Run:</para>
+ <screen>lctl dl</screen>
+ </listitem>
+ <listitem>
+ <para>Run the <literal>obdfilter-survey</literal> script with the parameters
+ <literal>case=network</literal> and
+ <literal>targets=<replaceable>hostname|ip_of_server</replaceable></literal>. For
+ example:</para>
+ <screen>$ nobjhi=2 thrhi=2 size=1024 targets="oss0 oss1" \
+ case=network sh odbfilter-survey</screen>
+ </listitem>
+ <listitem>
+ <para>On the server side, view the statistics at:</para>
+ <screen>/proc/fs/lustre/obdecho/<replaceable>echo_srv</replaceable>/stats</screen>
+ <para>where <literal><replaceable>echo_srv</replaceable></literal>
+ is the <literal>obdecho</literal> server created by the script.</para>
+ </listitem>
+ </orderedlist>
+ </section>
+ <section xml:id="dbdoclet.50438212_62662">
+ <title><indexterm><primary>benchmarking</primary><secondary>remote disk</secondary></indexterm>Testing Remote Disk Performance</title>
+ <para>The <literal>obdfilter-survey</literal> script can be run automatically or manually
+ against a network disk. To run the network disk test, start with a standard Lustre
+ configuration. No special setup is needed.</para>
+ <para><emphasis role="bold">To perform an automatic run:</emphasis></para>
+ <orderedlist>
+ <listitem>
+ <para>Start the Lustre OSTs.</para>
+ <para>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
+ </listitem>
+ <listitem>
+ <para>Verify that the <literal>obdecho</literal> module is loaded. Run:</para>
+ <screen>modprobe obdecho</screen>
+ </listitem>
+ <listitem>
+ <para>Run the <literal>obdfilter-survey</literal> script with the parameter
+ <literal>case=netdisk</literal>. For example:</para>
+ <screen>$ nobjhi=2 thrhi=2 size=1024 case=netdisk sh obdfilter-survey
+</screen>
+ </listitem>
+ </orderedlist>
+ <para><emphasis role="bold">To perform a manual run:</emphasis></para>
+ <orderedlist>
+ <listitem>
+ <para>Start the Lustre OSTs.</para>
+ <para>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
+ </listitem>
+ <listitem>
+ <para>Verify that the <literal>obdecho</literal> module is loaded. Run:</para>
+ <para>modprobe obdecho</para>
+ </listitem>
+ <listitem>
+ <para>Determine the OSC names.</para>
+ <para>On the OSS nodes to be tested, run the <literal>lctl dl</literal> command. The OSC device names are listed in the fourth column of the output. For example:</para>
+ <screen>$ lctl dl |grep obdfilter
+3 UP osc lustre-OST0000-osc-ffff88007754bc00 \
+ 54b91eab-0ea9-1516-b571-5e6df349592e 5
+4 UP osc lustre-OST0001-osc-ffff88007754bc00 \
+ 54b91eab-0ea9-1516-b571-5e6df349592e 5
+...
+</screen>
+ </listitem>
+ <listitem>
+ <para>List all OSCs you want to test.</para>
+ <para>Use the <literal>target=parameter</literal> to list the OSCs separated by spaces. List the individual OSCs by name separated by spaces using the format <literal><replaceable>fsname</replaceable>-<replaceable>OST_name</replaceable>-osc-<replaceable>instance</replaceable></literal> (for example, <literal>lustre-OST0000-osc-ffff88007754bc00</literal>). You <emphasis>do not have to specify an MDS or LOV.</emphasis></para>
+ </listitem>
+ <listitem>
+ <para>Run the <literal>obdfilter-survey</literal> script with the
+ <literal>target=<replaceable>osc</replaceable></literal> and
+ <literal>case=netdisk</literal>.</para>
+ <para>An example of a local test run with up to two objects (<literal>nobjhi</literal>), up to two threads (<literal>thrhi</literal>), and 1024 Mb (size) transfer size is shown below:</para>
+ <screen>$ nobjhi=2 thrhi=2 size=1024 \
+ targets="lustre-OST0000-osc-ffff88007754bc00 \
+ lustre-OST0001-osc-ffff88007754bc00" sh obdfilter-survey
+</screen>
+ </listitem>
+ </orderedlist>
</section>
- <section remap="h2">
- <title>24.3 <anchor xml:id="dbdoclet.50438212_26516" xreflabel=""/><anchor xml:id="dbdoclet.50438212_40624" xreflabel=""/>Testing OST Performance (obdfilter_survey<anchor xml:id="dbdoclet.50438212_marker-1289957" xreflabel=""/>)</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301303" xreflabel=""/>The obdfilter_survey script generates sequential I/O from varying numbers of threads and objects (files) to simulate the I/O patterns of a Lustre client.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301200" xreflabel=""/>The obdfilter_survey script can be run directly on the OSS node to measure the OST storage performance without any intervening network, or it can be run remotely on a Lustre client to measure the OST performance including network overhead.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303417" xreflabel=""/>The obdfilter_survey is used to characterize the performance of the following:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289960" xreflabel=""/><emphasis role="bold">Local file system</emphasis> - In this mode, the obdfilter_survey script exercises one or more instances of the obdfilter directly. The script may run on one or more OSS nodes, for example, when the OSSs are all attached to the same multi-ported disk subsystem.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289962" xreflabel=""/>Run the script using the case=disk parameter to run the test against all the local OSTs. The script automatically detects all local OSTs and includes them in the survey.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303453" xreflabel=""/>To run the test against only specific OSTs, run the script using the target= parameter to list the OSTs to be tested explicitly. If some OSTs are on remote nodes, specify their hostnames in addition to the OST name (for example, oss2:lustre-OST0004).</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303461" xreflabel=""/>All obdfilter instances are driven directly. The script automatically loads the obdecho module (if required) and creates one instance of echo_client for each obdfilter instance in order to generate I/O requests directly to the OST.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304398" xreflabel=""/>For more details, see <link xl:href="BenchmarkingTests.html#50438212_59319">Testing Local Disk Performance</link>.</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289963" xreflabel=""/><emphasis role="bold">Network</emphasis> - In this mode, the Lustre client generates I/O requests over the network but these requests are not sent to the OST file system. The OSS node runs the obdecho server to receive the requests but discards them before they are sent to the disk.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303500" xreflabel=""/>Pass the parameters case=network and target=<emphasis><hostname</emphasis>|<emphasis>IP_of_server></emphasis> to the script. For each network case, the script does the required setup.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304403" xreflabel=""/>For more details, see <link xl:href="BenchmarkingTests.html#50438212_36037">Testing Network Performance</link></para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304334" xreflabel=""/><emphasis role="bold">Remote file system over the network</emphasis> - In this mode the obdfilter_survey script generates I/O from a Lustre client to a remote OSS to write the data to the file system.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304336" xreflabel=""/>To run the test against all the local OSCs, pass the parameter case=netdisk to the script. Alternately you can pass the target= parameter with one or more OSC devices (e.g., lustre-OST0000-osc-ffff88007754bc00) against which the tests are to be run.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304408" xreflabel=""/>For more details, see <link xl:href="BenchmarkingTests.html#50438212_62662">Testing Remote Disk Performance</link>.</para>
- <informaltable frame="none">
+ <section remap="h3">
+ <title>Output Files</title>
+ <para>When the <literal>obdfilter-survey</literal> script runs, it creates a number of working
+ files and a pair of result files. All files start with the prefix defined in the variable
+ <literal>${rslt}</literal>.</para>
+ <informaltable frame="all">
<tgroup cols="2">
- <colspec colname="c1" colwidth="5*"/>
- <colspec colname="c2" colwidth="95*"/>
-
-
- <tbody>
+ <colspec colname="c1" colwidth="50*"/>
+ <colspec colname="c2" colwidth="50*"/>
+ <thead>
<row>
- <entry><para><inlinemediaobject><imageobject role="html">
- <imagedata fileref="./shared/caution.gif" scalefit="1"/>
- </imageobject>
-<imageobject role="fo">
- <imagedata contentdepth="100%" contentwidth="" depth="" fileref="./shared/caution.gif" scalefit="1" width="100%"/>
- </imageobject>
-</inlinemediaobject></para></entry>
- <entry><para><emphasis role="bold">Caution -</emphasis><anchor xml:id="dbdoclet.50438212_pgfId-1304418" xreflabel=""/>The obdfilter_survey script is destructive and should not be run on devices that containing existing data that needs to be preserved. Thus, tests using obdfilter_survey should be run before the Lustre file system is placed in production.</para></entry>
+ <entry>
+ <para><emphasis role="bold">File</emphasis></para>
+ </entry>
+ <entry>
+ <para><emphasis role="bold">Description</emphasis></para>
+ </entry>
</row>
- </tbody>
- </tgroup>
- </informaltable>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
+ </thead>
<tbody>
<row>
- <entry><para><emphasis role="bold">Note -</emphasis> If the obdfilter_survey test is terminated before it completes, some small amount of space is leaked. You can either ignore it or reformat the file system.</para></entry>
+ <entry>
+ <para> <literal>${rslt}.summary</literal></para>
+ </entry>
+ <entry>
+ <para> Same as stdout</para>
+ </entry>
</row>
- </tbody>
- </tgroup>
- </informaltable>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
- <tbody>
<row>
- <entry><para><emphasis role="bold">Note -</emphasis> The obdfilter_survey script is <emphasis>NOT</emphasis> scalable beyond tens of OSTs since it is only intended to measure the I/O performance of individual storage subsystems, not the scalability of the entire system.</para></entry>
+ <entry>
+ <para> <literal>${rslt}.script_*</literal></para>
+ </entry>
+ <entry>
+ <para> Per-host test script files</para>
+ </entry>
</row>
- </tbody>
- </tgroup>
- </informaltable>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
- <tbody>
<row>
- <entry><para><emphasis role="bold">Note -</emphasis><anchor xml:id="dbdoclet.50438212_pgfId-1297549" xreflabel=""/>The obdfilter_survey script must be customized, depending on the components under test and where the script’s working files should be kept. Customization variables are described at the beginning of the obdfilter_survey script. In particular, pay attention to the listed maximum values listed for each parameter in the script.</para></entry>
+ <entry>
+ <para> <literal>${rslt}.detail_tmp*</literal></para>
+ </entry>
+ <entry>
+ <para> Per-OST result files</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para> <literal>${rslt}.detail</literal></para>
+ </entry>
+ <entry>
+ <para> Collected result files for post-mortem</para>
+ </entry>
</row>
</tbody>
</tgroup>
</informaltable>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1289969" xreflabel=""/>24.3.1 <anchor xml:id="dbdoclet.50438212_59319" xreflabel=""/>Testing Local Disk Performance</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1298366" xreflabel=""/>The obdfilter_survey script can be run automatically or manually against a local disk. This script profiles the overall throughput of storage hardware, including the file system and RAID layers managing the storage, by sending workloads to the OSTs that vary in thread count, object count, and I/O size.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1298465" xreflabel=""/>When the obdfilter_survey script is run, it provides information about the performance abilities of the storage hardware and shows the saturation points.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303569" xreflabel=""/>The plot-obdfilter script generates from the output of the obdfilter_survey a CSV file and parameters for importing into a spreadsheet or gnuplot to visualize the data.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297883" xreflabel=""/>To run the obdfilter_survey script, create a standard Lustre configuration; no special setup is needed.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297701" xreflabel=""/><emphasis role="bold">To perform an automatic run:</emphasis></para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297574" xreflabel=""/> 1. Start the Lustre OSTs.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304489" xreflabel=""/>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297587" xreflabel=""/> 2. Verify that the obdecho module is loaded. Run:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1304502" xreflabel=""/>modprobe obdecho</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297591" xreflabel=""/> 3. Run the obdfilter_survey script with the parameter case=disk.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304517" xreflabel=""/>For example, to run a local test with up to two objects (nobjhi), up to two threads (thrhi), and 1024 MB transfer size (size):</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1297595" xreflabel=""/>$ nobjhi=2 thrhi=2 size=1024 case=disk sh obdfilter-survey
-<anchor xml:id="dbdoclet.50438212_pgfId-1304711" xreflabel=""/>
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299344" xreflabel=""/><emphasis role="bold">To perform a manual run:</emphasis></para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303657" xreflabel=""/> 1. Start the Lustre OSTs.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304535" xreflabel=""/>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305056" xreflabel=""/> 2. Verify that the obdecho module is loaded. Run:</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305057" xreflabel=""/>modprobe obdecho</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303655" xreflabel=""/> 3. Determine the OST names.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304551" xreflabel=""/>On the OSS nodes to be tested, run the lctldl command. The OST device names are listed in the fourth column of the output. For example:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1303717" xreflabel=""/>$ lctl dl |grep obdfilter
-<anchor xml:id="dbdoclet.50438212_pgfId-1303733" xreflabel=""/>0 UP obdfilter lustre-OST0001 lustre-OST0001_UUID 1159
-<anchor xml:id="dbdoclet.50438212_pgfId-1303751" xreflabel=""/>2 UP obdfilter lustre-OST0002 lustre-OST0002_UUID 1159
-<anchor xml:id="dbdoclet.50438212_pgfId-1303742" xreflabel=""/>...
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303687" xreflabel=""/> 4. List all OSTs you want to test.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304568" xreflabel=""/>Use the target= parameter to list the OSTs separated by spaces. List the individual OSTs by name using the format <emphasis><fsname>-<OSTnumber></emphasis> (for example, lustre-OST0001). You do not have to specify an MDS or LOV.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1303805" xreflabel=""/> 5. Run the obdfilter_survey script with the target= parameter.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304650" xreflabel=""/>For example, to run a local test with up to two objects (nobjhi), up to two threads (thrhi), and 1024 Mb (size) transfer size:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1303806" xreflabel=""/>$ nobjhi=2 thrhi=2 size=1024 targets=â€lustre-OST0001 \
-<anchor xml:id="dbdoclet.50438212_pgfId-1304757" xreflabel=""/>lustre-OST0002†sh obdfilter-survey
-</screen>
- </section>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1289982" xreflabel=""/>24.3.2 <anchor xml:id="dbdoclet.50438212_36037" xreflabel=""/>Testing Network Performance</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1289983" xreflabel=""/>The obdfilter_survey script can only be run automatically against a network; no manual test is provided.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297847" xreflabel=""/>To run the network test, a specific Lustre setup is needed. Make sure that these configuration requirements have been met.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305077" xreflabel=""/><emphasis role="bold">To perform an automatic run:</emphasis></para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305363" xreflabel=""/> 1. Start the Lustre OSTs.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305364" xreflabel=""/>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305365" xreflabel=""/> 2. Verify that the obdecho module is loaded. Run:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1305366" xreflabel=""/>modprobe obdecho</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305389" xreflabel=""/> 3. Start lctl and check the device list, which must be empty. Run:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1305940" xreflabel=""/>lctl dl
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305639" xreflabel=""/> 4. Run the obdfilter_survey script with the parameters case=network and <emphasis role="bold">targets=</emphasis><emphasis><hostname</emphasis>|<emphasis>ip_of_server></emphasis>. For example:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1304777" xreflabel=""/>$ nobjhi=2 thrhi=2 size=1024 targets=â€oss1 oss2†case=network sh obdfilte\
-r-survey
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297769" xreflabel=""/> 5. On the server side, view the statistics at:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1297820" xreflabel=""/>/proc/fs/lustre/obdecho/<emphasis><echo_srv></emphasis>/stats
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297824" xreflabel=""/>where <emphasis><echo_srv></emphasis> is the obdecho server created by the script.</para>
- </section>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1297766" xreflabel=""/>24.3.3 <anchor xml:id="dbdoclet.50438212_62662" xreflabel=""/>Testing Remote Disk Performance</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297832" xreflabel=""/>The obdfilter_survey script can be run automatically or manually against a network disk. To run the network disk test, start with a standard Lustre configuration. No special setup is needed.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297904" xreflabel=""/><emphasis role="bold">To perform an automatic run:</emphasis></para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305421" xreflabel=""/> 1. Start the Lustre OSTs.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305422" xreflabel=""/>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305423" xreflabel=""/> 2. Verify that the obdecho module is loaded. Run:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1305424" xreflabel=""/>modprobe obdecho</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297925" xreflabel=""/> 3. Run the obdfilter_survey script with the parameter case=netdisk. For example:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1297926" xreflabel=""/>$ nobjhi=2 thrhi=2 size=1024 case=netdisk sh obdfilter-survey
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1297909" xreflabel=""/><emphasis role="bold">To perform a manual run:</emphasis></para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305433" xreflabel=""/> 1. Start the Lustre OSTs.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305434" xreflabel=""/>The Lustre OSTs should be mounted on the OSS node(s) to be tested. The Lustre client is not required to be mounted at this time.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305435" xreflabel=""/> 2. Verify that the obdecho module is loaded. Run:</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305436" xreflabel=""/>modprobe obdecho</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305437" xreflabel=""/> 3. Determine the OSC names.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305438" xreflabel=""/>On the OSS nodes to be tested, run the lctldl command. The OSC device names are listed in the fourth column of the output. For example:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1305483" xreflabel=""/>$ lctl dl |grep obdfilter
-<anchor xml:id="dbdoclet.50438212_pgfId-1305461" xreflabel=""/>3 UP osc lustre-OST0000-osc-ffff88007754bc00 54b91eab-0ea9-1516-b571-5e6df3\
-49592e 5
-<anchor xml:id="dbdoclet.50438212_pgfId-1305462" xreflabel=""/>4 UP osc lustre-OST0001-osc-ffff88007754bc00 54b91eab-0ea9-1516-b571-5e6df3\
-49592e 5
-<anchor xml:id="dbdoclet.50438212_pgfId-1305515" xreflabel=""/>...
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305443" xreflabel=""/> 4. List all OSCs you want to test.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305444" xreflabel=""/>Use the target= parameter to list the OSCs separated by spaces. List the individual OSCs by name seperated by spaces using the format <emphasis><fsname>-<OST_name></emphasis>-osc-<emphasis><OSC_number></emphasis> (for example, lustre-OST0000-osc-ffff88007754bc00). You <emphasis role="bold">do not have to specify an MDS or LOV.</emphasis></para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305445" xreflabel=""/> 5. Run the <emphasis role="bold">o</emphasis>bdfilter_survey script with the target= parameter and case=netdisk.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305446" xreflabel=""/>An example of a local test run with up to two objects (nobjhi), up to two threads (thrhi), and 1024 Mb (size) transfer size is shown below:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1305576" xreflabel=""/>$ nobjhi=2 thrhi=2 size=1024 \
-<anchor xml:id="dbdoclet.50438212_pgfId-1305577" xreflabel=""/>targets="lustre-OST0000-osc-ffff88007754bc00 \
-<anchor xml:id="dbdoclet.50438212_pgfId-1305578" xreflabel=""/>lustre-OST0001-osc-ffff88007754bc00" \
-<anchor xml:id="dbdoclet.50438212_pgfId-1305579" xreflabel=""/>sh obdfilter-survey
+ <para>The <literal>obdfilter-survey</literal> script iterates over the given number of threads
+ and objects performing the specified tests and checks that all test processes have completed
+ successfully.</para>
+ <note>
+ <para>The <literal>obdfilter-survey</literal> script may not clean up properly if it is
+ aborted or if it encounters an unrecoverable error. In this case, a manual cleanup may be
+ required, possibly including killing any running instances of <literal>lctl</literal>
+ (local or remote), removing <literal>echo_client</literal> instances created by the script
+ and unloading <literal>obdecho</literal>.</para>
+ </note>
+ <section remap="h4">
+ <title>Script Output</title>
+ <para>The <literal>.summary</literal> file and <literal>stdout</literal> of the
+ <literal>obdfilter-survey</literal> script contain lines like:</para>
+ <screen>ost 8 sz 67108864K rsz 1024 obj 8 thr 8 write 613.54 [ 64.00, 82.00]
</screen>
- </section>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1290021" xreflabel=""/>24.3.4 Output Files</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1298013" xreflabel=""/>When the obdfilter_survey script runs, it creates a number of working files and a pair of result files. All files start with the prefix defined in the variable ${rslt}.</para>
+ <para>Where:</para>
<informaltable frame="all">
<tgroup cols="2">
<colspec colname="c1" colwidth="50*"/>
<colspec colname="c2" colwidth="50*"/>
<thead>
<row>
- <entry><para><emphasis role="bold"><anchor xml:id="dbdoclet.50438212_pgfId-1303053" xreflabel=""/>File</emphasis></para></entry>
- <entry><para><emphasis role="bold"><anchor xml:id="dbdoclet.50438212_pgfId-1303055" xreflabel=""/>Description</emphasis></para></entry>
+ <entry>
+ <para><emphasis role="bold">Parameter and value</emphasis></para>
+ </entry>
+ <entry>
+ <para><emphasis role="bold">Description</emphasis></para>
+ </entry>
</row>
</thead>
<tbody>
<row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303057" xreflabel=""/>${rslt}.summary</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303059" xreflabel=""/>Same as stdout</para></entry>
+ <entry>
+ <para> ost 8</para>
+ </entry>
+ <entry>
+ <para> Total number of OSTs being tested.</para>
+ </entry>
</row>
<row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303061" xreflabel=""/>${rslt}.script_*</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303063" xreflabel=""/>Per-host test script files</para></entry>
+ <entry>
+ <para> sz 67108864K</para>
+ </entry>
+ <entry>
+ <para> Total amount of data read or written (in KB).</para>
+ </entry>
</row>
<row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303065" xreflabel=""/>${rslt}.detail_tmp*</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303067" xreflabel=""/>Per-OST result files</para></entry>
+ <entry>
+ <para> rsz 1024</para>
+ </entry>
+ <entry>
+ <para> Record size (size of each echo_client I/O, in KB).</para>
+ </entry>
</row>
<row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303069" xreflabel=""/>${rslt}.detail</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1303071" xreflabel=""/>Collected result files for post-mortem</para></entry>
+ <entry>
+ <para> obj 8</para>
+ </entry>
+ <entry>
+ <para> Total number of objects over all OSTs.</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para> thr 8</para>
+ </entry>
+ <entry>
+ <para> Total number of threads over all OSTs and objects.</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para> write</para>
+ </entry>
+ <entry>
+ <para> Test name. If more tests have been specified, they all appear on the same line.</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para> 613.54</para>
+ </entry>
+ <entry>
+ <para> Aggregate bandwidth over all OSTs (measured by dividing the total number of MB by the elapsed time).</para>
+ </entry>
</row>
- </tbody>
- </tgroup>
- </informaltable>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1290022" xreflabel=""/>The obdfilter_survey script iterates over the given number of threads and objects performing the specified tests and checks that all test processes have completed successfully.</para>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
- <tbody>
<row>
- <entry><para><emphasis role="bold">Note -</emphasis><anchor xml:id="dbdoclet.50438212_pgfId-1298109" xreflabel=""/>The obdfilter_survey script may not clean up properly if it is aborted or if it encounters an unrecoverable error. In this case, a manual cleanup may be required, possibly including killing any running instances of lctl (local or remote), removing echo_client instances created by the script and unloading obdecho.</para></entry>
+ <entry>
+ <para> [64, 82.00]</para>
+ </entry>
+ <entry>
+ <para> Minimum and maximum instantaneous bandwidths on an individual OST.</para>
+ </entry>
</row>
</tbody>
</tgroup>
</informaltable>
- <section remap="h4">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1298104" xreflabel=""/>24.3.4.1 Script Output</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1298134" xreflabel=""/>The .summary file and stdout of the obdfilter_survey script contain lines like:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1294098" xreflabel=""/>ost 8 sz 67108864K rsz 1024 obj 8 thr 8 write 613.54 [ 64.00, 82.00]
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1294150" xreflabel=""/>Where:</para>
- <informaltable frame="all">
- <tgroup cols="2">
- <colspec colname="c1" colwidth="50*"/>
- <colspec colname="c2" colwidth="50*"/>
- <thead>
- <row>
- <entry><para><emphasis role="bold"><anchor xml:id="dbdoclet.50438212_pgfId-1294115" xreflabel=""/>Parameter and value</emphasis></para></entry>
- <entry><para><emphasis role="bold"><anchor xml:id="dbdoclet.50438212_pgfId-1294117" xreflabel=""/>Description</emphasis></para></entry>
- </row>
- </thead>
- <tbody>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294119" xreflabel=""/>ost 8</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294121" xreflabel=""/>Total number of OSTs being tested.</para></entry>
- </row>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294123" xreflabel=""/>sz 67108864K</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294125" xreflabel=""/>Total amount of data read or written (in KB).</para></entry>
- </row>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294127" xreflabel=""/>rsz 1024</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294129" xreflabel=""/>Record size (size of each echo_client I/O, in KB).</para></entry>
- </row>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294131" xreflabel=""/>obj 8</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294133" xreflabel=""/>Total number of objects over all OSTs.</para></entry>
- </row>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294135" xreflabel=""/>thr 8</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294137" xreflabel=""/>Total number of threads over all OSTs and objects.</para></entry>
- </row>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294139" xreflabel=""/>write</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294141" xreflabel=""/>Test name. If more tests have been specified, they all appear on the same line.</para></entry>
- </row>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294143" xreflabel=""/>613.54</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294145" xreflabel=""/>Aggregate bandwidth over all OSTs (measured by dividing the total number of MB by the elapsed time).</para></entry>
- </row>
- <row>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294147" xreflabel=""/>[64, 82.00]</para></entry>
- <entry><para> <anchor xml:id="dbdoclet.50438212_pgfId-1294149" xreflabel=""/>Minimum and maximum instantaneous bandwidths on an individual OST.</para></entry>
- </row>
- </tbody>
- </tgroup>
- </informaltable>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
- <tbody>
- <row>
- <entry><para><emphasis role="bold">Note -</emphasis><anchor xml:id="dbdoclet.50438212_pgfId-1290062" xreflabel=""/>Although the numbers of threads and objects are specified per-OST in the customization section of the script, the reported results are aggregated over all OSTs.</para></entry>
- </row>
- </tbody>
- </tgroup>
- </informaltable>
- </section>
- <section remap="h4">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1290063" xreflabel=""/>24.3.4.2 Visualizing Results</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1290064" xreflabel=""/>It is useful to import the obdfilter_survey script summary data (it is fixed width) into Excel (or any graphing package) and graph the bandwidth versus the number of threads for varying numbers of concurrent regions. This shows how the OSS performs for a given number of concurrently-accessed objects (files) with varying numbers of I/Os in flight.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305127" xreflabel=""/>It is also useful to monitor and record average disk I/O sizes during each test using the “disk io size†histogram in the file /proc/fs/lustre/obdfilter/ (see <link xl:href="LustreProc.html#50438271_55057">Watching the OST Block I/O Stream</link> for details). These numbers help identify problems in the system when full-sized I/Os are not submitted to the underlying disk. This may be caused by problems in the device driver or Linux block layer.</para>
- <para> */brw_stats</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1290066" xreflabel=""/>The plot-obdfilter script included in the I/O toolkit is an example of processing output files to a .csv format and plotting a graph using gnuplot.</para>
- </section>
+ <note>
+ <para>Although the numbers of threads and objects are specified per-OST in the customization section of the script, the reported results are aggregated over all OSTs.</para>
+ </note>
+ </section>
+ <section remap="h4">
+ <title>Visualizing Results</title>
+ <para>It is useful to import the <literal>obdfilter-survey</literal> script summary data (it
+ is fixed width) into Excel (or any graphing package) and graph the bandwidth versus the
+ number of threads for varying numbers of concurrent regions. This shows how the OSS
+ performs for a given number of concurrently-accessed objects (files) with varying numbers
+ of I/Os in flight.</para>
+ <para>It is also useful to monitor and record average disk I/O sizes during each test using the 'disk io size' histogram in the file <literal>/proc/fs/lustre/obdfilter/</literal> (see <xref linkend="dbdoclet.50438271_55057"/> for details). These numbers help identify problems in the system when full-sized I/Os are not submitted to the underlying disk. This may be caused by problems in the device driver or Linux block layer.</para>
+ <screen> */brw_stats</screen>
+ <para>The <literal>plot-obdfilter</literal> script included in the I/O toolkit is an example of processing output files to a .csv format and plotting a graph using <literal>gnuplot</literal>.</para>
</section>
</section>
- <section remap="h2">
- <title>24.4 <anchor xml:id="dbdoclet.50438212_85136" xreflabel=""/>Testing OST I/O Performance (ost_<anchor xml:id="dbdoclet.50438212_marker-1290067" xreflabel=""/>survey)</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1290069" xreflabel=""/>The ost_survey tool is a shell script that uses lfs setstripe to perform I/O against a single OST. The script writes a file (currently using dd) to each OST in the Lustre file system, and compares read and write speeds. The ost_survey tool is used to detect anomalies between otherwise identical disk subsystems.</para>
- <informaltable frame="none">
- <tgroup cols="1">
- <colspec colname="c1" colwidth="100*"/>
+ </section>
+ <section xml:id="dbdoclet.50438212_85136">
+ <title><indexterm>
+ <primary>benchmarking</primary>
+ <secondary>OST I/O</secondary>
+ </indexterm>Testing OST I/O Performance (<literal>ost-survey</literal>)</title>
+ <para>The <literal>ost-survey</literal> tool is a shell script that uses <literal>lfs
+ setstripe</literal> to perform I/O against a single OST. The script writes a file (currently
+ using <literal>dd</literal>) to each OST in the Lustre file system, and compares read and
+ write speeds. The <literal>ost-survey</literal> tool is used to detect anomalies between
+ otherwise identical disk subsystems.</para>
+ <note>
+ <para>We have frequently discovered wide performance variations across all LUNs in a cluster.
+ This may be caused by faulty disks, RAID parity reconstruction during the test, or faulty
+ network hardware.</para>
+ </note>
+ <para>To run the <literal>ost-survey</literal> script, supply a file size (in KB) and the Lustre
+ file system mount point. For example, run:</para>
+ <screen>$ ./ost-survey.sh 10 /mnt/lustre
+</screen>
+ <para>Typical output is:</para>
+ <screen>
+Average read Speed: 6.73
+Average write Speed: 5.41
+read - Worst OST indx 0 5.84 MB/s
+write - Worst OST indx 0 3.77 MB/s
+read - Best OST indx 1 7.38 MB/s
+write - Best OST indx 1 6.31 MB/s
+3 OST devices found
+Ost index 0 Read speed 5.84 Write speed 3.77
+Ost index 0 Read time 0.17 Write time 0.27
+Ost index 1 Read speed 7.38 Write speed 6.31
+Ost index 1 Read time 0.14 Write time 0.16
+Ost index 2 Read speed 6.98 Write speed 6.16
+Ost index 2 Read time 0.14 Write time 0.16
+</screen>
+ </section>
+ <section xml:id="mds_survey_ref">
+ <title><indexterm><primary>benchmarking</primary><secondary>MDS
+performance</secondary></indexterm>Testing MDS Performance (<literal>mds-survey</literal>)</title>
+ <para><literal>mds-survey</literal> is available in Lustre software release 2.2 and beyond. The
+ <literal>mds-survey</literal> script tests the local metadata performance using the
+ echo_client to drive different layers of the MDS stack: mdd, mdt, osd (the Lustre software
+ only supports mdd stack). It can be used with the following classes of operations:</para>
+
+ <itemizedlist>
+ <listitem>
+ <para><literal>Open-create/mkdir/create</literal></para>
+ </listitem>
+ <listitem>
+ <para><literal>Lookup/getattr/setxattr</literal></para>
+ </listitem>
+ <listitem>
+ <para><literal>Delete/destroy</literal></para>
+ </listitem>
+ <listitem>
+ <para><literal>Unlink/rmdir</literal></para>
+ </listitem>
+ </itemizedlist>
+ <para>These operations will be run by a variable number of concurrent threads and will test with the number of directories specified by the user. The run can be executed such that all threads operate in a single directory (dir_count=1) or in private/unique directory (dir_count=x thrlo=x thrhi=x).</para>
+
+ <para>The mdd instance is driven directly. The script automatically loads the obdecho module if required and creates instance of echo_client.</para>
+
+ <para>This script can also create OST objects by providing stripe_count greater than zero.</para>
+
+ <para><emphasis role="bold">To perform a run:</emphasis></para>
+ <orderedlist>
+ <listitem>
+ <para>Start the Lustre MDT.</para>
+ <para>The Lustre MDT should be mounted on the MDS node to be tested.</para>
+ </listitem>
+ <listitem>
+ <para>Start the Lustre OSTs (optional, only required when test with OST objects)</para>
+ <para>The Lustre OSTs should be mounted on the OSS node(s).</para>
+ </listitem>
+ <listitem>
+ <para>Run the <literal>mds-survey</literal> script as explain below</para>
+ <para>The script must be customized according to the components under test and where it should keep its working files. Customization variables are described as followed:</para>
+ <itemizedlist>
+ <listitem>
+ <para><literal>thrlo</literal> - threads to start testing. skipped if less than
+ <literal>dir_count</literal></para>
+ </listitem>
+ <listitem>
+ <para><literal>thrhi</literal> - maximum number of threads to test</para>
+ </listitem>
+ <listitem>
+ <para><literal>targets</literal> - MDT instance</para>
+ </listitem>
+ <listitem>
+ <para><literal>file_count</literal> - number of files per thread to test</para>
+ </listitem>
+ <listitem>
+ <para><literal>dir_count</literal> - total number of directories to test. Must be less
+ than or equal to <literal>thrhi</literal></para>
+ </listitem>
+ <listitem>
+ <para><literal>stripe_count </literal>- number stripe on OST objects</para>
+ </listitem>
+ <listitem>
+ <para><literal>tests_str</literal> - test operations. Must have at least "create" and
+ "destroy"</para>
+ </listitem>
+ <listitem>
+ <para><literal>start_number</literal> - base number for each thread to prevent name
+ collisions</para>
+ </listitem>
+ <listitem>
+ <para><literal>layer</literal> - MDS stack's layer to be tested</para>
+ </listitem>
+ </itemizedlist>
+ <para>Run without OST objects creation:</para>
+ <para>Setup the Lustre MDS without OST mounted. Then invoke the <literal>mds-survey</literal> script</para>
+ <screen>$ thrhi=64 file_count=200000 sh mds-survey</screen>
+ <para>Run with OST objects creation:</para>
+ <para>Setup the Lustre MDS with at least one OST mounted. Then invoke the
+ <literal>mds-survey</literal> script with <literal>stripe_count</literal>
+ parameter</para>
+ <screen>$ thrhi=64 file_count=200000 stripe_count=2 sh mds-survey</screen>
+ <para>Note: a specific MDT instance can be specified using targets variable.</para>
+ <screen>$ targets=lustre-MDT0000 thrhi=64 file_count=200000 stripe_count=2 sh mds-survey</screen>
+ </listitem>
+ </orderedlist>
+ <section remap="h3">
+ <title>Output Files</title>
+ <para>When the <literal>mds-survey</literal> script runs, it creates a number of working files and a pair of result files. All files start with the prefix defined in the variable <literal>${rslt}</literal>.</para>
+ <informaltable frame="all">
+ <tgroup cols="2">
+ <colspec colname="c1" colwidth="50*"/>
+ <colspec colname="c2" colwidth="50*"/>
+ <thead>
+ <row>
+ <entry>
+ <para><emphasis role="bold">File</emphasis></para>
+ </entry>
+ <entry>
+ <para><emphasis role="bold">Description</emphasis></para>
+ </entry>
+ </row>
+ </thead>
<tbody>
<row>
- <entry><para><emphasis role="bold">Note -</emphasis><anchor xml:id="dbdoclet.50438212_pgfId-1290070" xreflabel=""/>We have frequently discovered wide performance variations across all LUNs in a cluster. This may be caused by faulty disks, RAID parity reconstruction during the test, or faulty network hardware.</para></entry>
+ <entry>
+ <para> <literal>${rslt}.summary</literal></para>
+ </entry>
+ <entry>
+ <para> Same as stdout</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para> <literal>${rslt}.script_*</literal></para>
+ </entry>
+ <entry>
+ <para> Per-host test script files</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para> <literal>${rslt}.detail_tmp*</literal></para>
+ </entry>
+ <entry>
+ <para> Per-mdt result files</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para> <literal>${rslt}.detail</literal></para>
+ </entry>
+ <entry>
+ <para> Collected result files for post-mortem</para>
+ </entry>
</row>
</tbody>
</tgroup>
</informaltable>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1290071" xreflabel=""/>To run the ost_survey script, supply a file size (in KB) and the Lustre mount point. For example, run:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1290072" xreflabel=""/>$ ./ost-survey.sh 10 /mnt/lustre
-</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301765" xreflabel=""/>Typical output is:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1290073" xreflabel=""/>Average read Speed: 6.73
-<anchor xml:id="dbdoclet.50438212_pgfId-1290074" xreflabel=""/>Average write Speed: 5.41
-<anchor xml:id="dbdoclet.50438212_pgfId-1290075" xreflabel=""/>read - Worst OST indx 0 5.84 MB/s
-<anchor xml:id="dbdoclet.50438212_pgfId-1290076" xreflabel=""/>write - Worst OST indx 0 3.77 MB/s
-<anchor xml:id="dbdoclet.50438212_pgfId-1290077" xreflabel=""/>read - Best OST indx 1 7.38 MB/s
-<anchor xml:id="dbdoclet.50438212_pgfId-1290078" xreflabel=""/>write - Best OST indx 1 6.31 MB/s
-<anchor xml:id="dbdoclet.50438212_pgfId-1290079" xreflabel=""/>3 OST devices found
-<anchor xml:id="dbdoclet.50438212_pgfId-1290080" xreflabel=""/>Ost index 0 Read speed 5.84 Write speed \
- 3.77
-<anchor xml:id="dbdoclet.50438212_pgfId-1290081" xreflabel=""/>Ost index 0 Read time 0.17 Write time \
- 0.27
-<anchor xml:id="dbdoclet.50438212_pgfId-1290082" xreflabel=""/>Ost index 1 Read speed 7.38 Write speed \
- 6.31
-<anchor xml:id="dbdoclet.50438212_pgfId-1290083" xreflabel=""/>Ost index 1 Read time 0.14 Write time \
- 0.16
-<anchor xml:id="dbdoclet.50438212_pgfId-1290084" xreflabel=""/>Ost index 2 Read speed 6.98 Write speed \
- 6.16
-<anchor xml:id="dbdoclet.50438212_pgfId-1290085" xreflabel=""/>Ost index 2 Read time 0.14 Write time \
- 0.16
-</screen>
+ <para>The <literal>mds-survey</literal> script iterates over the given number of threads performing the specified tests and checks that all test processes have completed successfully.</para>
+ <note>
+ <para>The <literal>mds-survey</literal> script may not clean up properly if it is aborted or if it encounters an unrecoverable error. In this case, a manual cleanup may be required, possibly including killing any running instances of <literal>lctl</literal>, removing <literal>echo_client</literal> instances created by the script and unloading <literal>obdecho</literal>.</para>
+ </note>
</section>
- <section remap="h2">
- <title>24.5 <anchor xml:id="dbdoclet.50438212_58201" xreflabel=""/>Collecting Application Profiling Information (stats-collect)</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299523" xreflabel=""/>The stats-collect utility contains the following scripts used to collect application profiling information from Lustre clients and servers:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299524" xreflabel=""/>lstat.sh - Script for a single node that is run on each profile node.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299525" xreflabel=""/>gather_stats_everywhere.sh - Script that collect statistics.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299526" xreflabel=""/>config.sh - Script that contains customized configuration descriptions.</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299527" xreflabel=""/>The stats-collect utility requires:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299528" xreflabel=""/> Lustre to be installed and set up on your cluster</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1299529" xreflabel=""/> SSH and SCP access to these nodes without requiring a password</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <section remap="h3">
- <title><anchor xml:id="dbdoclet.50438212_pgfId-1299531" xreflabel=""/>24.5.1 Using stats-collect</title>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301821" xreflabel=""/>The stats-collect utility is configured by including profiling configuration variables in the config.sh script. Each configuration variable takes the following form, where 0 indicates statistics are to be collected only when the script starts and stops and <emphasis>n</emphasis> indicates the interval in seconds at which statistics are to be collected:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1301853" xreflabel=""/><emphasis><statistic></emphasis>_INTERVAL=<emphasis>[</emphasis>0<emphasis>|n]</emphasis></screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301887" xreflabel=""/>Statistics that can be collected include:</para>
- <itemizedlist><listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301893" xreflabel=""/>VMSTAT - Memory and CPU usage and aggregate read/write operations</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301898" xreflabel=""/>SERVICE - Lustre OST and MDT RPC service statistics</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301899" xreflabel=""/>BRW - OST block read/write statistics (brw_stats)</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301900" xreflabel=""/>SDIO - SCSI disk IO statistics (sd_iostats)</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301901" xreflabel=""/>MBALLOC - ldiskfs block allocation statistics</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301902" xreflabel=""/>IO - Lustre target operations statistics</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301903" xreflabel=""/>JBD - ldisfs journal statistics</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-<listitem>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301904" xreflabel=""/>CLIENT - Lustre OSC request statistics</para>
- </listitem>
-<listitem>
- <para> </para>
- </listitem>
-</itemizedlist>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1301989" xreflabel=""/>To collect profile information:</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1302001" xreflabel=""/> 1. Begin collecting statistics on each node specified in the config.sh script.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304893" xreflabel=""/>Starting the collect profile daemon on each node by entering:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1302024" xreflabel=""/>sh gather_stats_everywhere.sh config.sh start
+ <section remap="h4">
+ <title>Script Output</title>
+ <para>The <literal>.summary</literal> file and <literal>stdout</literal> of the <literal>mds-survey</literal> script contain lines like:</para>
+ <screen>mdt 1 file 100000 dir 4 thr 4 create 5652.05 [ 999.01,46940.48] destroy 5797.79 [ 0.00,52951.55] </screen>
+ <para>Where:</para>
+ <informaltable frame="all">
+ <tgroup cols="2">
+ <colspec colname="c1" colwidth="50*"/>
+ <colspec colname="c2" colwidth="50*"/>
+ <thead>
+ <row>
+ <entry>
+ <para><emphasis role="bold">Parameter and value</emphasis></para>
+ </entry>
+ <entry>
+ <para><emphasis role="bold">Description</emphasis></para>
+ </entry>
+ </row>
+ </thead>
+ <tbody>
+ <row>
+ <entry>
+ <para>mdt 1</para>
+ </entry>
+ <entry>
+ <para>Total number of MDT under test</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para>file 100000</para>
+ </entry>
+ <entry>
+ <para>Total number of files per thread to operate</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para>dir 4</para>
+ </entry>
+ <entry>
+ <para>Total number of directories to operate</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para>thr 4</para>
+ </entry>
+ <entry>
+ <para>Total number of threads operate over all directories</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para>create, destroy</para>
+ </entry>
+ <entry>
+ <para>Tests name. More tests will be displayed on the same line.</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para>565.05</para>
+ </entry>
+ <entry>
+ <para>Aggregate operations over MDT measured by dividing the total number of operations by the elapsed time.</para>
+ </entry>
+ </row>
+ <row>
+ <entry>
+ <para>[999.01,46940.48]</para>
+ </entry>
+ <entry>
+ <para>Minimum and maximum instantaneous operation seen on any individual MDT</para>
+ </entry>
+ </row>
+ </tbody>
+ </tgroup>
+ </informaltable>
+ <note>
+ <para>If script output has "ERROR", this usually means there is issue during the run such as running out of space on the MDT and/or OST. More detailed debug information is available in the ${rslt}.detail file</para>
+ </note>
+ </section>
+ </section>
+ <section xml:id="dbdoclet.50438212_58201">
+ <title><indexterm><primary>benchmarking</primary><secondary>application profiling</secondary></indexterm>Collecting Application Profiling Information (<literal>stats-collect</literal>)</title>
+ <para>The <literal>stats-collect</literal> utility contains the following scripts used to collect application profiling information from Lustre clients and servers:</para>
+ <itemizedlist>
+ <listitem>
+ <para><literal>lstat.sh</literal> - Script for a single node that is run on each profile node.</para>
+ </listitem>
+ <listitem>
+ <para><literal>gather_stats_everywhere.sh</literal> - Script that collect statistics.</para>
+ </listitem>
+ <listitem>
+ <para><literal>config.sh</literal> - Script that contains customized configuration descriptions.</para>
+ </listitem>
+ </itemizedlist>
+ <para>The <literal>stats-collect</literal> utility requires:</para>
+ <itemizedlist>
+ <listitem>
+ <para>Lustre software to be installed and set up on your cluster</para>
+ </listitem>
+ <listitem>
+ <para>SSH and SCP access to these nodes without requiring a password</para>
+ </listitem>
+ </itemizedlist>
+ <section remap="h3">
+ <title>Using <literal>stats-collect</literal></title>
+ <para>The stats-collect utility is configured by including profiling configuration variables in the config.sh script. Each configuration variable takes the following form, where 0 indicates statistics are to be collected only when the script starts and stops and <emphasis>n</emphasis> indicates the interval in seconds at which statistics are to be collected:</para>
+ <screen><replaceable>statistic</replaceable>_INTERVAL=<replaceable>0|n</replaceable></screen>
+ <para>Statistics that can be collected include:</para>
+ <itemizedlist>
+ <listitem>
+ <para><literal>VMSTAT</literal> - Memory and CPU usage and aggregate read/write operations</para>
+ </listitem>
+ <listitem>
+ <para><literal>SERVICE</literal> - Lustre OST and MDT RPC service statistics</para>
+ </listitem>
+ <listitem>
+ <para><literal>BRW</literal> - OST bulk read/write statistics (brw_stats)</para>
+ </listitem>
+ <listitem>
+ <para><literal>SDIO</literal> - SCSI disk IO statistics (sd_iostats)</para>
+ </listitem>
+ <listitem>
+ <para><literal>MBALLOC</literal> - <literal>ldiskfs</literal> block allocation statistics</para>
+ </listitem>
+ <listitem>
+ <para><literal>IO</literal> - Lustre target operations statistics</para>
+ </listitem>
+ <listitem>
+ <para><literal>JBD</literal> - ldiskfs journal statistics</para>
+ </listitem>
+ <listitem>
+ <para><literal>CLIENT</literal> - Lustre OSC request statistics</para>
+ </listitem>
+ </itemizedlist>
+ <para>To collect profile information:</para>
+ <para>Begin collecting statistics on each node specified in the config.sh script.</para>
+ <orderedlist>
+ <listitem>
+ <para>Starting the collect profile daemon on each node by entering:</para>
+ <screen>sh gather_stats_everywhere.sh config.sh start
</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1302184" xreflabel=""/> 2. Run the test.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1302047" xreflabel=""/> 3. Stop collecting statistics on each node, clean up the temporary file, and create a profiling tarball.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1304926" xreflabel=""/>Enter:</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1302051" xreflabel=""/>sh gather_stats_everywhere.sh config.sh stop <emphasis><log_name.tgz></emphasis></screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1305257" xreflabel=""/>When <emphasis><log_name.tgz></emphasis> is specified, a profile tarball /tmp/<emphasis><log_name.tgz></emphasis> is created.</para>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1302055" xreflabel=""/> 4. Analyze the collected statistics and create a csv tarball for the specified profiling data.</para>
- <screen><anchor xml:id="dbdoclet.50438212_pgfId-1302059" xreflabel=""/>sh gather_stats_everywhere.sh config.sh analyse log_tarball.tgz csv
+ </listitem>
+ <listitem>
+ <para>Run the test.</para>
+ </listitem>
+ <listitem>
+ <para>Stop collecting statistics on each node, clean up the temporary file, and create a profiling tarball.</para>
+ <para>Enter:</para>
+ <screen>sh gather_stats_everywhere.sh config.sh stop <replaceable>log_name</replaceable>.tgz</screen>
+ <para>When <literal><replaceable>log_name</replaceable>.tgz</literal>
+ is specified, a profile tarball <literal><replaceable>/tmp/log_name</replaceable>.tgz</literal> is created.</para>
+ </listitem>
+ <listitem>
+ <para>Analyze the collected statistics and create a csv tarball for the specified profiling data.</para>
+ <screen>sh gather_stats_everywhere.sh config.sh analyse log_tarball.tgz csv
</screen>
- <para><anchor xml:id="dbdoclet.50438212_pgfId-1302020" xreflabel=""/> </para>
- </section>
+ </listitem>
+ </orderedlist>
</section>
</section>
-</article>
+</chapter>
git://git.whamcloud.com / doc / manual.git / blobdiff