https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 https://jira.whamcloud.com/browse/LU-12345 Lustre <p>From 8c48f7e88e293b9dd422bd8884842aea85d30b22<br/> ..<br/> We found performance regression when using bigalloc with "nodelalloc"<br/> (1MB cluster size):<br/> 1. mke2fs -C 1048576 -O ^has_journal,bigalloc /dev/sda<br/> 2. mount -o nodelalloc /dev/sda /test/<br/> 3. time dd if=/dev/zero of=/test/io bs=1048576 count=1024<br/> The "dd" will cost about 2 seconds to finish, but if we mke2fs without<br/> "bigalloc", "dd" will only cost less than 1 second.<br/>  <br/>  </p> LU-12345

backport - ext4: optimize ext4_find_delalloc_range() in nodelalloc mode

Bug Major Resolved Fixed Artem Blagodarenko Artem Blagodarenko patch Tue, 28 May 2019 11:08:18 +0000 Tue, 1 Nov 2022 16:25:49 +0000 Sat, 1 Jun 2019 14:37:48 +0000 Lustre 2.13.0 Lustre 2.12.5 0 5 <p>I don't think osd-ldiskfs is using delalloc? This may affect local ext4 performance but I don't think it will affect Lustre. </p> <p>I remember one reason do not use delalloc is missing suport for ordered writes. i.e. allocation is driven by VM as process of memory cleanup/release, not by JBD which is required to know when specific pages have been commited. this is different from ZFS where data allocation is done as part of commiting TXG.</p> <p>&gt; I don't think osd-ldiskfs is using delalloc? This may affect local ext4 performance but I don't think it will affect Lustre.</p> <p>During IOR single thread 4K random re-write over 16GB file on one of our machine <br/>     the CPU usage on the OSS is 100% and resulting IOPS are extremely<br/>     low. Here is a snippet of perf report:<br/>  <br/>    |-<del>89.87%</del>- ldiskfs_es_find_delayed_extent_range<br/>     |          ldiskfs_fiemap<br/>     |          osd_is_mapped<br/>     |          osd_declare_write_commit<br/>     |          ofd_commitrw_write<br/>     |          ofd_commitrw<br/>     |          tgt_brw_write<br/>     |          tgt_request_handle<br/>     |          ptlrpc_server_handle_request<br/>     |          ptlrpc_main<br/>     |          kthread<br/>     |          ret_from_fork</p> <p><a href="https://jira.whamcloud.com/secure/ViewProfile.jspa?name=adilger" class="user-hover" rel="adilger">adilger</a>,</p> <p>There is "ext4: optimize ext4_find_delalloc_range() in nodelalloc<br/>     mode" patch that prevents executin<br/>     ldiskfs_es_find_delayed_extent_range() if no delalloc enabled. The<br/>     same optimization also added for ldiskfs_find_delayed_extent()<br/>     function that improve performance dromaticaly.<br/>  <br/>     Here is results of testing on two node system.<br/>     Without the patch:<br/>     avg-cpu:  %user   %nice %system %iowait  %steal   %idle<br/>                0.00    0.00   56.30    0.06    0.00   43.63<br/>  <br/>      Device:         rrqm/s   wrqm/s     r/s     w/s    rMB/s    wMB/s<br/>      avgrq-sz avgqu-sz   await r_await w_await  svctm  %util<br/>      sds               0.00     0.00    0.00 1174.00     0.00     4.59<br/>      8.00     0.84    0.71    0.00    0.71   0.01   1.20<br/>  <br/>      With patch:<br/>      08/29/2018 01:13:22 AM<br/>      avg-cpu:  %user   %nice %system %iowait  %steal   %idle<br/>                 0.00    0.00    4.13   30.37    0.00   65.50<br/>  <br/>     Device:         rrqm/s   wrqm/s     r/s     w/s    rMB/s      wMB/s<br/>     avgrq-sz avgqu-sz   await r_await w_await  svctm %util<br/>     sds               0.00     0.00    0.00 54117.82     0.00     211.43<br/>     8.00   152.59    2.82    0.00    2.82   0.02 99.01</p> <p>Artem Blagodarenko (c17828@cray.com) uploaded a new patch: <a href="https://review.whamcloud.com/34982" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/34982</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-12345" title="backport - ext4: optimize ext4_find_delalloc_range() in nodelalloc mode" class="issue-link" data-issue-key="LU-12345"><del>LU-12345</del></a> ldiskfs: optimize nodelalloc mode<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: 1<br/> Commit: 7e74f2751f8cfdce2915d9b92259cdfc78d3a8d4</p> <p>Other than this (obviously significant) performance problems, have you seen other problems with <tt>bigalloc</tt>? I'm worried that there are hidden problems because the <tt>osd-ldiskfs</tt> code is dealing with blocks instead of chunks in various places (<tt>chunk</tt>, <tt>C2B</tt>, and <tt>B2C</tt> do not appear anywhere in that code).</p> <p><a href="https://jira.whamcloud.com/secure/ViewProfile.jspa?name=adilger" class="user-hover" rel="adilger">adilger</a> going to make bigalloc testing in near future.</p> <p>Oleg Drokin (green@whamcloud.com) merged in patch <a href="https://review.whamcloud.com/34982/" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/34982/</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-12345" title="backport - ext4: optimize ext4_find_delalloc_range() in nodelalloc mode" class="issue-link" data-issue-key="LU-12345"><del>LU-12345</del></a> ldiskfs: optimize nodelalloc mode<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: <br/> Commit: af48ae8bff289b2bc083a888efeafa3c48df91e2</p> <p>Landed for 2.13</p> <p>Minh Diep (mdiep@whamcloud.com) uploaded a new patch: <a href="https://review.whamcloud.com/37538" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/37538</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-12345" title="backport - ext4: optimize ext4_find_delalloc_range() in nodelalloc mode" class="issue-link" data-issue-key="LU-12345"><del>LU-12345</del></a> ldiskfs: optimize nodelalloc mode<br/> Project: fs/lustre-release<br/> Branch: b2_12<br/> Current Patch Set: 1<br/> Commit: bbf850c5ac55b631081848a67831d23ac4a241a8</p> <p>Oleg Drokin (green@whamcloud.com) merged in patch <a href="https://review.whamcloud.com/37538/" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/37538/</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-12345" title="backport - ext4: optimize ext4_find_delalloc_range() in nodelalloc mode" class="issue-link" data-issue-key="LU-12345"><del>LU-12345</del></a> ldiskfs: optimize nodelalloc mode<br/> Project: fs/lustre-release<br/> Branch: b2_12<br/> Current Patch Set: <br/> Commit: f91552d85cd086b37f78524113cadf799c045220</p> Related LU-12103 LU-16286 Development Rank 1|i00h1j: Rank (Obsolete) 9223372036854775807 Severity