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-3886 Lustre <p>This problem is similar with <a href="https://jira.whamcloud.com/browse/LU-3846" title="Sanity 56u error with two OSTs" class="issue-link" data-issue-key="LU-3846"><del>LU-3846</del></a> and <a href="https://jira.whamcloud.com/browse/LU-3858" title="sanity test_27A: @@@@@@ FAIL: stripe count -1 != 1" class="issue-link" data-issue-key="LU-3858"><del>LU-3858</del></a>. The test suit should wait for a few seconds after it clear the stripe of the directory. Otherwise, the newly created entries under the directory will have 2 stripe counts rather than 1.</p> LU-3886

sanity test_56a: @@@@@@ FAIL: /usr/bin/lfs getstripe --obd wrong: found 6, expected 3

Bug Major Resolved Cannot Reproduce Emoly Liu Li Xi Thu, 5 Sep 2013 10:51:04 +0000 Tue, 17 Mar 2020 03:24:41 +0000 Tue, 17 Mar 2020 03:24:41 +0000 Lustre 2.5.0 0 4 <p>Emoly</p> <p>Could you please comment on this one?</p> <p>Thanks</p> <p>Peter</p> <p>"lfs setstripe -d" should be enough to clear directory striping information. LiXi, could you tell me how you hit this problem? </p> <p>IMO, we can add "lfs getstripe -v" after "setstripe -d" to print that striping information, and see if this problem will happen again.</p> <p>I've post the script (run-sanity.sh) to hit this problem (and <a href="https://jira.whamcloud.com/browse/LU-3858" title="sanity test_27A: @@@@@@ FAIL: stripe count -1 != 1" class="issue-link" data-issue-key="LU-3858"><del>LU-3858</del></a>, <a href="https://jira.whamcloud.com/browse/LU-3846" title="Sanity 56u error with two OSTs" class="issue-link" data-issue-key="LU-3846"><del>LU-3846</del></a>). It will hit the problem every time it runs.</p> <p>What is interesting is that when I add a sleep into the test suit, the problem is gone. That makes me believe that the problem is similar with <a href="https://jira.whamcloud.com/browse/LU-3858" title="sanity test_27A: @@@@@@ FAIL: stripe count -1 != 1" class="issue-link" data-issue-key="LU-3858"><del>LU-3858</del></a>. Emoly, would you please check the attachment of <a href="https://jira.whamcloud.com/browse/LU-3858" title="sanity test_27A: @@@@@@ FAIL: stripe count -1 != 1" class="issue-link" data-issue-key="LU-3858"><del>LU-3858</del></a> first? I.e. <a href="https://jira.hpdd.intel.com/secure/attachment/13414/run.sh" class="external-link" target="_blank" rel="nofollow noopener">https://jira.hpdd.intel.com/secure/attachment/13414/run.sh</a>. It shows that the effect of default stripe is delayed. Thanks!</p> <p>test_56a() { # was test_56<br/> rm -rf $DIR/$tdir<br/> $SETSTRIPE -d $DIR<br/> test_mkdir -p $DIR/$tdir/dir<br/> NUMFILES=3<br/> NUMFILESx2=$(($NUMFILES * 2))<br/> sleep 10 # This will fix the problem.<br/> for i in `seq 1 $NUMFILES` ; do<br/> touch $DIR/$tdir/file$i<br/> touch $DIR/$tdir/dir/file$i<br/> done<br/> ......</p> <p>I just ran the script of "https://jira.hpdd.intel.com/secure/attachment/13414/run.sh" on my local VM. It showed me 10000 times "No errors after xxx iters".</p> <p>My step is:<br/> 1. mount lustre<br/> 2. mkdir /mnt/lustre/d<br/> 3. sh run.sh /mnt/lustre/d</p> <p>I tried several times, no error happened.</p> <p>Oh, sorry, plase run on lustre mount point '/mnt/lustre' rather than its directory '/mnt/lustre/dir', i.e. sh run.sh /mnt/lustre/</p> <p>I got following output:</p> <p>No error after 640 iters<br/> -1 != 1<br/> Does not become correct after 0 seconds<br/> Does not become correct after 1 seconds<br/> Does not become correct after 2 seconds<br/> Does not become correct after 3 seconds<br/> Does not become correct after 4 seconds<br/> Does not become correct after 5 seconds<br/> Does not become correct after 6 seconds<br/> Become correct after 7 seconds</p> <p>Yes, this time I hit that. I will investigate it.</p> <p>The problem you found by run.sh is probably related to the following code:</p> <p>When we set stripe for root(mount point), set_default is enabled in ll_dir_ioctl()</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> <span class="code-keyword">case</span> LL_IOC_LOV_SETSTRIPE: { ... <span class="code-object">int</span> set_default = 0; ... <span class="code-keyword">if</span> (inode-&gt;i_sb-&gt;s_root == file-&gt;f_dentry) set_default = 1; <span class="code-comment">/* in v1 and v3 cases lumv1 points to data */</span> rc = ll_dir_setstripe(inode, lumv1, set_default); </pre> </div></div> <p>Then, in ll_dir_setstripe() if set_default=1, we will call ll_send_mgc_param() to set information asynchronously.</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> <span class="code-keyword">if</span> (set_default &amp;&amp; mgc-&gt;u.cli.cl_mgc_mgsexp) { <span class="code-comment">/* Set root stripesize */</span> <span class="code-comment">/* Set root stripecount */</span> <span class="code-comment">/* Set root stripeoffset */</span> } </pre> </div></div> <p>Since you run setstripe very frequently and many times in run.sh, the config log queue might be very long (bottleneck), and mgs will take more time to process it.</p> <p>BTW, can you hit this problem if you don't use run.sh, just run sanity.sh regularly?</p> <p>Yeah, I hit the problem of 'found 6, expected 3' every time when I run sanity.sh. </p> <p>I've hit this problem with lustre-master tag 2.6.92. Results at <a href="https://testing.hpdd.intel.com/test_sets/37e63f92-9f0d-11e4-91b3-5254006e85c2" class="external-link" target="_blank" rel="nofollow noopener">https://testing.hpdd.intel.com/test_sets/37e63f92-9f0d-11e4-91b3-5254006e85c2</a></p> <p>Haven't seen this in a long time.</p> Duplicate LU-7071 Related LU-3846 LU-3858 Development End date Tue, 1 Sep 2015 10:51:04 +0000 Rank 1|hzw073: Rank (Obsolete) 10111 Severity Start date Thu, 5 Sep 2013 10:51:04 +0000