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-14383 Lustre <p>The following race between mkdir and migrate may end with failure of mkdir:</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre> mkdir $DIR1/$tdir mkdir $DIR2/$tdir/dir2 &amp; $LFS migrate -m 1 $DIR1/$tdir </pre> </div></div> <p>Please confirm whether it is expected behavior has to be fixed.</p> LU-14383

race between lookup and migrate

Bug Minor Closed Not a Bug WC Triage Vladimir Saveliev Fri, 29 Jan 2021 12:24:57 +0000 Wed, 12 Jan 2022 13:33:17 +0000 Wed, 12 Jan 2022 13:33:17 +0000 0 3 <p>Vladimir Saveliev (c17830@cray.com) uploaded a new patch: <a href="https://review.whamcloud.com/41364" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/41364</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-14383" title="race between lookup and migrate" class="issue-link" data-issue-key="LU-14383"><del>LU-14383</del></a> tests: lookup and migrate race<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: 1<br/> Commit: e9a235febb1ab331929ef57c84e95cbe479596e1</p> <p>This looks to be expected in current implementation: "mkdir $DIR2/$tdir/dir2" tried to create "dir2" under current layout, however if server found the layout is changed, it may return error. To solve this race, client needs to refresh layout of $tdir and try again.</p> <blockquote><p>This looks to be expected in current implementation: "mkdir $DIR2/$tdir/dir2" tried to create "dir2" under current layout, however if server found the layout is changed, it may return error.</p></blockquote> <p>ok</p> <p>Btw, the same applies to stat:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> @@ -4631,12 +4631,13 @@ test_80c() { [ $MDSCOUNT -lt 2 ] &amp;&amp; skip <span class="code-quote">"needs &gt;= 2 MDTs"</span> &amp;&amp; <span class="code-keyword">return</span> mkdir $DIR1/$tdir + touch $DIR1/$tdir/file #define OBD_FAIL_MDS_OBJECT_LOCK_DELAY 0x16b do_facet mds1 $LCTL set_param fail_loc=0x8000016b - mkdir $DIR2/$tdir/dir2 &amp; - MKDIRPID=$! + stat $DIR2/$tdir/file &amp; + STATPID=$! $LFS migrate -m 1 $DIR1/$tdir - wait $MKDIRPID + wait $STATPID [ $? -eq 0 ] || error <span class="code-quote">"stat failed"</span> } </pre> </div></div> <p>This results in:</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>== sanityn test 80c: Lookup and migrate race ========================================================= 18:04:30 (1612796670) fail_loc=0x8000016b stat: cannot stat '/mnt/lustre2/d80c.sanityn/file': No such file or directory </pre> </div></div> <p>If this is ok as well, then sanityn.sh:test_80b() should not break its "accessing the migrating directory" loop with:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> stat $migrate_dir2/file5 &gt; /dev/<span class="code-keyword">null</span> || { echo <span class="code-quote">"stat file5 fails"</span> <span class="code-keyword">break</span> } </pre> </div></div> <blockquote><p>To solve this race, client needs to refresh layout of $tdir and try again.</p></blockquote> <p>WIth -ENOENT clients are not able to decide whether they are to try again. Would it be possible to return -EAGAIN in case of race with migrate?</p> <p>This issue should be fixed by FID map, which is tracked under <a href="https://jira.whamcloud.com/browse/LU-7607" title="Preserve inode number after MDT migration " class="issue-link" data-issue-key="LU-7607">LU-7607</a>. The patch is on <a href="https://review.whamcloud.com/#/c/38233/" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/#/c/38233/</a>, but I don't have time to update it recently.</p> Development Rank 1|i01ks7: Rank (Obsolete) 9223372036854775807 Severity