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-7754 Lustre <p>ONLY=300k bash sanity.sh:</p> <p>[ 89.828294] LNet: Service thread pid 4249 was inactive for 40.00s. The thread might be hung, or it might only be slow and will resume later. Dumping the stack trace for debugging purposes:<br/> [ 89.831356] Pid: 4249, comm: mdt01_001<br/> [ 89.831895] <br/> [ 89.831895] Call Trace:<br/> [ 89.832451] <span class="error">&#91;&lt;ffffffff810ac6de&gt;&#93;</span> ? getrawmonotonic+0x2e/0xc0<br/> [ 89.833222] <span class="error">&#91;&lt;ffffffff810828c5&gt;&#93;</span> __cond_resched+0x25/0x40<br/> [ 89.834035] <span class="error">&#91;&lt;ffffffff814e521a&gt;&#93;</span> _cond_resched+0x2a/0x40<br/> [ 89.834810] <span class="error">&#91;&lt;ffffffff814e5f11&gt;&#93;</span> mutex_lock+0x11/0x40<br/> [ 89.835565] <span class="error">&#91;&lt;ffffffffa07741b4&gt;&#93;</span> dmu_tx_assign+0x284/0x500 <span class="error">&#91;zfs&#93;</span><br/> [ 89.836410] <span class="error">&#91;&lt;ffffffffa0a1ae32&gt;&#93;</span> osd_trans_start+0xb2/0x410 <span class="error">&#91;osd_zfs&#93;</span><br/> [ 89.837282] <span class="error">&#91;&lt;ffffffffa032de15&gt;&#93;</span> top_trans_start+0x255/0x9c0 <span class="error">&#91;ptlrpc&#93;</span><br/> [ 89.838090] <span class="error">&#91;&lt;ffffffffa0bb88f9&gt;&#93;</span> lod_trans_start+0x59/0x60 <span class="error">&#91;lod&#93;</span><br/> [ 89.838854] <span class="error">&#91;&lt;ffffffffa0ae3cdf&gt;&#93;</span> mdd_trans_start+0xf/0x20 <span class="error">&#91;mdd&#93;</span><br/> [ 89.839594] <span class="error">&#91;&lt;ffffffffa0acf1a0&gt;&#93;</span> mdd_create+0x1170/0x1c70 <span class="error">&#91;mdd&#93;</span></p> LU-7754

DNE3: osd-zfs gets into a livelock if transaction is too big

Bug Minor Open Unresolved Alex Zhuravlev Alex Zhuravlev dne3 zfs Sun, 7 Feb 2016 14:30:44 +0000 Tue, 27 Aug 2019 17:51:50 +0000 Lustre 2.8.0 0 4 <p>Alex Zhuravlev (alexey.zhuravlev@intel.com) uploaded a new patch: <a href="http://review.whamcloud.com/18341" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/18341</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-7754" title="DNE3: osd-zfs gets into a livelock if transaction is too big" class="issue-link" data-issue-key="LU-7754">LU-7754</a> osd: osd-zfs should not wait indefinitely for a TXG<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: 1<br/> Commit: ebef50ed6f5032697f05c4bcc20c7fb329423a17</p> <p>Your patch turn this from a hang into a failure. That is an improvement, but it doesn't explain why this test failed? Do you have an unusual config (small MDT?) or is there some regression that makes the transaction too large?</p> <p>sanity/300k tries to create a big striped directory:</p> <p> $LFS setdirstripe -i 0 -c512 $DIR/$tdir/striped_dir</p> <p>with default MDSSIZE=200000 DMU fails to start such a big transaction.</p> <p>How large is the transaction? Do we have a larger MDS size in our testing?</p> <p>I guess this is because we don't run DNE + ZFS by default. </p> <p>transaction calculations:<br/> mem 4986830848, asize 119683940352, fsize 8506441728, usize 8497152000</p> <p>it seem to fail because of insufficient memory: 4986830848 (4755MB) is needed while the test system had 4GB in total.</p> <p>That is <tt>4755 MB / 512 stripes = 9 MB/stripe</tt> which seems like a lot of space to reserve? I thought we got away from O(n^2) transaction sizes for striped directories?</p> Related LU-9096 LU-8267 Development Rank 1|hzy0lr: Rank (Obsolete) 9223372036854775807 Severity