Whamcloud Community JIRA https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 [LU-11991] conf-sanity test_32xx fails for separate MGS https://jira.whamcloud.com/browse/LU-11991 Lustre <p>It is regression after patch <a href="https://review.whamcloud.com/33422" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/33422</a> "<tt><a href="https://jira.whamcloud.com/browse/LU-11556" title="conf-sanity test 32b crashes on MDT umount with “ASSERTION( atomic_read(&amp;d-&gt;ld_ref) == 0 ) failed: Refcount is 1”" class="issue-link" data-issue-key="LU-11556">LU-11556</a> tests: fix set_persistent_param_and_check breakage</tt>". With separate MGS set_persistent_param_and_check run at mgs node, but test stop it, and run local MDS+MGS.</p> LU-11991 conf-sanity test_32xx fails for separate MGS Bug Minor Open Unresolved Alexander Boyko Alexander Boyko Fri, 22 Feb 2019 11:20:05 +0000 Wed, 19 Feb 2020 19:41:20 +0000 0 3 <p>I see you are resolving this with <a href="https://review.whamcloud.com/#/c/33589." class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/#/c/33589.</a> Why the separate ticket?</p> <p>Only skipping for 32xx at <a href="https://review.whamcloud.com/#/c/33589" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/#/c/33589</a></p> <p>Actually, looking at patch <a href="https://review.whamcloud.com/33422" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/33422</a> "<tt><a href="https://jira.whamcloud.com/browse/LU-11556" title="conf-sanity test 32b crashes on MDT umount with “ASSERTION( atomic_read(&amp;d-&gt;ld_ref) == 0 ) failed: Refcount is 1”" class="issue-link" data-issue-key="LU-11556">LU-11556</a> tests: fix set_persistent_param_and_check breakage</tt>" it isn't clear that this introduced the problem that you describe. It was just fixing a problem where "<tt>$node</tt>" was being passed to <tt>set_persistent_param_and_check()</tt> instead of "<tt>$facet</tt>", it doesn't do anything related to starting/stopping the MGS device? Is there some other patch that introduced this breakage, or am I missing the connection?</p> Related LU-10717 LU-8110 Development Rank 1|i00c1z: Rank (Obsolete) 9223372036854775807 Severity