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-9726 Lustre <p>When I fetch the <a href="https://build.hpdd.intel.com/job/lustre-b2_10/arch=x86_64,build_type=client,distro=el7,ib_stack=inkernel/2/artifact/artifacts/SRPMS/lustre-2.10.0_RC1-1.src.rpm" class="external-link" target="_blank" rel="nofollow noopener">client SRPM</a> from the URL in the Environment field, it doesn't simply rebuild in a clean-build environment such as Copr (which is basically mock<span class="error">&#91;chain&#93;</span>):</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>checking for Linux sources... checking for ... no configure: error: Kernel source could not be found. error: Bad exit status from /var/tmp/rpm-tmp.0DFyxx (%build) </pre> </div></div> <p>You can see the <a href="https://copr-be.cloud.fedoraproject.org/results/managerforlustre/lustre-client/epel-7-x86_64/00573166-lustre/builder-live.log" class="external-link" target="_blank" rel="nofollow noopener">full log</a> of the build if you wish.</p> <p>As you can see in that log though, <em>kernel-devel</em> was installed:</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>Dependencies Resolved ================================================================================ Package Arch Version Repository Size ================================================================================ Installing: kernel-devel x86_64 3.10.0-514.26.1.el7 updates 13 M ... </pre> </div></div> <p>Is the Lustre autoconf macro that is looking for the kernel source not defaulting to <em>/usr/src/kernels/*</em> perhaps?</p> <a href="https://build.hpdd.intel.com/job/lustre-b2_10/arch=x86_64,build_type=client,distro=el7,ib_stack=inkernel/2/">https://build.hpdd.intel.com/job/lustre-b2_10/arch=x86_64,build_type=client,distro=el7,ib_stack=inkernel/2/</a> LU-9726

client SRPM does not rebuild

Bug Critical Open Unresolved Minh Diep Brian Murrell Fri, 30 Jun 2017 11:28:18 +0000 Wed, 19 Jul 2017 12:40:26 +0000 Lustre 2.10.0 0 4 <p>The problem seems to be in the nature of the checks for kernel source in <em>/usr/src/kernels/</em>:</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>for DEFAULT_LINUX in /lib/modules/$(uname -r)/{source,build} /usr/src/linux; do AS_IF([readlink -q -e $DEFAULT_LINUX &gt;/dev/null], [break]) done if test "$DEFAULT_LINUX" = "/lib/modules/$(uname -r)/source"; then PATHS="/lib/modules/$(uname -r)/build" fi </pre> </div></div> <p>The problem is that in a clean-room type build system like <em>Copr</em>, or even just <em>mock</em>, the build is done in a chroot so the kernel that is running on the system that has the chroot may not reflect the contents of the <em>kernel-devel</em> that was installed for the build.</p> <p>The fix is probably to add <span class="error">&#91;some subset -- perhaps just the highest version of&#93;</span> <em>/usr/src/kernels/*</em> to the end of {$PATHS} in the above search for a kernel source/devel location.</p> <p>Minh </p> <p>Can you please assist with this one?</p> <p>Thanks</p> <p>Peter</p> <p><a href="https://jira.whamcloud.com/secure/ViewProfile.jspa?name=brian" class="user-hover" rel="brian">brian</a>, I assume you're building client only. However, I don't see a '--disable-server' being passed to ./configure</p> <p><a href="https://jira.whamcloud.com/secure/ViewProfile.jspa?name=mdiep" class="user-hover" rel="mdiep">mdiep</a>: To be clear, I am trying to <span class="error">&#91;re&#93;</span>build the produced SRPM so I don't control the configure options.  The specfile inside the SRPM controls that.</p> <p>Ultimately, the SRPM output product of an RPM build should be rebuildable.  If I use the produced <em>lustre-client</em> SRPM file and then use <tt>rpmbuild --rebuild &lt;SRPM file&gt;</tt> that should rebuild the lustre-client binary RPM.</p> <p><a href="https://jira.whamcloud.com/secure/ViewProfile.jspa?name=brian" class="user-hover" rel="brian">brian</a> I agree. I don't think lustre packaging source any differently between server and client and that needs to be fixed. could you verify that you can build lustre client with passing --disable-server into rpmbuild?</p> <blockquote> <p>can build lustre client with passing --disable-server into rpmbuild</p></blockquote> <p>The argument is <tt>--without-servers</tt> and that only addresses the instructing of a client build, which is not what this ticket is reporting.  This is indeed another issue with the SRPM, but different than this ticket.</p> <p>The problem that this ticket is reporting is that kernel source detection/selection does not work in clean-room SRPM <span class="error">&#91;re&#93;</span>build (i.e. <tt>rpmbuild -<del>rebuild lustre-client</del>....src.rpm</tt> environments. It's limited to autodetection only on systems where the kernel you want to build for is the kernel the build system is booted on.  It should, at minimum use whatever is in <em>/usr/src/kernels/</em> particularly if there is only one item in that directory if no other location can be found.</p> <p>And using <em>/usr/src/kernels/</em> is important because the <em>/lib/modules/*/{build,source}</em> links only exist if you install a kernel which shouldn't be necessary in a build chroot such as <em>mock</em> creates.</p> <p>To clarify, the original issue described in this ticket, the need to add /usr/src/kernels/* to the search path for kernel source is actually addressed in <a href="https://jira.whamcloud.com/browse/LU-9775" title="find kernel-devel even if the current kernel is not installed in the build root" class="issue-link" data-issue-key="LU-9775"><del>LU-9775</del></a>.</p> <p>The second issue in this ticket, that the <em>lustre-client</em> SRPM re-build does not set the <tt>–disable-servers</tt> still needs to be addressed.</p> <p>Probably the easiest way to deal with that is not not require the option be set but to detect if the kernel source is patched for a server build or not automatically.</p> Development Rank 1|hzzfx3: Rank (Obsolete) 9223372036854775807 Severity