Whamcloud Community JIRA

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-2148] Kernel update for latest FC18 kernel https://jira.whamcloud.com/browse/LU-2148 Lustre LU-2148

Kernel update for latest FC18 kernel

Improvement Critical Resolved Fixed Yang Sheng Yang Sheng Thu, 11 Oct 2012 10:29:16 +0000 Tue, 23 Apr 2013 17:13:38 +0000 Wed, 13 Mar 2013 09:31:03 +0000 Lustre 2.4.0 Lustre 2.4.0 0 11 Landed for 2.4 This needs further work to actually build under fc18. I've got an fc18 builder online and reviews are using it so that the necessary changes to lbuild etc can be made. NOTE: The review build will always show fc18 as building successfully, this is so that we do not stop reviews landing. You will need to look at the actual output for fc18 and see if it is a README or rpms. In either case you can then review the console to ensure everything is OK. Hi. Chiris, Thanks for the explain. I'll commit a patch for fc18 build. Patch commit to: <a href="http://review.whamcloud.com/#change,5194" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/#change,5194</a> I got following error when trying to install build lustre-reviews 12744 fc18 client with the command: loadjenkinsbuild -b 12744 -j lustre-reviews -p test-fc18-x86_64 -t client -d fc18 -a x86_64 --packages="gcc,opensm" -n client-4 -r the system hang when running post install scripts: <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>Sending SIGTERM to remaining processes... Sending SIGKILL to remaining processes... Unmounting file systems. Unmounted /mnt/sysimage/sys/fs/selinux. Unmounted /mnt/sysimage/sys. Unmounted /mnt/sysimage/proc. Not all loop devices detached, 3 left. Detaching DM devices. Not all DM devices detached, 1 left. Cannot finalize remaining file systems and devices, giving up. [ 559.498648] Restarting system. ? </pre> </div></div> Sarah: You will be able to ssh to the node at this point; Best to ssh to it as root and then reboot it once. This should then be good Chris, I can ssh to the node as root, but reboot didn't work for me, there is no lustre installed on the node. I found the kernel version is 3.7.4-204.fc18.x86_64 which is different from the 3.6.10_4.fc18.x86_64 used for lustre. Here is the build link: <a href="http://build.whamcloud.com/job/lustre-reviews/12744/arch=x86_64,build_type=client,distro=fc18,ib_stack=inkernel/" class="external-link" target="_blank" rel="nofollow noopener">http://build.whamcloud.com/job/lustre-reviews/12744/arch=x86_64,build_type=client,distro=fc18,ib_stack=inkernel/</a> [root@client-5 ~]# rpm -qa|grep lustre [root@client-5 ~]# uname -a Linux client-5.lab.whamcloud.com 3.7.4-204.fc18.x86_64 #1 SMP Wed Jan 23 16:44:29 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux When I think about it I'm not surprised, I've never seen it install fc18 - last time I looked we didn't even have Lustre for FC18. I'll investigate. So the problem at the moment is this; <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java">yum install -y kernel-3.6.10-4.fc18.x86_64 lustre-client-modules lustre-client lustre-client-tests + yum install -y kernel-3.6.10-4.fc18.x86_64 lustre-client-modules lustre-client lustre-client-tests Resolving Dependencies --> Running transaction check ---> Package kernel.x86_64 0:3.6.10-4.fc18 will be installed ---> Package lustre-client.x86_64 0:2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e will be installed --> Processing Dependency: libnetsnmpagent.so.30()(64bit) for package: lustre-client-2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e.x86_64 --> Processing Dependency: libnetsnmp.so.30()(64bit) for package: lustre-client-2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e.x86_64 --> Processing Dependency: libnetsnmpmibs.so.30()(64bit) for package: lustre-client-2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e.x86_64 --> Processing Dependency: libsensors.so.4()(64bit) for package: lustre-client-2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e.x86_64 ---> Package lustre-client-modules.x86_64 0:2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e will be installed ---> Package lustre-client-tests.x86_64 0:2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e will be installed --> Processing Dependency: liblustreapi.so()(64bit) for package: lustre-client-tests-2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e.x86_64 --> Running transaction check ---> Package lm_sensors-libs.x86_64 0:3.3.2-5.fc18 will be installed ---> Package lustre-client-tests.x86_64 0:2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e will be installed --> Processing Dependency: liblustreapi.so()(64bit) for package: lustre-client-tests-2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e.x86_64 ---> Package net-snmp-agent-libs.x86_64 1:5.7.2-5.fc18 will be installed ---> Package net-snmp-libs.x86_64 1:5.7.2-5.fc18 will be installed --> Finished Dependency Resolution Error: Package: lustre-client-tests-2.3.59-3.6.10_4.fc18.x86_64_g6f0f53e.x86_64 (lustre-build) Requires: liblustreapi.so()(64bit) You could try using --skip-broken to work around the problem You could try running: rpm -Va --nofiles --nodigest </pre> </div></div> where is liblustreapi.so I would welcome someone else looking into this. Yang Sheng, I need you to look into the reason why the build will not load on Fedora 18. I have client-7 reserved and you can make use of this node. The best way for you to see the issue is to use this loadjenkinsbuild command. loadjenkinsbuild -b 12744 -j lustre-reviews -p test-fc18-x86_64 -t client -d fc18 -a x86_64 --packages="gcc,opensm" -n client-7 -r --user chris Once the install is complete you will see if you login as root (password lustre) the error above in ks-post.log Frank and I spent some time trying to work out what is missing, but for some reason the rpms will not install. You might want to compare the rpms built with those of you own test install to see if there is a difference. Is it possible that the lustre .spec file needs to indicate that liblustreapi.so is provided by the .rpm file that installs it? It is a common spec file between all builds, so whilst it is an error somewhere I fear it is no obvious. After investigate, I think this problem should relate to fc18 self. The rpms lost the Provides information. I am continue checking. HI, Chris, I found a workaround. Please add line '%__elf_flags % {nli} ' to build node ~/.rpmmacros and rebuild the rpms. The issue will be fixed. This seemed to work. What is the best way to make this work for an end user trying to build on Fedora - or is Fedora no supported and therefore not an issue worth worrying about? Hi, Chris, I have made a patch to fix this issue. commit to: <a href="http://review.whamcloud.com/5469" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/5469</a> I am trying to provision the fc18 client and got following error, the el6 works fine: <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>sarah@brent:~/test_logs$ loadjenkinsbuild -b 13610 -j lustre-reviews -p test-fc18-x86_64 -t client -d fc18 -a x86_64 --packages="gcc,opensm" -n client-18 -r Error: can't convert nil into String. Try --help for help. </pre> </div></div> Looks like 13610 hasn't product the rpms at all. I'll investigate. Hi, Sarah, The <a href="http://build.whamcloud.com/job/lustre-reviews/13661/" class="external-link" target="_blank" rel="nofollow noopener">http://build.whamcloud.com/job/lustre-reviews/13661/</a> can be used now. build lustre-reviews/13661 works for me, now I am running sanity.sh The fc18 has join to our build & autotest system, So i close this ticket. Further work will traced on lu-2686 Duplicate LU-2810 Related LU-812 LU-1337 LU-1994 LU-2686 Development Rank 1|hzva2v: Rank (Obsolete) 5167