Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-14902

Crash while building TokuDB on Fedora 27

Details

    • Bug
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Not a Bug
    • 10.2(EOL)
    • N/A

    Description

      I get segmentation fault while linking libtokuportability_static_conv.a:

      [ 74%] Building CXX object storage/spider/CMakeFiles/spider.dir/hs_client/string_util.cpp.o
      [ 74%] Linking CXX shared module ha_spider.so
      [ 74%] Built target spider
      Scanning dependencies of target test_sql_discovery
      [ 74%] Building CXX object storage/test_sql_discovery/CMakeFiles/test_sql_discovery.dir/test_sql_discovery.cc.o
      [ 74%] Linking CXX shared module ha_test_sql_discovery.so
      [ 74%] Built target test_sql_discovery
      [ 74%] Linking CXX static library libtokuportability_static_conv.a
      Error running link command: Segmentation fault
      make[2]: *** [storage/tokudb/PerconaFT/portability/CMakeFiles/tokuportability_static_conv.dir/build.make:330: storage/tokudb/PerconaFT/portability/libtokuportability_static_conv.a] Error 1
      make[1]: *** [CMakeFiles/Makefile2:7743: storage/tokudb/PerconaFT/portability/CMakeFiles/tokuportability_static_conv.dir/all] Error 2
      make: *** [Makefile:163: all] Error 2
      

      It happened to me last week and still happens with current code:

      [openxs@fc23 server]$ git branch
      * 10.3
      [openxs@fc23 server]$ git log -1
      commit fe79ac5b0e48395e55a4fb78196198b8542cc3d5 (HEAD -> 10.3, origin/HEAD, origin/10.3)
      Author: Marko Mäkelä <marko.makela@mariadb.com>
      Date:   Tue Jan 9 13:45:39 2018 +0200
       
          MDEV-14837 Duplicate primary keys are allowed after ADD COLUMN / UPDATE
       
          This bug affected tables where the PRIMARY KEY contains variable-length
          columns, and ROW_FORMAT is COMPACT or DYNAMIC.
       
          rec_init_offsets_comp_ordinary(): Do not short-cut the parsing
          of the record header for records that contain explicit values
          for instantly added columns.
       
          rec_copy_prefix_to_buf(): Copy more header for records that
          contain explicit values for instantly added columns.
      

      cmake command line was (from fc -l):

      1001     git pull
      1002     git submodule update
      1003     make clean
      1004     ls -l CMake*.txt
      1005     rm CMakeCache.txt
      1006     cmake . -DCMAKE_BUILD_TYPE=RelWithDebInfo -DWITH_SSL=system -DWITH_ZLIB=bundled -DENABLED_LOCAL_INFILE=1 -DWITH_JEMALLOC=system -DWITH_WSREP=ON -DWITH_INNODB_DISALLOW_WRITES=ON -DPLUGIN_MROONGA=NO -DCMAKE_INSTALL_PREFIX=/home/openxs/dbs/maria10.3
      1007     time make -j 4
      1008     make
      

      make -j 4 failed with MyRocks, but it may be a topic for other bug report...

      Attachments

        Activity

          serg, if you mean buildbot Fedora27 VM, it's not the latest, that's the point. It has 4.13.15-300.fc27.x86_64, and according to my comments above, there everything worked, but on 4.14.13-300.fc27.x86_64 I got the segfault.

          elenst Elena Stepanova added a comment - serg , if you mean buildbot Fedora27 VM, it's not the latest, that's the point. It has 4.13.15-300.fc27.x86_64 , and according to my comments above, there everything worked, but on 4.14.13-300.fc27.x86_64 I got the segfault.

          Do you have any way for me to repeat/observe this? I've upgraded the VM (sudo dnf upgrade), but didn't get any crash either

          serg Sergei Golubchik added a comment - Do you have any way for me to repeat/observe this? I've upgraded the VM ( sudo dnf upgrade ), but didn't get any crash either

          I tried that too, got the current is fedora27-amd64 4.17.5-100.fc27.x86_64, and it didn't crash to me either. It appears that the problem is gone on this version (I tried 10.2 server of the time when the crash was observed, so it's not changes in the code). I could try to find an image of 4.14.13-300, but I don't see much point in it.

          valerii, can you confirm that it doesn't crash for you anymore on the recent Fedora 27?

          elenst Elena Stepanova added a comment - I tried that too, got the current is fedora27-amd64 4.17.5-100.fc27.x86_64, and it didn't crash to me either. It appears that the problem is gone on this version (I tried 10.2 server of the time when the crash was observed, so it's not changes in the code). I could try to find an image of 4.14.13-300, but I don't see much point in it. valerii , can you confirm that it doesn't crash for you anymore on the recent Fedora 27?

          I was able to build current 10.2 successfully today:

          ...
          [100%] Building CXX object extra/mariabackup/CMakeFiles/mariabackup.dir/encryption_plugin.cc.o
          [100%] Building CXX object extra/mariabackup/CMakeFiles/mariabackup.dir/__/__/sql/net_serv.cc.o
          [100%] Building C object extra/mariabackup/CMakeFiles/mariabackup.dir/__/__/libmysqld/libmysql.c.o
          [100%] Linking CXX executable mariabackup
          [100%] Built target mariabackup
           
          real    24m24.436s
          user    60m49.615s
          sys     4m31.455s
          [openxs@fc23 server]$ fc -l
          ...
          997      sudo dnf update
          998      cd git/server/
          999      git branch
          1000     git checkout 10.2
          1001     git pull
          1002     git submodule update
          1003     rm CMakeCache.txt
          1004     cmake . -DCMAKE_BUILD_TYPE=RelWithDebInfo -DWITH_SSL=system -DWITH_ZLIB=bundled -DENABLED_LOCAL_INFILE=1 -DWITH_JEMALLOC=system -DWITH_WSREP=ON -DWITH_INNODB_DISALLOW_WRITES=ON -DPLUGIN_MROONGA=NO -DCMAKE_INSTALL_PREFIX=/home/openxs/dbs/maria10.2
          1005     time make -j 4
          [openxs@fc23 server]$
          

          Current kernel is a bit older (no reboot happened yet):

          [openxs@fc23 server]$ gcc --version
          gcc (GCC) 7.3.1 20180303 (Red Hat 7.3.1-5)
          Copyright (C) 2017 Free Software Foundation, Inc.
          This is free software; see the source for copying conditions.  There is NO
          warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
           
          [openxs@fc23 server]$ uname -a
          Linux fc23 4.17.3-100.fc27.x86_64 #1 SMP Tue Jun 26 14:19:03 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
          

          But we can probably close this report as not repeatable with recent kernels any more.

          valerii Valerii Kravchuk added a comment - I was able to build current 10.2 successfully today: ... [100%] Building CXX object extra/mariabackup/CMakeFiles/mariabackup.dir/encryption_plugin.cc.o [100%] Building CXX object extra/mariabackup/CMakeFiles/mariabackup.dir/__/__/sql/net_serv.cc.o [100%] Building C object extra/mariabackup/CMakeFiles/mariabackup.dir/__/__/libmysqld/libmysql.c.o [100%] Linking CXX executable mariabackup [100%] Built target mariabackup   real 24m24.436s user 60m49.615s sys 4m31.455s [openxs@fc23 server]$ fc -l ... 997 sudo dnf update 998 cd git/server/ 999 git branch 1000 git checkout 10.2 1001 git pull 1002 git submodule update 1003 rm CMakeCache.txt 1004 cmake . -DCMAKE_BUILD_TYPE=RelWithDebInfo -DWITH_SSL=system -DWITH_ZLIB=bundled -DENABLED_LOCAL_INFILE=1 -DWITH_JEMALLOC=system -DWITH_WSREP=ON -DWITH_INNODB_DISALLOW_WRITES=ON -DPLUGIN_MROONGA=NO -DCMAKE_INSTALL_PREFIX=/home/openxs/dbs/maria10.2 1005 time make -j 4 [openxs@fc23 server]$ Current kernel is a bit older (no reboot happened yet): [openxs@fc23 server]$ gcc --version gcc (GCC) 7.3.1 20180303 (Red Hat 7.3.1-5) Copyright (C) 2017 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.   [openxs@fc23 server]$ uname -a Linux fc23 4.17.3-100.fc27.x86_64 #1 SMP Tue Jun 26 14:19:03 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux But we can probably close this report as not repeatable with recent kernels any more.

          Closing as "not a bug" in the sense that apparently it wasn't a MariaDB code problem, see comments above.

          elenst Elena Stepanova added a comment - Closing as "not a bug" in the sense that apparently it wasn't a MariaDB code problem, see comments above.

          People

            Unassigned Unassigned
            valerii Valerii Kravchuk
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Git Integration

                Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.