Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-8125

Assertion `!tmp_gco->next_gco || tmp_gco->last_sub_id > sub_id' failed at 10.1-patched/sql/rpl_parallel.cc:189: void finish_event_group(rpl_parallel_thread*, uint64, rpl_parallel_entry*, rpl_group_info*)



    • Type: Bug
    • Status: Closed (View Workflow)
    • Priority: Major
    • Resolution: Duplicate
    • Affects Version/s: 10.1
    • Fix Version/s: 10.1.5
    • Component/s: Replication
    • Labels:


      _Note: line numbers from commit 2f25c653ade1e73aa2b1aa77af9a4898bacb2330, patched with http://lists.askmonty.org/pipermail/commits/2015-May/007819.html_

      10.1-patched/sql/rpl_parallel.cc:189: void finish_event_group(rpl_parallel_thread*, uint64, rpl_parallel_entry*, rpl_group_info*): Assert
      ion `!tmp_gco->next_gco || tmp_gco->last_sub_id > sub_id' failed.
      150508 19:54:55 [ERROR] mysqld got signal 6 ;

      #0  0x00007f6bf0861f8c in pthread_kill () from /lib/x86_64-linux-gnu/libpthread.so.0
      #1  0x00007f6bf2c882e4 in my_write_core (sig=6) at /home/elenst/git/10.1-patched/mysys/stacktrace.c:456
      #2  0x00007f6bf2644aef in handle_fatal_signal (sig=6) at /home/elenst/git/10.1-patched/sql/signal_handler.cc:266
      #3  <signal handler called>
      #4  0x00007f6befecc425 in raise () from /lib/x86_64-linux-gnu/libc.so.6
      #5  0x00007f6befecfb8b in abort () from /lib/x86_64-linux-gnu/libc.so.6
      #6  0x00007f6befec50ee in ?? () from /lib/x86_64-linux-gnu/libc.so.6
      #7  0x00007f6befec5192 in __assert_fail () from /lib/x86_64-linux-gnu/libc.so.6
      #8  0x00007f6bf25b0c7b in finish_event_group (rpt=0x7f6bc3c5dc40, sub_id=14255, entry=0x7f6bc3c8c370, rgi=0x7f6bc1ecd000) at /home/elenst/git/10.1-patched/sql/rpl_parallel.cc:189
      #9  0x00007f6bf25b2da8 in handle_rpl_parallel_thread (arg=0x7f6bc3c5dc40) at /home/elenst/git/10.1-patched/sql/rpl_parallel.cc:980
      #10 0x00007f6bf085ce9a in start_thread () from /lib/x86_64-linux-gnu/libpthread.so.0
      #11 0x00007f6beff89cbd in clone () from /lib/x86_64-linux-gnu/libc.so.6
      #12 0x0000000000000000 in ?? ()

      To reproduce,

      • clone lp:~elenst/randgen/rqg-mdev8125
      • cd rqg-mdev8125
      • run

        perl ./runall-trials.pl --trials=10 --threads=8 --duration=200 --queries=100M --reporters=Backtrace,ErrorLog,ReplicationThreadRestarter --mysqld=--log_bin_trust_function_creators=1 --engine=InnoDB --grammar=mdev8125.yy --gendata=conf/runtime/metadata_stability.zz --rpl_mode=statement --mysqld=--slave-parallel-mode=optimistic --mysqld=--slave-parallel-threads=20 --basedir=<basedir> --vardir=<vardir>

      It will run the same test 10 times (or until the first crash).
      The master vardir will be <vardir>. The slave vardir will be <vardir>_slave.
      Data and logs will be in these folders, correspondingly.

      If it crashes sporadically on another reason, you can try to run the same line with --force option. In this case it will run all 10 times, and will save vardirs for the failed attempts.

      It can also hit MDEV-8113 and hang (visually, it will just stop saying anything for long time, which shouldn't normally happen because trials are fairly short, 200 seconds each plus some time to sync servers, and it's verbose enough; usually it prints complaints about semantic errors – duplicate column names, and so on; as long as it doesn't abort, it's normal).
      Currently this type of deadlock isn't recognized by the tool, so the only reliable way to get out of it is interrupt everything and start again.
      But maybe you already have a patch for the bug, it should help.


          Issue Links



              knielsen Kristian Nielsen
              elenst Elena Stepanova
              0 Vote for this issue
              1 Start watching this issue



                  Git Integration

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