Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-19484

Server crashes in fix_semijoin_strategies_for_picked_join_order

Details

    • Bug
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Duplicate
    • 10.2.23
    • N/A
    • N/A
    • None

    Description

      This crash happened simultaneously on 3 nodes of a galera cluster.
      Seems related to MDEV-18467 which has been fixed in 10.3 but no mention of 10.2.
      Tried the use case from that ticket but did not crash the server.

      stack_bottom = 0x7f9d38485cf8 thread_stack 0x49000
      		 
      *** buffer overflow detected ***: /usr/sbin/mysqld terminated
      		 
      ======= Backtrace: =========
      		 
      /lib/x86_64-linux-gnu/libc.so.6(+0x70bfb)[0x7fa704e5fbfb]
      		 
      /lib/x86_64-linux-gnu/libc.so.6(__fortify_fail+0x37)[0x7fa704ee8437]
      		 
      /lib/x86_64-linux-gnu/libc.so.6(+0xf7570)[0x7fa704ee6570]
      		 
      /lib/x86_64-linux-gnu/libc.so.6(+0xf93aa)[0x7fa704ee83aa]
      		 
      /usr/sbin/mysqld(my_addr_resolve+0xe2)[0x55877900a922]
      		 
      /usr/sbin/mysqld(my_print_stacktrace+0x1bb)[0x558778ff11eb]
      		 
      /usr/sbin/mysqld(handle_fatal_signal+0x41d)[0x558778a9001d]
      		 
      /lib/x86_64-linux-gnu/libpthread.so.0(+0x110e0)[0x7fa7067820e0]
      		 
      /usr/sbin/mysqld(_Z45fix_semijoin_strategies_for_picked_join_orderP4JOIN+0x8f)[0x5587789fabcf]
      		 
      /usr/sbin/mysqld(_ZN4JOIN20get_best_combinationEv+0xc1)[0x558778922531]
      		 
      /usr/sbin/mysqld(+0x50d472)[0x55877893c472]
      		 
      /usr/sbin/mysqld(_ZN4JOIN14optimize_innerEv+0x9ab)[0x558778941f4b]
      		 
      /usr/sbin/mysqld(_ZN4JOIN8optimizeEv+0x42)[0x5587789443d2]
      		 
      /usr/sbin/mysqld(_Z12mysql_selectP3THDP10TABLE_LISTjR4ListI4ItemEPS4_jP8st_orderS9_S7_S9_yP13select_resultP18st_select_lex_unitP13st_select_lex+0x87d)[0x55877894634d]
      		 
      /usr/sbin/mysqld(_Z13handle_selectP3THDP3LEXP13select_resultm+0x135)[0x5587789464f5]
      		 
      /usr/sbin/mysqld(+0x4022bf)[0x5587788312bf]
      		 
      /usr/sbin/mysqld(_Z21mysql_execute_commandP3THD+0x3469)[0x5587788f13e9]
      		 
      /usr/sbin/mysqld(_Z11mysql_parseP3THDPcjP12Parser_statebb+0x28a)[0x5587788f6e4a]
      		 
      /usr/sbin/mysqld(+0x4c864f)[0x5587788f764f]
      		 
      /usr/sbin/mysqld(_Z16dispatch_command19enum_server_commandP3THDPcjbb+0x1a7c)[0x5587788f97fc]
      		 
      /usr/sbin/mysqld(_Z10do_commandP3THD+0x176)[0x5587788fa8a6]
      		 
      /usr/sbin/mysqld(_Z24do_handle_one_connectionP7CONNECT+0x25a)[0x5587789c4c0a]
      		 
      /usr/sbin/mysqld(handle_one_connection+0x3d)[0x5587789c4d7d]
      		 
      /lib/x86_64-linux-gnu/libpthread.so.0(+0x74a4)[0x7fa7067784a4]
      		 
      /lib/x86_64-linux-gnu/libc.so.6(clone+0x3f)[0x7fa704ed7d0f]

      Attachments

        Issue Links

          is duplicated by

          Bug - A problem which impairs or prevents the functions of the product. MDEV-13607 MariaDB crash in fix_semijoin_strategies_for_picked_join_order

          • Critical - Crashes, loss of data, severe memory leak.
          • Closed
          relates to

          Bug - A problem which impairs or prevents the functions of the product. MDEV-18479 Assertion `join->best_read < double(1.79769313486231570815e+308L)' or server crashes in JOIN::fix_all_splittings_in_plan after EXPLAIN

          • Major - Major loss of function.
          • Closed

          Activity

            Ascending order - Click to sort in descending order
            julien.fritsch Julien Fritsch added a comment - - edited

            rpizzi, I asked igor if this is possible to fix MDEV-18467 but he said "that this is a bug in the code of the feature that was added 10.3.". So it can't be fixed in 10.2. Anyway, alice is trying now to reproduce your bug. Once done, we'll know what this bug is about and the root cause. Before that, we can't say much yet.

            julien.fritsch Julien Fritsch added a comment - - edited rpizzi , I asked igor if this is possible to fix MDEV-18467 but he said "that this is a bug in the code of the feature that was added 10.3.". So it can't be fixed in 10.2. Anyway, alice is trying now to reproduce your bug. Once done, we'll know what this bug is about and the root cause. Before that, we can't say much yet.
            alice Alice Sherepa added a comment - - edited

            julien.fritsch, I am not able to reproduce this bug as there is no data about crashing query,..
            There is a big chance that it is the same bug as MDEV-13607 (and MDEV-18479, derived from MDEV-13607), it produces the same stacktrace on 10.2.23. But we can not garantee that it is a duplicate as we do not have the reproducible test case for it.

            alice Alice Sherepa added a comment - - edited julien.fritsch , I am not able to reproduce this bug as there is no data about crashing query,.. There is a big chance that it is the same bug as MDEV-13607 (and MDEV-18479 , derived from MDEV-13607 ), it produces the same stacktrace on 10.2.23. But we can not garantee that it is a duplicate as we do not have the reproducible test case for it.

            People

              Unassigned Unassigned
              rpizzi Rick Pizzi (Inactive)
              Votes:
              1 Vote for this issue
              Watchers:
              6 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.