Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-583

LP:884728 - [ERROR] mysqld: Aria engine: log initialization failed

    Details

    • Type: Bug
    • Status: Closed
    • Priority: Minor
    • Resolution: Duplicate
    • Affects Version/s: 10.0.0, 5.5.28, 5.3.9
    • Fix Version/s: None
    • Component/s: None
    • Labels:

      Description

      MariaDB failed to start on one of our systems:

      Nov  1 13:35:03 server1 mysqld_safe: Starting mysqld daemon with databases from /var/lib/mysql
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [Warning] 'master-host' is deprecated and will be removed in a future release. Please use ''CHANGE MASTER'' instead.
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [Note] Plugin 'BLACKHOLE' is disabled.
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [Note] Plugin 'CSV' is disabled.
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [Note] Plugin 'FEDERATED' is disabled.
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [ERROR] mysqld: Aria engine: log initialization failed
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [ERROR] Plugin 'Aria' init function returned error.
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [ERROR] Plugin 'Aria' registration as a STORAGE ENGINE failed.
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [ERROR] Failed to initialize plugins.
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [ERROR] Aborting
      Nov  1 13:35:03 server1 mysqld: 
      Nov  1 13:35:03 server1 mysqld: 111101 13:35:03 [Note] /usr/sbin/mysqld: Shutdown complete
      Nov  1 13:35:03 server1 mysqld: 
      Nov  1 13:35:03 server1 mysqld_safe: mysqld from pid file /var/run/mysqld/mysqld.pid ended
      

      Removing /var/lib/mysql/aria* allowed MariaDB to start again. I'll upload the corrupt Aria files to FTP (we verified on another system that it indeed fails to start with those files).

        Gliffy Diagrams

          Attachments

            Activity

            Hide
            dreasvandonselaar Dreas van Donselaar added a comment -

            Re: [ERROR] mysqld: Aria engine: log initialization failed
            Exactly the same Aria bug occurred on another server after a storage issue. Removing the aria_log_control file allowed MariaDB 5.2.9 to start again. I'll upload another collection of data files to FTP.

            Show
            dreasvandonselaar Dreas van Donselaar added a comment - Re: [ERROR] mysqld: Aria engine: log initialization failed Exactly the same Aria bug occurred on another server after a storage issue. Removing the aria_log_control file allowed MariaDB 5.2.9 to start again. I'll upload another collection of data files to FTP.
            Hide
            ratzpo Rasmus Johansson added a comment -

            Launchpad bug id: 884728

            Show
            ratzpo Rasmus Johansson added a comment - Launchpad bug id: 884728
            Hide
            elenst Elena Stepanova added a comment -

            The problem can still be seen while trying to read logs:

            • unpack launchpad_884728_by_Dreas_example_2.tgz,
            • cd to the created 'aria' folder,
            • run aria_read_log.
              A non-debug version fails with "Aria engine: log initialization failed .. Can't init loghandler (0)"; a debug version fails with "translog_init_with_table: Assertion `sure_page <= last_page' failed".

            However, there is no indication of how the corrupted logs were produced. Knowing the history, chances are that it happened due to a power failure or a hardware failure. I suppose if we want to do anything at all about the bug, it's to check that the failure is really caused by the log being corrupted and not by the contents.

            Show
            elenst Elena Stepanova added a comment - The problem can still be seen while trying to read logs: unpack launchpad_884728_by_Dreas_example_2.tgz, cd to the created 'aria' folder, run aria_read_log. A non-debug version fails with "Aria engine: log initialization failed .. Can't init loghandler (0)"; a debug version fails with "translog_init_with_table: Assertion `sure_page <= last_page' failed". However, there is no indication of how the corrupted logs were produced. Knowing the history, chances are that it happened due to a power failure or a hardware failure. I suppose if we want to do anything at all about the bug, it's to check that the failure is really caused by the log being corrupted and not by the contents.
            Hide
            elenst Elena Stepanova added a comment -

            I had the exact same assertion failure reproducible while running out of space during recovery test, which caused a problem with copying the server datadir and created inconsistent data for recovery. Thus, I assume the guess about initially corrupted datadir was correct, so I'm closing the bug as a duplicate MDEV-727 (as the latter has a test case and a longer explanation of how it all happened); but MDEV-727 was also closed as a duplicate, since at the end it was a consequence of a failure which made the server run out of space in the first place.

            Show
            elenst Elena Stepanova added a comment - I had the exact same assertion failure reproducible while running out of space during recovery test, which caused a problem with copying the server datadir and created inconsistent data for recovery. Thus, I assume the guess about initially corrupted datadir was correct, so I'm closing the bug as a duplicate MDEV-727 (as the latter has a test case and a longer explanation of how it all happened); but MDEV-727 was also closed as a duplicate, since at the end it was a consequence of a failure which made the server run out of space in the first place.

              People

              • Assignee:
                monty Michael Widenius
                Reporter:
                dreasvandonselaar Dreas van Donselaar
              • Votes:
                0 Vote for this issue
                Watchers:
                1 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: