• unknown's avatar
    BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 · 642eda22
    unknown authored
    using TPC-B):
     
    Problem: A RBR event can contain incomplete row data (only key value and
    fields which have been changed). In that case, when the row is unpacked
    into record and written to a table, the missing fields get incorrect NULL
    values leading to master-slave inconsistency.
     
    Solution: Use values found in slave's table for columns which are not given
    in the rows event. The code for writing a single row uses the following 
    algorithm: 
    
    1. unpack row_data into table->record[0],
    2. try to insert record,
    3. if duplicate record found, fetch it into table->record[0],
    4. unpack row_data into table->record[0],
    5. write table->record[0] into the table.
    
    Where row_data is the row as stored in the data area of a rows event. 
    Thus:
    
    a) unpacking of row_data happens at the time when row is written into 
     a table,
    
    b) when unpacking (in step 4), only columns present in row_data are 
     overwritten - all other columns remain as they were found in the table.
     
    Since all data needed for the above algorithm is stored inside 
    Rows_log_event class, functions which locate and write rows are turned 
    into methods of that class.
    
    replace_record()     -> Rows_log_event::write_row()
    find_and_fetch_row() -> Rows_log_event::find_row()
    
    Both methods take row data from event's data buffer - the row being 
    processed is pointed by m_curr_row. They unpack the data as needed into 
    table's record buffers record[0] or record[1]. When row is unpacked, 
    m_curr_row_end is set to point at next row in the data buffer.
    
    Other changes introduced in this changeset:
    
    - Change signature of unpack_row(): don't report errors and don't
    setup table's rw_set here. Errors can happen only when setting default 
    values in prepare_record() function and are detected there.
     
    - In Rows_log_event and derived classes, don't pass arguments to
    the execution primitives (do_...() member functions) but use class
    members instead.
    
    - Move old row handling code into log_event_old.cc to be used by 
    *_rows_log_event_old classes.
    
    Also, a new test rpl_ndb_2other is added which tests basic replication 
    from master using ndb tables to slave storing the same tables using 
    (possibly) different engine (myisam,innodb).
      
    Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. 
    However, these tests doesn't work for various reasons and currently are 
    disabled (see BUG#19227).
      
    The new test differs from the ones it is based on as follows:
      
    1. Single test tests replication with different storage engines on slave 
    (myisam, innodb, ndb).
      
    2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing 
    original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test 
    which doesn't contain tests using partitioned tables as these don't work 
    currently. Instead, it tests replication to a slave which has more or 
    less columns than master.
      
    3. Include file include/rpl_multi_engine3.inc is replaced with 
    include/rpl_multi_engine2.inc. The later differs by performing slightly 
    different operations (updating more than one row in the table) and 
    clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" 
    as replication of "DELETE" doesn't work well in this setting.
      
    4. Slave must use option --log-slave-updates=0 as otherwise execution of 
    replication events generated by ndb fails if table uses a different 
    storage engine on slave (see BUG#29569).
    
    
    sql/log_event.cc:
      - Initialization of new Rows_log_event members.
      - Fixing some typos in documentation.
      
      In Rows_log_event::do_apply_event:
      - Set COMPLETE_ROWS_F flag (when master and slave have the same number of 
      columns and all colums are present in the row)
      - Move initialization of tables write/read sets here, outside the rows
      processing loop (and out of unpack_row() function).
      - Remove calls to do_prepare_row() - no longer needed.
      - Add code managing m_curr_row and m_curr_row_end pointers.
      
      - Change signatures of row processing methods of Rows_log_event and it
      descendants - now most arguments are taken from class members.
      - Remove do_prepare_row() methods which are no longer used.
      - The auto_afree_ptr template is moved to rpl_utility.h (so that it can
      be used in log_event_old.cc).
      - Removed copy_extra_fields() function - no longer used.
      
      In Rows_log_event::write_row (former replace_record):
      - The old code is moved to log_event_old.cc.
      - Use prepare_record() and non-destructive unpack_current_row() to fill record
      with data.
      - In case a record being inserted already exists on slave and row data is 
      incomplete use the record found and non-destructive unpack_current_row() to 
      combine new column values with existing ones.
      - More debug info added.
      
      In Rows_log_event::find_row (former find_and_fetch_row function):
      - The old code is moved to log_event_old.cc.
      - Unpacking of the row is moved here.
      - In case of search using PK, the key data is prepared here.
      - More debug info added.
      
      - Remove initialization of Rows_log_event::m_after_image buffer which is no
      longer used. 
      - Use new row unpacking methods in Update_rows_log_event::do_exec_row() to 
      create before and after image.
      
      Note: all existing code used by Rows_log_event::do_apply_event() has been moved
      to log_event_old.cc to be used by *_rows_log_event_old classes.
    sql/log_event.h:
      - Add new COMPLETE_ROWS_F flag in Rows_log_event.
      - Add Rows_log_event members describing the row being processed.
      - Add a pointer to key buffer which is used in derived classes.
      - Add new methods: find__row(), write_row() and unpack_current_row().
      - Change signatures of do_...() methods (replace method arguments by
      class members).
      - Remove do_prepare_row() method which is no longer used.
      - Update method documentation.
      - Add Old_rows_log_event class, which contains the old row processing code, as
      a friend of Rows_log_event so that it can access all members of an event 
      instance.
    sql/log_event_old.cc:
      Move here old implementation of Rows_log_event::do_apply_event() and 
      helper methods.
    sql/log_event_old.h:
      - Define new class Old_rows_log_event encapsulating old version of
      Rows_log_event::do_apply_event() and the helper methods.
      - Add the Old_rows_log_event class as a base for *_old versions of RBR event
      classes, ensure that the old version of do_apply_event() is called.
      - For *_old classes, declare the helper methods used in the old version of
      do_apply_event().
    sql/rpl_record.cc:
      - Make unpack_row non-destructive for columns not present in the row.
      - Don't fill read/write set here as it is done outside these functions.
      - Move initialization of a record with default values to a separate
      function prepare_record().
    sql/rpl_record.h:
      - Change signature of unpack_row().
      - Declare function prepare_record().
    sql/rpl_utility.cc:
      Make tabe_def::calc_field_size() a const method.
    sql/rpl_utility.h:
      Make table_def::calc_field_size() a const method.
      
      Move auto_afree_ptr template here so that it can be re-used (currently
      in log_event.cc and log_event_old.cc). Similar with DBUG_PRINT_BITSET 
      macro.
    mysql-test/extra/rpl_tests/rpl_ndb_2multi_basic.test:
      Modification of rpl_ndb_2multi_eng test. Tests with partitioned tables 
      are removed and a setup with slave having different number of columns 
      than master is added.
    mysql-test/include/rpl_multi_engine2.inc:
      Modification of rpl_multi_engine3.inc which operates on more rows and
      replaces "DELETE FROM t1" with "TRUNCATE TABLE t1" as the first form
      doesn't replicate in NDB -> non-NDB setting (BUG#28538).
    mysql-test/suite/rpl_ndb/r/rpl_ndb_2other.result:
      Results of the test.
    mysql-test/suite/rpl_ndb/t/rpl_ndb_2other-slave.opt:
      Test options. --log-slave-updates=0 is compulsory as otherwise non-NDB 
      slave applying row events from NDB master will fail when trying to log
      them.
    mysql-test/suite/rpl_ndb/t/rpl_ndb_2other.test:
      Test replication of NDB table to slave using other engine. The main test
      is in extra/rpl_tests/rpl_ndb_2multi_basic.test. It is included here
      several times with different settings of default storage engine on slave.
    642eda22
rpl_ndb_2multi_basic.test 3.42 KB