Problem: "Under high load, the slave registering to the master can timeout 
during the COM_REGISTER_SLAVE execution. This causes an error, which 
prevents the slave from connecting at all."

Fix: Do not abort immediately, but retry registering on master.


sql/slave.cc:
  Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop
    - retry connecting after a failed slave registration on master.
      Reconnecting has been factored out (see try_to_reconnect() and 
      check_io_slave_killed()) to get rid of similar code used during 
      registering on master/dump requesting/event reading.
      All messages have been moved to reconnect_messages[] array 
      for easy usage and maintenance.
    - now one can force slave to reconnect using 
      ./mtr --mysqld=--loose-debug=d,FORCE_SLAVE_TO_RECONNECT_{REG, DUMP, EVENT}