There are two issues: 
      1. 6.0 uses the obsolate master-*** server options;
      2. the test is not deterministic in that although master vs slave consistency is
         fine, two runs of the test can have different results. The reason of the
         non-determinism is the combination of 
         a chosen way to demo results and the ndb_autoincrement_prefetch_sz feature.

The current patch fixes the 2nd issue by putting out results via diff_table macro
instead of the former run-sensitive method.
The 1st issue is going to be fixed by a separate patch to 6.0.

mysql-test/suite/rpl_ndb/r/rpl_ndb_circular_2ch.result:
  results changed
mysql-test/suite/rpl_ndb/t/rpl_ndb_circular_2ch.test:
  Test is refined to put out results via diff_table macro instead of the former run-sensitive
  method.