From: Nuno Carvalho
Date: August 21 2012 10:33am
Subject: bzr push into mysql-5.6 branch (nuno.carvalho:4136 to 4137) Bug#14116526
List-Archive: http://lists.mysql.com/commits/144588
X-Bug: 14116526
Message-Id: <20120821103306.31806.50605.4137@localhost>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

 4137 Nuno Carvalho	2012-08-21
      BUG#14116526: RPL.RPL_GTID_SQL_UNTIL_BEFORE_AFTER FAILS WITH RESULT LENGTH MISMATCH ON TRUNK.
      
      rpl_gtid_sql_until_before_after has sporadic failures on pb2 on valgrind
      runs. On slow/overloaded platforms SQL_THREAD_WAIT_AFTER_GTIDS may
      timeout before reaching the condition, which query result is not being
      logged due to nondeterministic return value.
      
      To avoid SQL_THREAD_WAIT_AFTER_GTIDS returns before reaching the
      condition, disabled timeout from SQL_THREAD_WAIT_AFTER_GTIDS query.

    modified:
      mysql-test/suite/rpl/r/rpl_gtid_sql_until_before_after.result
      mysql-test/suite/rpl/t/rpl_gtid_sql_until_before_after.test
 4136 Manish Kumar	2012-08-21
      BUG#14381701 - DELETING FROM A HASH WHILE ITERATING OVER IT IS WRONG
      
      Problem - The problem is that in sql/rpl_rli_pdb.cc file we delete
                from a hash while iterating over it. This is wrong since 
                delete operation might rearrange elements in hash, 
                so "iterator" will become invalid and, unless extra steps
                are taken, such iteration is likely to miss elements which 
                should have been deleted otherwise.
      
      Fix - The problem is fixed by storing the address of the Hash elements 
            that needs to be deleted based on the usage, in the Dynamic array
            of pointers in a for loop. In a second iteration we pop the 
            elements of the dynamic array and delete the Hash element from 
            that address. This will not cause any inconsistency.
     @ sql/rpl_rli_pdb.cc
        Added the new Dynamic array " hash_element "  which is used to store the
        hash entries that need to be deleted.

    modified:
      sql/rpl_rli_pdb.cc
=== modified file 'mysql-test/suite/rpl/r/rpl_gtid_sql_until_before_after.result'
--- a/mysql-test/suite/rpl/r/rpl_gtid_sql_until_before_after.result	2012-05-08 17:11:38 +0000
+++ b/mysql-test/suite/rpl/r/rpl_gtid_sql_until_before_after.result	2012-08-21 10:27:20 +0000
@@ -111,7 +111,7 @@ c1
 9
 Sync slave with master
 include/start_slave.inc
-SELECT SQL_THREAD_WAIT_AFTER_GTIDS('MASTER_UUID:10', 300);
+SELECT SQL_THREAD_WAIT_AFTER_GTIDS('MASTER_UUID:10');
 SELECT * FROM t1;
 c1
 2

=== modified file 'mysql-test/suite/rpl/t/rpl_gtid_sql_until_before_after.test'
--- a/mysql-test/suite/rpl/t/rpl_gtid_sql_until_before_after.test	2012-05-08 17:11:38 +0000
+++ b/mysql-test/suite/rpl/t/rpl_gtid_sql_until_before_after.test	2012-08-21 10:27:20 +0000
@@ -139,7 +139,7 @@ SELECT * FROM t1;
 --replace_result $master_uuid MASTER_UUID
 --enable_query_log
 --disable_result_log
---eval SELECT SQL_THREAD_WAIT_AFTER_GTIDS('$master_uuid:10', 300)
+--eval SELECT SQL_THREAD_WAIT_AFTER_GTIDS('$master_uuid:10')
 --enable_result_log
 SELECT * FROM t1;
 

No bundle (reason: useless for push emails).