Comment 5 for bug 982381

Work around for getting a table sync issue when super privileges are not available and not syncing between a master and a slave. This code change adds an option called "--no-bin-log-format" to tell pt-table-sync that is is okay to avoid using "SET @@binlog_format=STATEMENT" .

This addresses only one of many possible places where the super privilege is needed

In pt-table-sync there is original code looking like this (I intentionally did not provide diff/patch formatted changes, since these are changes to the distributed perl code):

sub get_cxn {
...
   # Ensure statement-based replication.
   # http://code.google.com/p/maatkit/issues/detail?id=95
   $sql = '/*!50105 SET @@binlog_format="STATEMENT"*/';
   PTDEBUG && _d($dbh, $sql);
   $dbh->do($sql);

The changed code is:

   # Ensure statement-based replication.
   # http://code.google.com/p/maatkit/issues/detail?id=95
   if ( $o->get('bin-log-format') ) {
       $sql = '/*!50105 SET @@binlog_format="STATEMENT"*/';
       PTDEBUG && _d($dbh, $sql);
       $dbh->do($sql);
   }

which requires that the bin-log-format option be added to the POD section, So I added this:

original:

=item --[no]bin-log

default: yes

Log to the binary log (C<SET SQL_LOG_BIN=1>).

Specifying C<--no-bin-log> will C<SET SQL_LOG_BIN=0>.

=item --buffer-in-mysql

Changed code:

=item --[no]bin-log

default: yes

Log to the binary log (C<SET SQL_LOG_BIN=1>).

Specifying C<--no-bin-log> will C<SET SQL_LOG_BIN=0>.

=item --[no]bin-log-format

default: yes

Force binlog_format to STATEMENT (C<SET @@binlog_format=STATEMENT>).

Specifying C<--no-bin-log> will prevent binlog_format modification.

=item --buffer-in-mysql

I'd like to understand the percona toolkit build and tests better so that I can offer suggestions for code enhancements (especially for making tools work with systems like Amazon's RDS where super privilege will never be permitted). Is there a percona toolkit developer "getting started guide" that I may have missed? I understand only Percona employees will make changes to the official code, so I my purpose is to provide helpful and working code suggestions that are easy to read, test, and consider for code change candidates.