pt-table-checksum will set binlog_format to statement for it's own session.
As for data drift, there are many causes for it but with statement
based replication for your general replication configuration leaves
you open to non-deterministic functions causing diffs throughout your
On 9/11/13, Rafał Radecki <radecki.rafal@stripped> wrote:
> Hi All.
> I use binlog_format = row for my production databases. In this format most
> binlog_format changes are sent not as SQL statements but in some other
> I understand that when binlog_format = statement is used, queries on master
> and slave can give different results, but should pt-table-cheksum be used
> in situation when we use binlog_format = row?
> I've seen opinions that regardles of binlog_format the data on slave and
> master may differ. What are your experiences?
> Best regards,