1

We had to debug our web application for an event that occurred some days ago, and all we have are the MySQL replication binlogs (statement replication). Can we depend on those as being exactly the same SQL queries as our web application was executing on master?

Our application is doing a lot of updates on single row in one table, but those are always queries like UPDATE y SET x = x + 1 ... or x = x - 2, never UPDATE y SET x = 23 ..., and in the binlog we have found many x = 23 kind of updates. Can it be that those strictly assigning UPDATE queries are in fact created by replication mechanism?

Michał Bielecki
  • 66
  • 4
  • That is strange. Do you also see `x=x+1` updates in the binlog? – Rick James May 02 '15 at 18:11
  • Yeah, `x=x+1` were also there. Actually application was doing those `SET x=23` kind of updates, but under certain conditions. So answering my own question: queries that you will find in binlog are the same that are being executed on master database, but you will only find insert/update queries there (no selects obviously), and their order might be different (you need to look at timestamps in comments in binlog). – Michał Bielecki May 14 '15 at 14:24

1 Answers1

0

Actually application was doing those SET x=23 kind of updates, but under certain conditions.

So answering my own question: queries that you will find in binlog are the same that are being executed on master database, but you will only find insert/update queries there (no selects obviously), and their order might be different (you need to look at timestamps in comments in binlog).

Michał Bielecki
  • 66
  • 4