MySQL binary log reporting execution time as UINT_MAX

Question

Looking through the MySQL Binlogs, I see some data that just does not make sense. There are queries that say they have been running for 4294967295 seconds (which also happens to be UINT_MAX on this system).

The queries that are reported with these execution times vary, but they are typically updates to small tables. There is no way that these queries have been running for that long.

Percona has a blog stating that the exec_time value can be different than one would expect if the server is a MySQL replication slave, but this is not the case.

Are these strange values for exec_time in the binary log expected, or is this a bug in MySQL?

Answer 1

The unsigned value 4294967295 has the same binary representation as signed -1. So I suggest that it is reporting "-1 seconds".

Either your machines are not in sync on time, or it is a bug in MySQL.

I have several seen other cases where "the clock ran backwards", even on machines that are synced with the time service.

Consider finding or filing a bug with bugs.mysql.com