This is a bit dangerous isn't it? What happens when a process accidentally updates one of the columns and not the other?
It would be safer/easier to write a function (either in perl or the db) to convert between the two formats, and only store one of them. I've not used mysql much, but unix_timestamp() might be useful there.
Regarding the original problem - getting thousands of lines into a project before discovering bugs like this is a good example of why writing tests as you go (or before you go in some cases) is such a sanity-saver. :)