It depends on what DB, but it's basically the following to "ignore" it,

INSERT
  INTO records (PriKey, Foo, Bar)
  VALUES (1, 'Hello', World')
  ON DUPLICATE KEY
    UPDATE Foo=Foo;
[download]

Some track the attempts, though; using the equivalent to a column named Attempts. But I am not sure how useful that is in practice tbh:

INSERT
  INTO records (PriKey, Foo, Bar)
  VALUES (1, 'Hello', World')
  ON DUPLICATE KEY
    UPDATE Attempts=Attempts+1;
[download]

> Since I can choose the DB, what is your opinion on MySQL vs. Postgres?

It really depends; frankly I don't do anything these days that even requires a DB server so usually I go with DBD::SQLite. That has it's own things to deal with (forks, etc) but it makes my life a lot easier not to have to maintain DB servers and all that stuff. YMMV based on the application though.

Most relevant from the YMMV link,

SQLite does not compete with client/server databases. SQLite competes with fopen().