There are more experienced synthetic chemists here than I am, but I am happy to throw out some ideas. If the aldehyde loses an α-H to base, the carbanion could attack the carbonyl group of a different molecule. One, have you demonstrated that you made the desired crude product prior to work-up? In discussions with a more experienced synthetic chemist, I became convinced that isolating the steps (synthesis, work-up, purification) was a good strategy. Two, are you following a literature protocol? If so, you might email the authors; in my experience, they usually reply, but they do not always know what the problem is. Three, my experience with DIBAL-H is limited to using it to reduce Weinreb amides to aldehydes. In that application, I have used either NaHSO4 or KHSO4 to work up the reaction. However, I would be hesitant to suggest this protocol unless it has been shown to work in the reduction of another lactone.