How do you catch schema drift before it breaks your production database?

There is a mini heart attack moment when you realize your code review says one thing, your migration history says another, and the actual live database is doing something completely different. A single silent failure in a database trigger can take down your whole sign up flow with a vague error message that’s impossible to debug on launch day.

Keeping code and production in sync feels like a constant juggling act. What’s your go to method for keeping schemas honest automated diffs in the pipeline or just strict pre deployment checks?