Cache Invalidation Strategies, Stale Data Versus Complexity

hey everyone, I'm wrestling with cache invalidation in a recent project. We're using Redis for caching API responses, and while it's helped read performance a lot, keeping the cache updated with database changes is proving to be a real headache. We've looked at time-based expiration, but that can lead to stale data. Event-driven invalidation sounds good but adds so much complexity. How do you balance the need for fresh data with the engineering effort required for cache invalidation in your systems? What are your go-to strategies and any pitfalls to watch out for