Strategies for managing feature flags in large-scale distributed systems?

Hey everyone, I'm curious to hear about your experiences and best practices for managing feature flags in large-scale distributed systems. We're growing and our current home-grown solution is starting to show its limits. Specifically, what approaches have you found effective for managing flag definitions, targeting rules, and ensuring consistency across microservices? Are there particular tools or patterns that have worked well for rollouts, A/B testing, and kill switches? Also, how do you handle the lifecycle of flags, from creation to eventual deprecation and cleanup? Any war stories or successful strategies would be much appreciated!