Worked on the dragonflydb/dragonfly repository to address a Kubernetes-specific issue affecting thread allocation accuracy. Implemented a fix ensuring that the proactor_threads setting is properly respected in Kubernetes environments, overriding CPU limits from cgroups when the relevant flag is enabled. The solution involved backend development and DevOps skills, utilizing C++, Python, and YAML to modify resource management logic and add comprehensive test coverage. Tests validated the new behavior for both command line flags and environment variables, improving reliability and predictability of thread allocation. This work enhanced stability and resource utilization for Kubernetes deployments, focusing on robust, test-driven engineering practices.