Cause: The openshift-router process was ignoring the SIGTERM shutdown signal for a brief moment as it was starting. Consequence: If a SIGTERM was sent while the router process was starting up, it would ignore the signal. This means the container would ignore a kubernetes shutdown request resulting in the container taking 1 hour to shutdown (terminationGracePeriodSeconds). Fix: Propagate the SIGTERM handler in GO code to the cache initialization function. Result: The router now responds to SIGTERM signals during it initialization.