Yes and no. If there's a network glitch while we're polling to see if the instance is "running", it will result in the "pending-to-running" time being high but the "running-to-port-closed" time being low.