The wait loop added in16c0107probes `docker version` but does not attempt to start the service if it is stopped, leading to 30 failed retries and a timeout when the runner's pre-installed Docker service has not started automatically. Add service status diagnostics and attempt to start the Docker service on each retry if it exists but is not running. Also redirect stderr to null so `docker version` error messages don't flood the log. Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com> (cherry picked from commit0741f195aa) Signed-off-by: Cory Snider <csnider@mirantis.com>