unify starting local cluster with attaching to existing cluster (#327)

examples/alexnet/driver.py

@@ -15,7 +15,7 @@ parser.add_argument("--label-file", default="train.txt", type=str, help="File co
 if __name__ == "__main__":
   args = parser.parse_args()
   num_workers = 4
-  ray.services.start_ray_local(num_workers=num_workers)
+  ray.init(start_ray_local=True, num_workers=num_workers)
 
   # Note we do not do sess.run(tf.initialize_all_variables()) because that would
   # result in a different initialization on each worker. Instead, we initialize

examples/hyperopt/driver.py

@@ -10,7 +10,7 @@ from tensorflow.examples.tutorials.mnist import input_data
 import hyperopt
 
 if __name__ == "__main__":
-  ray.services.start_ray_local(num_workers=3)
+  ray.init(start_ray_local=True, num_workers=3)
 
   # The number of sets of random hyperparameters to try.
   trials = 2

examples/lbfgs/driver.py

@@ -8,7 +8,7 @@ import tensorflow as tf
 from tensorflow.examples.tutorials.mnist import input_data
 
 if __name__ == "__main__":
-  ray.services.start_ray_local(num_workers=16)
+  ray.init(start_ray_local=True, num_workers=16)
 
   # Define the dimensions of the data and of the model.
   image_dimension = 784

examples/rl_pong/driver.py

@@ -108,7 +108,7 @@ def compute_gradient(model):
   return policy_backward(eph, epx, epdlogp, model), reward_sum
 
 if __name__ == "__main__":
-  ray.services.start_ray_local(num_workers=10)
+  ray.init(start_ray_local=True, num_workers=10)
 
   # Run the reinforcement learning
   running_reward = None