def echo_v1(_):
return "v1"
def echo_v2(_):
return "v2"
# specify the router policy as RoundRobin
serve.init(blocking=True, queueing_policy=serve.RoutePolicy.RoundRobin)
# create a service
serve.create_endpoint("my_endpoint", "/echo", blocking=True)
# create first backend
serve.create_backend(echo_v1, "echo:v1")
# create second backend
serve.create_backend(echo_v2, "echo:v2")
# link and split the service to two backends
serve.split("my_endpoint", {"echo:v1": 0.5, "echo:v2": 0.5})
while True:
resp = requests.get("http://127.0.0.1:8000/echo").json()
print(pformat_color_json(resp))
print("...Sleeping for 2 seconds...")
time.sleep(2)
def split(endpoint, traffic):
ray.init(address="auto")
serve.init()
serve.split(endpoint, json.loads(traffic))