def test_v2_async_mode():
# model loading is async
os.environ["SERVING_SPEC_ENV"] = json.dumps(asyncspec)
context = GraphContext()
nuclio_init_hook(context, globals(), serving_subkind)
context.logger.info("model initialized")
context.logger.info("test not ready, should return err 408")
event = MockEvent("", path="/v2/models/m5/ready", method="GET")
resp = context.mlrun_handler(context, event)
assert (
resp.status_code == 408
), f"didnt get proper ready resp, expected 408, got {resp.status_code}"
event = MockEvent(testdata, path="/v2/models/m5/infer")
resp = context.mlrun_handler(context, event)
context.logger.info("model responded")
logger.info(resp)
assert (resp.status_code !=
200), f"expected failure, got {resp.status_code} {resp.body}"
event = MockEvent('{"model": "m5", "inputs": [5]}')
event.trigger = "stream"
resp = context.mlrun_handler(context, event)
context.logger.info("model responded")
logger.info(resp)
data = json.loads(resp.body)
assert data["outputs"] == 5, f"wrong model response {data}"
def test_v2_raised_err():
os.environ["SERVING_SPEC_ENV"] = json.dumps(raiser_spec)
context = GraphContext()
nuclio_init_hook(context, globals(), serving_subkind)
event = MockEvent(testdata, path="/v2/models/m6/infer")
resp = context.mlrun_handler(context, event)
context.logger.info(f"model responded: {resp}")
assert resp.status_code == 400, "expecting Response() with status_code 400"
def init_context(context):
nuclio_init_hook(context, globals(), "serving_v2")
def init_ctx(spec=spec, context=None):
os.environ["SERVING_SPEC_ENV"] = json.dumps(spec)
context = context or GraphContext()
nuclio_init_hook(context, globals(), serving_subkind)
return context
def init_ctx():
os.environ["SERVING_SPEC_ENV"] = json.dumps(spec)
context = MockContext()
nuclio_init_hook(context, globals(), serving_subkind)
return context
