Docker Tensorflow-Serving Predictions too large

最近在使用TensorFlow Serving的时候，如果图片较大，会出现如下的报错
“grpc_message”:“Received message larger than max (14830491 vs. 4194304)”

经过查询，是因为grpc 默认只能传输4M的文件，而我们超过的这个阈值，所以要设置这个阈值

MAX_MESSAGE_LENGTH = 429496729   #传输最大设置

class Grpc_Reco():
    '通过GRPC的方式调用方向检测服务'
    def __init__(self):
        channel = grpc.insecure_channel("172.1.0.1:8000",
options=[('grpc.max_send_message_length', MAX_MESSAGE_LENGTH),  #加入options参数
('grpc.max_receive_message_length', MAX_MESSAGE_LENGTH)])        
        # print(channel)
        self.stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
        self.request = predict_pb2.PredictRequest()
        self.request.model_spec.name = "mymodel"  # 模型名称
        self.request.model_spec.signature_name = "predict_images"  # 签名名称
        print ('grpc is inited')