support change limit sizes of messages

python/paddle_serving_client/__init__.py

@@ -111,7 +111,7 @@ class Client(object):
         self.result_handle_ = PredictorRes()
         self.client_handle_ = PredictorClient()
         self.client_handle_.init(path)
-        read_env_flags = ["profile_client", "profile_server"]
+        read_env_flags = ["profile_client", "profile_server", "max_body_size"]
         self.client_handle_.init_gflags([sys.argv[
             0]] + ["--tryfromenv=" + ",".join(read_env_flags)])
         self.feed_names_ = [var.alias_name for var in model_conf.feed_var]

python/paddle_serving_server/__init__.py

@@ -89,6 +89,7 @@ class Server(object):
         self.num_threads = 4
         self.port = 8080
         self.reload_interval_s = 10
+        self.max_body_size = 64 * 1024 * 1024
         self.module_path = os.path.dirname(paddle_serving_server.__file__)
         self.cur_path = os.getcwd()
         self.use_local_bin = False
@@ -100,6 +101,14 @@ class Server(object):
     def set_num_threads(self, threads):
         self.num_threads = threads
 
+    def set_max_body_size(self, body_size):
+        if body_size >= self.max_body_size:
+            self.max_body_size = body_size
+        else:
+            print(
+                "max_body_size is less than default value, will use default value in service."
+            )
+
     def set_port(self, port):
         self.port = port
 
@@ -292,7 +301,8 @@ class Server(object):
                   "-resource_file {} " \
                   "-workflow_path {} " \
                   "-workflow_file {} " \
-                  "-bthread_concurrency {} ".format(
+                  "-bthread_concurrency {} " \
+                  "-max_body_size {} ".format(
                       self.bin_path,
                       self.workdir,
                       self.infer_service_fn,
@@ -304,7 +314,8 @@ class Server(object):
                       self.resource_fn,
                       self.workdir,
                       self.workflow_fn,
-                      self.num_threads)
+                      self.num_threads,
+                      self.max_body_size)
         print("Going to Run Command")
         print(command)
         os.system(command)

python/paddle_serving_server/serve.py

@@ -41,6 +41,11 @@ def parse_args():  # pylint: disable=doc-string-missing
         "--device", type=str, default="cpu", help="Type of device")
     parser.add_argument(
         "--mem_optim", type=bool, default=False, help="Memory optimize")
+    parser.add_argument(
+        "--max_body_size",
+        type=int,
+        default=64 * 1024 * 1024,
+        help="Limit sizes of messages")
     return parser.parse_args()
 
 
@@ -52,6 +57,7 @@ def start_standard_model():  # pylint: disable=doc-string-missing
     workdir = args.workdir
     device = args.device
     mem_optim = args.mem_optim
+    max_body_size = args.max_body_size
 
     if model == "":
         print("You must specify your serving model")
@@ -72,6 +78,7 @@ def start_standard_model():  # pylint: disable=doc-string-missing
     server.set_op_sequence(op_seq_maker.get_op_sequence())
     server.set_num_threads(thread_num)
     server.set_memory_optimize(mem_optim)
+    server.set_max_body_size(max_body_size)
 
     server.load_model_config(model)
     server.prepare_server(workdir=workdir, port=port, device=device)

python/paddle_serving_server_gpu/__init__.py

@@ -46,6 +46,11 @@ def serve_args():
         "--name", type=str, default="None", help="Default service name")
     parser.add_argument(
         "--mem_optim", type=bool, default=False, help="Memory optimize")
+    parser.add_argument(
+        "--max_body_size",
+        type=int,
+        default=64 * 1024 * 1024,
+        help="Limit sizes of messages")
     return parser.parse_args()
 
 
@@ -114,6 +119,7 @@ class Server(object):
         self.num_threads = 4
         self.port = 8080
         self.reload_interval_s = 10
+        self.max_body_size = 64 * 1024 * 1024
         self.module_path = os.path.dirname(paddle_serving_server.__file__)
         self.cur_path = os.getcwd()
         self.check_cuda()
@@ -126,6 +132,14 @@ class Server(object):
     def set_num_threads(self, threads):
         self.num_threads = threads
 
+    def set_max_body_size(self, body_size):
+        if body_size >= self.max_body_size:
+            self.max_body_size = body_size
+        else:
+            print(
+                "max_body_size is less than default value, will use default value in service."
+            )
+
     def set_port(self, port):
         self.port = port
 
@@ -324,7 +338,8 @@ class Server(object):
                   "-workflow_path {} " \
                   "-workflow_file {} " \
                   "-bthread_concurrency {} " \
-                  "-gpuid {} ".format(
+                  "-gpuid {} " \
+                  "-max_body_size {} ".format(
                       self.bin_path,
                       self.workdir,
                       self.infer_service_fn,
@@ -337,7 +352,8 @@ class Server(object):
                       self.workdir,
                       self.workflow_fn,
                       self.num_threads,
-                      self.gpuid,)
+                      self.gpuid,
+                      self.max_body_size)
         print("Going to Run Comand")
         print(command)
 

python/paddle_serving_server_gpu/serve.py

@@ -35,6 +35,7 @@ def start_gpu_card_model(index, gpuid, args):  # pylint: disable=doc-string-miss
     thread_num = args.thread
     model = args.model
     mem_optim = args.mem_optim
+    max_body_size = args.max_body_size
     workdir = "{}_{}".format(args.workdir, gpuid)
 
     if model == "":
@@ -56,6 +57,7 @@ def start_gpu_card_model(index, gpuid, args):  # pylint: disable=doc-string-miss
     server.set_op_sequence(op_seq_maker.get_op_sequence())
     server.set_num_threads(thread_num)
     server.set_memory_optimize(mem_optim)
+    server.set_max_body_size(max_body_size)
 
     server.load_model_config(model)
     server.prepare_server(workdir=workdir, port=port, device=device)