Use ipc instead of tcp in zmq (#1566)

examples/frontend_language/usage/llava_video/srt_example_llava_v.py

@@ -223,7 +223,6 @@ if __name__ == "__main__":
         model_path=args.model_path,  # "liuhaotian/llava-v1.6-vicuna-7b",
         tokenizer_path=tokenizer_path,
         port=cur_port,
-        additional_ports=[cur_port + 1, cur_port + 2, cur_port + 3, cur_port + 4],
         json_model_override_args=json.dumps(model_override_args),
         tp_size=1,
     )

python/sglang/bench_latency.py

@@ -66,9 +66,8 @@ from sglang.srt.model_executor.forward_batch_info import ForwardBatch
 from sglang.srt.model_executor.model_runner import ModelRunner
 from sglang.srt.sampling.sampling_params import SamplingParams
 from sglang.srt.server import _set_envs_and_config
-from sglang.srt.server_args import ServerArgs
+from sglang.srt.server_args import PortArgs, ServerArgs
 from sglang.srt.utils import (
-    allocate_init_ports,
     configure_logger,
     kill_child_process,
     suppress_other_loggers,
@@ -127,11 +126,7 @@ def load_model(server_args, tp_rank):
     suppress_other_loggers()
     rank_print = print if tp_rank == 0 else lambda *args, **kwargs: None
 
-    server_args.port, server_args.additional_ports = allocate_init_ports(
-        server_args.port,
-        server_args.additional_ports,
-        server_args.dp_size,
-    )
+    port_args = PortArgs.init_new(server_args)
     model_config = ModelConfig(
         server_args.model_path,
         server_args.trust_remote_code,
@@ -143,7 +138,7 @@ def load_model(server_args, tp_rank):
         gpu_id=tp_rank,
         tp_rank=tp_rank,
         tp_size=server_args.tp_size,
-        nccl_port=server_args.additional_ports[-1],
+        nccl_port=port_args.nccl_ports[0],
         server_args=server_args,
     )
     rank_print(f"max_total_num_tokens={model_runner.max_total_num_tokens}")

python/sglang/srt/managers/detokenizer_manager.py

@@ -59,10 +59,10 @@ class DetokenizerManager:
         # Init inter-process communication
         context = zmq.Context(2)
         self.recv_from_scheduler = context.socket(zmq.PULL)
-        self.recv_from_scheduler.bind(f"tcp://127.0.0.1:{port_args.detokenizer_port}")
+        self.recv_from_scheduler.bind(f"ipc://{port_args.detokenizer_ipc_name}")
 
         self.send_to_tokenizer = context.socket(zmq.PUSH)
-        self.send_to_tokenizer.connect(f"tcp://127.0.0.1:{port_args.tokenizer_port}")
+        self.send_to_tokenizer.connect(f"ipc://{port_args.tokenizer_ipc_name}")
 
         if server_args.skip_tokenizer_init:
             self.tokenizer = None

python/sglang/srt/managers/scheduler.py

@@ -96,14 +96,10 @@ class Scheduler:
 
         if self.tp_rank == 0:
             self.recv_from_tokenizer = context.socket(zmq.PULL)
-            self.recv_from_tokenizer.bind(
-                f"tcp://127.0.0.1:{port_args.scheduler_input_port}"
-            )
+            self.recv_from_tokenizer.bind(f"ipc://{port_args.scheduler_input_ipc_name}")
 
             self.send_to_detokenizer = context.socket(zmq.PUSH)
-            self.send_to_detokenizer.connect(
-                f"tcp://127.0.0.1:{port_args.detokenizer_port}"
-            )
+            self.send_to_detokenizer.connect(f"ipc://{port_args.detokenizer_ipc_name}")
         else:
             self.recv_from_tokenizer = self.send_to_detokenizer = None
 

python/sglang/srt/managers/tokenizer_manager.py

@@ -84,12 +84,10 @@ class TokenizerManager:
         # Init inter-process communication
         context = zmq.asyncio.Context(2)
         self.recv_from_detokenizer = context.socket(zmq.PULL)
-        self.recv_from_detokenizer.bind(f"tcp://127.0.0.1:{port_args.tokenizer_port}")
+        self.recv_from_detokenizer.bind(f"ipc://{port_args.tokenizer_ipc_name}")
 
         self.send_to_scheduler = context.socket(zmq.PUSH)
-        self.send_to_scheduler.connect(
-            f"tcp://127.0.0.1:{port_args.scheduler_input_port}"
-        )
+        self.send_to_scheduler.connect(f"ipc://{port_args.scheduler_input_ipc_name}")
 
         # Read model args
         self.model_path = server_args.model_path

python/sglang/srt/mem_cache/memory_pool.py

@@ -16,7 +16,6 @@ limitations under the License.
 """Memory pool."""
 
 import logging
-from abc import ABC, abstractmethod
 from typing import List, Tuple, Union
 
 import numpy as np
@@ -62,9 +61,11 @@ class BaseTokenToKVPool:
         self,
         size: int,
         dtype: torch.dtype,
+        device: str,
     ):
         self.size = size
         self.dtype = dtype
+        self.device = device
         if dtype == torch.float8_e5m2:
             # NOTE: Store as torch.uint8 because Tensor index_put is not implemented for torch.float8_e5m2
             self.store_dtype = torch.uint8
@@ -84,7 +85,7 @@ class BaseTokenToKVPool:
         select_index = self.free_slots[:need_size]
         self.free_slots = self.free_slots[need_size:]
 
-        return torch.tensor(select_index, dtype=torch.int32, device="cuda")
+        return torch.tensor(select_index, dtype=torch.int32, device=self.device)
 
     def free(self, free_index: torch.Tensor):
         self.free_slots = np.concatenate((self.free_slots, free_index.cpu().numpy()))
@@ -123,7 +124,7 @@ class MHATokenToKVPool(BaseTokenToKVPool):
         layer_num: int,
         device: str,
     ):
-        super().__init__(size, dtype)
+        super().__init__(size, dtype, device)
 
         # [size, head_num, head_dim] for each layer
         # The padded slot 0 is used for writing dummy outputs from padded tokens.
@@ -187,7 +188,7 @@ class MLATokenToKVPool(BaseTokenToKVPool):
         layer_num: int,
         device: str,
     ):
-        super().__init__(size, dtype)
+        super().__init__(size, dtype, device)
 
         self.kv_lora_rank = kv_lora_rank
         # The padded slot 0 is used for writing dummy outputs from padded tokens.

python/sglang/srt/server.py

@@ -24,6 +24,7 @@ import json
 import logging
 import multiprocessing as mp
 import os
+import random
 import threading
 import time
 from http import HTTPStatus
@@ -68,9 +69,9 @@ from sglang.srt.openai_api.protocol import ModelCard, ModelList
 from sglang.srt.server_args import PortArgs, ServerArgs
 from sglang.srt.utils import (
     add_api_key_middleware,
-    allocate_init_ports,
     assert_pkg_version,
     configure_logger,
+    is_port_available,
     kill_child_process,
     maybe_set_triton_cache_manager,
     prepare_model_and_tokenizer,
@@ -302,18 +303,7 @@ def launch_server(
     _set_envs_and_config(server_args)
 
     # Allocate ports for inter-process communications
-    server_args.port, server_args.additional_ports = allocate_init_ports(
-        server_args.port,
-        server_args.additional_ports,
-        server_args.dp_size,
-    )
-    ports = server_args.additional_ports
-    port_args = PortArgs(
-        tokenizer_port=ports[0],
-        scheduler_input_port=ports[1],
-        detokenizer_port=ports[2],
-        nccl_ports=ports[3:],
-    )
+    port_args = PortArgs.init_new(server_args)
     logger.info(f"{server_args=}")
 
     # If using model from www.modelscope.cn, first download the model.
@@ -499,17 +489,16 @@ class Runtime:
         self.server_args = ServerArgs(*args, log_level=log_level, **kwargs)
 
         # Pre-allocate ports
-        self.server_args.port, self.server_args.additional_ports = allocate_init_ports(
-            self.server_args.port,
-            self.server_args.additional_ports,
-            self.server_args.dp_size,
-        )
+        for port in range(10000, 40000):
+            if is_port_available(port):
+                break
+            port += 1
+        self.server_args.port = port
 
         self.url = self.server_args.url()
-        self.generate_url = (
-            f"http://{self.server_args.host}:{self.server_args.port}/generate"
-        )
+        self.generate_url = self.url + "/generate"
 
+        # NOTE: We store pid instead of proc to fix some issues during __delete__
         self.pid = None
         pipe_reader, pipe_writer = mp.Pipe(duplex=False)
 

python/sglang/srt/server_args.py

@@ -19,9 +19,10 @@ import argparse
 import dataclasses
 import logging
 import random
-from typing import List, Optional, Union
+import tempfile
+from typing import List, Optional
 
-from sglang.srt.utils import is_hip, is_ipv6
+from sglang.srt.utils import is_hip, is_ipv6, is_port_available
 
 logger = logging.getLogger(__name__)
 
@@ -46,7 +47,6 @@ class ServerArgs:
     # Port
     host: str = "127.0.0.1"
     port: int = 30000
-    additional_ports: Optional[Union[List[int], int]] = None
 
     # Memory and scheduling
     mem_fraction_static: Optional[float] = None
@@ -134,11 +134,6 @@ class ServerArgs:
             else:
                 self.mem_fraction_static = 0.88
 
-        if isinstance(self.additional_ports, int):
-            self.additional_ports = [self.additional_ports]
-        elif self.additional_ports is None:
-            self.additional_ports = []
-
         if self.random_seed is None:
             self.random_seed = random.randint(0, 1 << 30)
 
@@ -199,13 +194,6 @@ class ServerArgs:
         parser.add_argument(
             "--port", type=int, default=ServerArgs.port, help="The port of the server."
         )
-        parser.add_argument(
-            "--additional-ports",
-            type=int,
-            nargs="*",
-            default=[],
-            help="The additional ports specified for the server.",
-        )
         parser.add_argument(
             "--tokenizer-mode",
             type=str,
@@ -625,16 +613,31 @@ def prepare_server_args(argv: List[str]) -> ServerArgs:
 
 @dataclasses.dataclass
 class PortArgs:
-    # The port for tokenizer to receive inputs from detokenizer (zmq)
-    tokenizer_port: int
-    # The port for scheduler (rank 0) to receive inputs from tokenizer (zmq)
-    scheduler_input_port: int
-    # The port for detokenizer to receive inputs from scheduler (zmq)
-    detokenizer_port: int
+    # The ipc filename for tokenizer to receive inputs from detokenizer (zmq)
+    tokenizer_ipc_name: str
+    # The ipc filename for scheduler (rank 0) to receive inputs from tokenizer (zmq)
+    scheduler_input_ipc_name: str
+    # The ipc filename for detokenizer to receive inputs from scheduler (zmq)
+    detokenizer_ipc_name: str
 
     # The port for nccl initialization for multiple TP groups (torch.dist)
     nccl_ports: List[int]
 
+    @classmethod
+    def init_new(self, server_args):
+        port = server_args.port + 1
+        while True:
+            if is_port_available(port):
+                break
+            port += 1
+
+        return PortArgs(
+            tokenizer_ipc_name=tempfile.NamedTemporaryFile(delete=False).name,
+            scheduler_input_ipc_name=tempfile.NamedTemporaryFile(delete=False).name,
+            detokenizer_ipc_name=tempfile.NamedTemporaryFile(delete=False).name,
+            nccl_ports=[port],
+        )
+
 
 class LoRAPathAction(argparse.Action):
     def __call__(self, parser, namespace, values, option_string=None):

python/sglang/srt/utils.py

@@ -177,35 +177,6 @@ def is_port_available(port):
             return False
 
 
-def allocate_init_ports(
-    port: Optional[int] = None,
-    additional_ports: Optional[List[int]] = None,
-    dp_size: int = 1,
-):
-    """Allocate ports for all connections."""
-    if additional_ports:
-        ret_ports = [port] + additional_ports
-    else:
-        ret_ports = [port]
-
-    ret_ports = list(set(x for x in ret_ports if is_port_available(x)))
-    cur_port = ret_ports[-1] + 1 if len(ret_ports) > 0 else 10000
-
-    # HTTP + Tokenizer + Controller + Detokenizer + dp_size * 1 (nccl)
-    num_ports_needed = 4 + dp_size
-    while len(ret_ports) < num_ports_needed:
-        if cur_port not in ret_ports and is_port_available(cur_port):
-            ret_ports.append(cur_port)
-        cur_port += 1
-
-    if port is not None and ret_ports[0] != port:
-        logger.warning(
-            f"WARNING: Port {port} is not available. Use port {ret_ports[0]} instead."
-        )
-
-    return ret_ports[0], ret_ports[1:num_ports_needed]
-
-
 def is_multimodal_model(model_architectures):
     if (
         "LlavaLlamaForCausalLM" in model_architectures