truss 0.11.18__tar.gz → 0.11.18rc501__tar.gz

{truss-0.11.18 → truss-0.11.18rc501}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: truss
-Version: 0.11.18
+Version: 0.11.18rc501
 Summary: A seamless bridge from model development to model delivery
 Project-URL: Repository, https://github.com/basetenlabs/truss
 Project-URL: Homepage, https://truss.baseten.co

{truss-0.11.18 → truss-0.11.18rc501}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "truss"
-version = "0.11.18"
+version = "0.11.18rc501"
 description = "A seamless bridge from model development to model delivery"
 authors = [
     { name = "Pankaj Gupta", email = "no-reply@baseten.co" },

{truss-0.11.18 → truss-0.11.18rc501}/truss/templates/control/control/helpers/inference_server_process_controller.py

@@ -5,6 +5,7 @@ from pathlib import Path
 from typing import List, Optional
 
 from helpers.context_managers import current_directory
+from shared.util import kill_child_processes
 
 INFERENCE_SERVER_FAILED_FILE = Path("~/inference_server_crashed.txt").expanduser()
 TERMINATION_TIMEOUT_SECS = 120.0
@@ -46,17 +47,21 @@ class InferenceServerProcessController:
             self._inference_server_ever_started = True
             self._logged_unrecoverable_since_last_restart = False
 
+    def _terminate_children_and_process(self):
+        """Kill child processes first, then parent. Prevents port binding conflicts."""
+        # Use 2s timeout for faster restarts (was 120s default)
+        kill_child_processes(self._inference_server_process.pid, timeout=2)
+        self._inference_server_process.terminate()
+
     def stop(self):
         if self._inference_server_process is not None:
-            self._inference_server_process.terminate()
+            self._terminate_children_and_process()
             self._inference_server_process.wait()
-            # Introduce delay to avoid failing to grab the port
-            time.sleep(3)
 
         self._inference_server_started = False
 
     def terminate_with_wait(self):
-        self._inference_server_process.terminate()
+        self._terminate_children_and_process()
         self._inference_server_terminated = True
         termination_check_attempts = int(
             TERMINATION_TIMEOUT_SECS / TERMINATION_CHECK_INTERVAL_SECS

{truss-0.11.18 → truss-0.11.18rc501}/truss/templates/shared/util.py

@@ -1,7 +1,7 @@
 import multiprocessing
 import os
 import sys
-from typing import List
+from typing import List, Optional
 
 import psutil
 
@@ -62,7 +62,7 @@ def all_processes_dead(procs: List[multiprocessing.Process]) -> bool:
     return True
 
 
-def kill_child_processes(parent_pid: int):
+def kill_child_processes(parent_pid: int, timeout: Optional[float] = CHILD_PROCESS_WAIT_TIMEOUT_SECONDS):
     try:
         parent = psutil.Process(parent_pid)
     except psutil.NoSuchProcess:
@@ -71,7 +71,7 @@ def kill_child_processes(parent_pid: int):
     for process in children:
         process.terminate()
     gone, alive = psutil.wait_procs(
-        children, timeout=CHILD_PROCESS_WAIT_TIMEOUT_SECONDS
+        children, timeout=timeout
     )
     for process in alive:
         process.kill()