feat(training): release V0.8 自调参 Agent

training_server/tuning/process.py

@@ -0,0 +1,46 @@
+"""Shared GPU lease and process-group lifecycle helpers."""
+
+from __future__ import annotations
+
+import os
+import signal
+import subprocess
+import threading
+from contextlib import suppress
+
+
+class ResourceBusyError(RuntimeError):
+    pass
+
+
+class GpuLease:
+    def __init__(self):
+        self.lock = threading.RLock()
+        self.owner: str | None = None
+
+    def acquire(self, owner: str) -> None:
+        with self.lock:
+            if self.owner is not None and self.owner != owner:
+                raise ResourceBusyError(f"计算资源正由 {self.owner} 使用")
+            self.owner = owner
+
+    def release(self, owner: str) -> None:
+        with self.lock:
+            if self.owner == owner:
+                self.owner = None
+
+    def public(self) -> str | None:
+        with self.lock:
+            return self.owner
+
+
+def terminate_process(process: subprocess.Popen[str], grace_seconds: float = 5.0) -> None:
+    if process.poll() is not None:
+        return
+    with suppress(ProcessLookupError):
+        os.killpg(process.pid, signal.SIGTERM)
+    try:
+        process.wait(timeout=grace_seconds)
+    except subprocess.TimeoutExpired:
+        with suppress(ProcessLookupError):
+            os.killpg(process.pid, signal.SIGKILL)