d97777e74a
Once a model overflowed VRAM and Studio later died in a way that bypasses the graceful path (SIGHUP from a closed terminal, SIGKILL/OOM, or a direct-uvicorn launch), the llama-server child was orphaned and kept holding GPU memory. GPU placement is recomputed every load from a live free-VRAM probe, so the leftover process made every subsequent load (even a tiny model, even after a restart) spill to system RAM until it was killed by hand. The main llama-server spawn now records its PID to a pidfile under the active studio root, removed on _kill_process. The startup reaper kills that exact PID first (path-independent, so it catches an orphan the install-root match misses), verifying it is still a llama-server to guard against PID reuse; the pidfile only ever names a Studio-spawned server, so unrelated user processes (vllm, games) are never touched. The existing root-gated enumeration stays as a fallback. A belt-and-suspenders kill is also wired into the FastAPI lifespan shutdown (before hardware/cache teardown) to cover the direct-uvicorn path that run.py's signal handler does not. PR_SET_PDEATHSIG is intentionally not used on the main spawn: llama-server is launched on a pooled asyncio.to_thread worker, and a thread-scoped death signal could prematurely kill a healthy server. The reaper runs before any model loads on the next start, so it fully covers the user-visible problem. Adds tests for the pidfile reap (kills a recorded live server, skips a reused non-llama PID, cleans a stale/missing pidfile, clears on kill) and for the lifespan kill (runs first, errors swallowed).
66 lines
2.5 KiB
Python
66 lines
2.5 KiB
Python
# SPDX-License-Identifier: AGPL-3.0-only
|
|
# Copyright 2026-present the Unsloth AI Inc. team. All rights reserved. See /studio/LICENSE.AGPL-3.0
|
|
|
|
"""Resilient FastAPI lifespan shutdown cleanup.
|
|
|
|
On an abrupt shutdown (Windows console-close, interpreter teardown racing
|
|
uvicorn) the loop's default executor may already be dead, so an unguarded
|
|
``asyncio.to_thread`` raise here would abort the nested-lifespan unwind and
|
|
surface as "Application shutdown failed". Dependency-injected so it can be
|
|
unit-tested without the heavy backend import graph.
|
|
"""
|
|
|
|
import asyncio
|
|
import contextvars
|
|
import types
|
|
from typing import Callable, Optional
|
|
|
|
import structlog
|
|
|
|
logger = structlog.get_logger(__name__)
|
|
|
|
|
|
async def run_lifespan_shutdown(
|
|
terminate_downloads: Callable[[], None],
|
|
clear_compiled_cache: Callable[[], None],
|
|
hw_module: types.ModuleType,
|
|
kill_llama_server: Optional[Callable[[], None]] = None,
|
|
) -> None:
|
|
"""Run each shutdown step guarded so one failure can't skip the others; never raise."""
|
|
# Kill the llama-server child first, before clearing hardware/cache state, so a
|
|
# direct-uvicorn shutdown (which bypasses run.py's signal handler) cannot orphan a
|
|
# GPU process. The signal / _graceful_shutdown path already covers SIGTERM/SIGINT.
|
|
if kill_llama_server is not None:
|
|
try:
|
|
kill_llama_server()
|
|
except Exception as exc:
|
|
logger.warning("kill_llama_server failed at shutdown: %s", exc)
|
|
|
|
loop = asyncio.get_running_loop()
|
|
# Copy context for parity with asyncio.to_thread. Schedule and await
|
|
# separately so a dead executor (raises at submit) runs inline, while a
|
|
# body exception (raised at await) is logged, not re-run.
|
|
ctx = contextvars.copy_context()
|
|
try:
|
|
future = loop.run_in_executor(None, ctx.run, terminate_downloads)
|
|
except RuntimeError:
|
|
# Executor gone: run inline on the loop thread.
|
|
try:
|
|
ctx.run(terminate_downloads)
|
|
except Exception as exc:
|
|
logger.warning("terminate_downloads (inline) failed at shutdown: %s", exc)
|
|
else:
|
|
try:
|
|
await future
|
|
except Exception as exc:
|
|
logger.warning("terminate_downloads failed at shutdown: %s", exc)
|
|
|
|
try:
|
|
hw_module.DEVICE = None
|
|
except Exception as exc:
|
|
logger.warning("clearing hardware DEVICE failed at shutdown: %s", exc)
|
|
|
|
try:
|
|
clear_compiled_cache()
|
|
except Exception as exc:
|
|
logger.warning("clear_compiled_cache failed at shutdown: %s", exc)
|