furyhawk/deer-flow
1
0
Fork 0
mirror of https://github.com/bytedance/deer-flow.git synced 2026-05-23 16:35:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(sandbox): cleanup dead containers and avoid lock-held liveness checks

Browse Source 
Agent-Logs-Url: https://github.com/bytedance/deer-flow/sessions/96707445-0f8b-4901-8ef3-d8e5667f8a05

Co-authored-by: WillemJiang <219644+WillemJiang@users.noreply.github.com>
This commit is contained in:
copilot-swe-agent[bot]
2026-05-11 00:09:09 +00:00
committed by GitHub
parent b67c2a4e56
commit dad3997459
2 changed files with 66 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/packages/harness/deerflow/community/aio_sandbox/aio_sandbox_provider.py
+37 -16
View File
@@ -624,24 +624,45 @@ class AioSandboxProvider(SandboxProvider):
"""
with self._lock:
sandbox = self._sandboxes.get(sandbox_id)
if sandbox is not None:
self._last_activity[sandbox_id] = time.time()
# Auto-restart: detect crashed containers and evict from cache
# so the next acquire() transparently recreates them.
if self._config.get("auto_restart", True):
info = self._sandbox_infos.get(sandbox_id)
if info and not self._backend.is_alive(info):
logger.warning(f"Sandbox {sandbox_id} container is not alive, evicting from cache for auto-restart")
self._sandboxes.pop(sandbox_id, None)
self._sandbox_infos.pop(sandbox_id, None)
self._last_activity.pop(sandbox_id, None)
thread_ids = [tid for tid, sid in self._thread_sandboxes.items() if sid == sandbox_id]
for tid in thread_ids:
del self._thread_sandboxes[tid]
return None
return sandbox
if sandbox is None:
return None
self._last_activity[sandbox_id] = time.time()
auto_restart = self._config.get("auto_restart", True)
info = self._sandbox_infos.get(sandbox_id) if auto_restart else None
if not info:
return sandbox
if self._backend.is_alive(info):
return sandbox
info_to_destroy = None
with self._lock:
current_sandbox = self._sandboxes.get(sandbox_id)
current_info = self._sandbox_infos.get(sandbox_id)
if current_sandbox is None:
return None
if current_info is not info:
self._last_activity[sandbox_id] = time.time()
return current_sandbox
logger.warning(f"Sandbox {sandbox_id} container is not alive, evicting from cache for auto-restart")
self._sandboxes.pop(sandbox_id, None)
self._sandbox_infos.pop(sandbox_id, None)
self._last_activity.pop(sandbox_id, None)
self._warm_pool.pop(sandbox_id, None)
thread_ids = [tid for tid, sid in self._thread_sandboxes.items() if sid == sandbox_id]
for tid in thread_ids:
del self._thread_sandboxes[tid]
info_to_destroy = info
if info_to_destroy:
try:
self._backend.destroy(info_to_destroy)
except Exception as e:
logger.warning(f"Failed to cleanup dead sandbox {sandbox_id}: {e}")
return None
def release(self, sandbox_id: str) -> None:
"""Release a sandbox from active use into the warm pool.