fix: 断点丢失时 goal-resume 拒绝成 failed,阻断自动恢复空转

2026-10-29 长任务实弹:Docker goal 断点续跑在 host-state 里找
不到任务时抛 KeyError,worker 把无结论异常退出判 interrupted,
自动恢复再排队、再 KeyError,12 秒一轮空转 49 次(人工暂停
拦下,第 50 次人工续跑又烧掉),文件 32 章原样未动。
修法:host.py goal-resume 先查任务在不在 state 里,不在就打印
拒绝行、正常退出;sandbox 三处解析(本地、Docker resume、
Docker 断点)见拒绝行即判 failed 终态,不进自动恢复循环。
单测:host 缺任务拒绝不炸 + 三处解析拒绝判 failed。
This commit is contained in:
host committed 2026-10-29 14:14:13 +08:00
1 parent e48b756f96
commit a5f2e4e091
4 files changed
+138 -3

No files matched your search

+31 -3
View File
@@ -117,6 +117,19 @@ def _parse_goal_output(stdout):
return None, None
def _goal_resume_refusal(stdout):
"""host.py goal-resume 的拒绝行(断点丢失时),无则返回 None。
断点丢失不是可续跑的中断:host-state 里没有任务,任何一次
续跑都会原样失败;调用方必须判 failed 终态,绝不能判
interrupted——否则 worker 自动恢复会 12 秒一轮空转到上限。
"""
for line in (stdout or "").splitlines():
if line.startswith("goal-resume 拒绝:"):
return line.strip()
return None
def _goal_interrupted_reason(stdout):
"""goal interrupted 标记行的原因,无标记返回 None。
@@ -569,8 +582,13 @@ class LocalExecutor(Executor):
stderr = getattr(result, "stderr", "") or ""
met, goal_error = _parse_goal_output(stdout_full)
interrupted_reason = _goal_interrupted_reason(stdout_full)
refusal = _goal_resume_refusal(stdout_full)
combined = stdout_full + stderr
if "已存在" in combined:
if refusal is not None and met is None:
# 断点丢失拒绝:终态 failed,不进自动恢复(见 helper)。
met, goal_error = None, refusal
status = "failed"
elif "已存在" in combined:
met, goal_error = None, (
"host 任务已存在:goal 失败请用继续而非重试——"
"同任务重试会撞已存在 host-state,worker 会先"
@@ -1388,7 +1406,12 @@ class DockerExecutor(Executor):
stderr = getattr(result, "stderr", "") or ""
met, goal_error = _parse_goal_output(stdout_full)
interrupted_reason = _goal_interrupted_reason(stdout_full)
if interrupted_reason is not None:
refusal = _goal_resume_refusal(stdout_full)
if refusal is not None and met is None:
# 断点丢失拒绝:终态 failed,不进自动恢复(见 helper)。
status = "failed"
goal_error = refusal
elif interrupted_reason is not None:
# host 侧已记 interrupted:异常中断不是失败,云端跟
# 着记 interrupted,任务可 /goal-resume 断点续跑。
status = "interrupted"
@@ -1484,7 +1507,12 @@ class DockerExecutor(Executor):
stderr = getattr(result, "stderr", "") or ""
met, goal_error = _parse_goal_output(stdout_full)
interrupted_reason = _goal_interrupted_reason(stdout_full)
if interrupted_reason is not None:
refusal = _goal_resume_refusal(stdout_full)
if refusal is not None and met is None:
# 断点丢失拒绝:终态 failed,不进自动恢复(见 helper)。
status = "failed"
goal_error = refusal
elif interrupted_reason is not None:
# host 侧已记 interrupted:异常中断不是失败,云端跟
# 着记 interrupted,任务可 /goal-resume 断点续跑。
status = "interrupted"