fix(daily_check): 期望窗口未到不应判 missed

work #07 (2026-07-03):daily_check 在日内运行时(如 09:00/12:00/15:00)
误把'还没到时间'的任务标 missed。SCHEDULE 里有 window_end 但 _classify_task
没用到——直接判 last_success vs today_start。

修复:判定 missed 之前先比较 now vs window_end;未到则返回 not_yet
(新状态,不进 alerts),已过但今日未成功走原 missed 路径。解析失败
回落旧逻辑不引入风险。

验证 4 个用例:12:00 kline_daily→not_yet, 18:00 kline_daily→missed,
18:00 longhubang→not_yet, 22:30 longhubang→missed。全部正确。

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
gao
2026-07-03 18:25:50 +08:00
parent b351bd7595
commit f7a2cbe5a8
2 changed files with 90 additions and 0 deletions
+23
View File
@@ -133,6 +133,28 @@ def _classify_task(
last_failure = _parse_ts(row.get("last_failure_at")) last_failure = _parse_ts(row.get("last_failure_at"))
status_field = row.get("status", "idle") status_field = row.get("status", "idle")
# 期望窗口是否已过?未到窗口不应判 missed
if spec and spec.get("window_end") and not status_field == "running":
try:
hh, mm = spec["window_end"].split(":")
window_end_dt = datetime.combine(check_date, datetime.min.time()).replace(
hour=int(hh), minute=int(mm)
)
if now < window_end_dt:
return {
"dataset_id": dataset_id,
"name": row.get("name", ""),
"status": "not_yet",
"reason": f"期望窗口 {spec['window_end']} 未到(now={now.strftime('%H:%M')}",
"last_success_at": row.get("last_success_at"),
"last_failure_at": row.get("last_failure_at"),
"last_error": row.get("last_error"),
"needs_resync": row.get("needs_resync", 0),
"status_field": status_field,
}
except (ValueError, AttributeError):
pass # 解析失败回落到旧逻辑
# 还在跑(卡死检测) # 还在跑(卡死检测)
if status_field == "running": if status_field == "running":
started = _parse_ts(row.get("started_at")) started = _parse_ts(row.get("started_at"))
@@ -399,6 +421,7 @@ def print_summary(report: dict[str, Any]) -> None:
"running": "🔄", "running": "🔄",
"never_run": "", "never_run": "",
"not_expected": "", "not_expected": "",
"not_yet": "",
"disabled": "🚫", "disabled": "🚫",
}.get(t["status"], "·") }.get(t["status"], "·")
print(f" {icon} [{t['status']:>13}] {t['dataset_id']:<20} {t['reason']}") print(f" {icon} [{t['status']:>13}] {t['dataset_id']:<20} {t['reason']}")
@@ -0,0 +1,67 @@
# 2026-07-03 — work #07: daily_check 分类逻辑 bug — 期望窗口未到也判 missed
## 发现
12/12 全部 ok 后跑 daily_check 复验,看到 report 里有个问题:
```python
# 旧逻辑(伪码)
if last_success and last_success >= today_start:
return "ok"
elif last_failure and last_failure > last_success and last_failure >= today_start:
return "failed"
else:
return "missed" # ❌ 不管窗口是否已过
```
跑一个反例(today=07-03 12:00, last_success=07-02 15:56, window_end=15:45):
- **期望**not_yet(窗口 15:30-15:45 还没到)
- **实际**missed(误报)
每天 23:00 跑 daily_check 时这个 bug 不可见(窗口都已过);但在 09:00 / 12:00 / 15:00 这种日内 check 会误报,让"daily check 视图"看起来比实际严重。
## 根因
`SCHEDULE` 表里有 `window_end`(如 kline_daily=15:45)但 `_classify_task` 函数没用它 —— 直接基于 `last_success_at` vs `today_start` 判定。
## 解决
`_classify_task` 在判定"今日未成功"之前先看 `now < window_end`
- 窗口未到 → 返回 `status=not_yet`(新状态,不进 alerts
- 窗口已过 + 今日未成功 → 走原 missed 路径
- 解析失败(window_end 格式异常)→ 回落旧逻辑(不引入新风险)
新 status `not_yet` 加 icon `⏳` 到 print_summary。
## 验证
4 个测试用例:
| 时间点 | task | window_end | 期望 | 实际 |
|---|---|---|---|---|
| 12:00 | kline_daily | 15:45 | not_yet | ✅ not_yet |
| 18:00 | kline_daily | 15:45 | missed | ✅ missed |
| 18:00 | longhubang | 22:05 | not_yet | ✅ not_yet |
| 22:30 | longhubang | 22:05 | missed | ✅ missed |
mock 5 个 task 跑 build_report
```
counts: {'ok': 2, 'missed': 1, 'not_yet': 1, 'not_expected': 1}
overall: warning
stock_basic status=ok
kline_daily status=missed
longhubang status=not_yet
stock_node status=not_expected
mairui_ma_daily status=ok
```
## 关联
- [[market-data-overview]]12 task 的窗口时间表
- 这次修改让"在指定时间"语义在 check 阶段也正确(不再把"还没到时间"误报为 missed
## 不在范围(未来)
- 给"stuck" 任务也加窗口判断(running 任务超过 24h 才算 stuck,目前是 2h
- 增加"in_window"状态(任务正在窗口内)— 现版 "not_yet" 模糊覆盖了
- 历史回看:让 check_date 接受任意一天,校验过去的 missed(现在只查今天)