fix: meta_watchdog删除L0 agents_health_check检查项——MoFin无该组件(deploy/profile-scripts无agents_health_check.py),且L1 functional_health_check已覆盖健康检查功能,检查项是死代码导致误报(检查MoFin路径的last_health_check.json,但AgentsMeeting版写的是AgentsMeeting路径)

This commit is contained in:
hmo
2026-08-13 23:28:19 +08:00
parent 535867e16f
commit 5a97442520
+3 -5
View File
@@ -1,14 +1,15 @@
#!/usr/bin/env python3
"""meta_watchdog.py — L4 自检系统的自检(看门狗的看门狗)
检查 L0-L3 各自检组件本身是否在正常运转:
- L0 agents_health_check: last_health_check.json 是否 <10min
检查 L1-L3 各自检组件本身是否在正常运转:
- L1 functional_health_check: functional_health.json 是否 <20min(交易时段)
- L2 system_hygiene_audit: hygiene_report.json 是否 <26h(每日)
- L3 self_repair: repair_state.json 存在性 + cron 是否注册
- mofin_health 采集: mofin_health.json 是否 <20min(交易时段)
- XMPP 桥: :5805 是否可发(self_repair 的报备通道)
2026-08-13 删除 L0 agents_health_check 检查项:MoFin 无该组件,且 L1 functional_health_check 已覆盖健康检查功能,检查项是死代码)
任何一层死了 → 推 XMPP 点名(这是最后的兜底,必须直达用户)。
频率:每小时(cron)。输出 gateway/logs/meta_watchdog.json。
"""
@@ -18,9 +19,6 @@ from datetime import datetime
OUT = '/home/hmo/MoFin/gateway/logs/meta_watchdog.json'
LAYERS = [
{"layer": "L0 agents_health_check", "file": "/home/hmo/MoFin/gateway/temp/last_health_check.json",
"max_age_min": 10, "when": "always",
"repair": "crontab */5 agents_health_check.py 停摆,检查系统 crontab"},
{"layer": "L1 functional_health", "file": "/home/hmo/MoFin/gateway/logs/functional_health.json",
"max_age_min": 25, "when": "trading",
"repair": "L1 cron 停摆,检查 hermes cron 引擎"},