Atlas 技術版 > 系統服務存活監控 watchdog

系統服務存活監控 watchdog NO_AGENT P0 risk=HIGH

1. 做什麼 / 2. 為什麼存在

系統服務存活監控 watchdog：每小時檢查 Gateway/WebUI/cron ticker/embedding server 等，故障即推 Telegram。

3. Schedule / Mode

mode=NO_AGENT state=scheduled enabled=1
schedule=0 * * * *
next_run=2026-10-21T05:00:00 last_run=2026-10-21T03:00:41
deliver=telegram model=— provider=—

4. Implementation

script=/Users/exion/.hermes/scripts/system_watchdog.py
monitor_script=— monitor_url=—
skills=[]
workdir=—

5. Effective instruction sources

prompt_sha256=40741e3d8dd435c8… chars=66
skill_files={}

Nemotron Embed Server 健康監控：每小時探測 8457，失敗時輸出錯誤診斷推送 Telegram（正常時靜默）。…

6. Actual rules / thresholds

rule	value	source
TIMEOUT	15	CODE_STATIC @system_watchdog.py:28
HEARTBEAT_FILE	/Users/exion/.hermes/cron/ticker_heartbeat	CODE_STATIC @system_watchdog.py:66
TICKER_SUCCESS_FILE	/Users/exion/.hermes/cron/ticker_last_success	CODE_STATIC @system_watchdog.py:67
STALE_AFTER	200	CODE_STATIC @system_watchdog.py:68

7/8. Inputs & Outputs (resources)

kind	path	name	access
SQLITE_DB	/Users/exion/.hermes/mnemosyne/data/mnemosyne.db		unknown
SQLITE_TABLE	self	working_memory	read
FILE_READ	/Users/exion/.hermes/cron/ticker_heartbeat		read
FILE_READ	/Users/exion/.hermes/cron/ticker_last_success		read
HTTP	http://127.0.0.1:8457/health		execute
HTTP	http://127.0.0.1:8457/v1/embeddings		execute
SUBPROCESS		launchctl list	execute
SUBPROCESS		ps -ax -o pid=,command=	execute

9/10. Upstream / Downstream

Upstream: P0-3 Phase 3: importance 重評執行（30天數據）
Downstream: —
Transitive downstream (0): —

Edges (typed evidence)

type	hard?	from	to	resource	origin	conf	evidence
DATA	soft	P0-3 Phase 3: importance 重評執行（30天數據）	系統服務存活監控 watchdog	sqlite:working_memory	STATIC	0.75	table 'working_memory' write@3440ecd342a8(line 163) -> read@827cf16b631c(line 215)

11. Shared resources

/Users/exion/.hermes/mnemosyne/data/mnemosyne.db — shared with 2 jobs
working_memory — shared with 2 jobs

12. Recent executions

failure_streak=0
last_error=—
delivery_error=—

14. Drift history

2026-09-05T06:57 JOB_CHANGED job 827cf16b631c (系統服務存活監控 watchdog) changed: mode=NO_AGENT state=scheduled
2026-09-05T06:57 TEMPORAL_DEPENDENCY_HAZARD '系統服務存活監控 watchdog' fires 00:00 but depends on 'P0-3 Phase 3: importance 重評執行（30天數據）' (earliest 09:00) via DATA [sqlite:working_memory]
2026-09-05T06:57 JOB_CHANGED job 827cf16b631c (系統服務存活監控 watchdog) changed: mode=NO_AGENT state=scheduled
2026-09-05T00:38 JOB_ADDED job 827cf16b631c appeared (系統服務存活監控 watchdog)
2026-09-04T23:49 TEMPORAL_DEPENDENCY_HAZARD '系統服務存活監控 watchdog' fires 00:00 but depends on 'P0-3 Phase 3: importance 重評執行（30天數據）' (earliest 09:00) via DATA [sqlite:working_memory]
2026-09-04T23:25 TEMPORAL_DEPENDENCY_HAZARD '系統服務存活監控 watchdog' fires 00:00 but depends on 'P0-3 Phase 3: importance 重評執行（30天數據）' (earliest 09:00) via DATA [sqlite:working_memory]
2026-09-04T22:51 TEMPORAL_DEPENDENCY_HAZARD '系統服務存活監控 watchdog' fires 00:00 but depends on 'P0-3 Phase 3: importance 重評執行（30天數據）' (earliest 09:00) via DATA [sqlite:working_memory]
2026-09-04T22:51 JOB_ADDED job 827cf16b631c appeared (系統服務存活監控 watchdog)
2026-09-03T20:00 TEMPORAL_DEPENDENCY_HAZARD '系統服務存活監控 watchdog' fires 00:00 but depends on 'P0-3 Phase 3: importance 重評執行（30天數據）' (earliest 09:00) via DATA [sqlite:working_memory]
2026-09-03T20:00 JOB_ADDED job 827cf16b631c appeared (系統服務存活監控 watchdog)

16. Impact / change risk

change_risk=HIGH