記錄 Observability fail-safe 區塊失敗

docs/memory/claude_inventory_validation_20260513.md

@@ -17,6 +17,7 @@
 - Elephant Alpha short-circuit：`log_ai_call` 遙測失敗仍不阻擋省成本 return，但已改為 warning + stack。
 - Claude cost throttle：成本節流檢查失敗仍維持 Claude 可用，但已改為 warning + stack，避免成本保護失效無跡可查。
 - `ai_call_logger` caller registry：registry 匯入失敗仍不阻擋 LLM 遙測，但已改為 warning + stack。
+- Observability route：promotion review RAG 相似查詢、PPT audit history 缺表、host health probe 寫入、MCP 24h summary 缺表等 fail-safe 區塊已改成 debug/warning log，不再完全靜默。
 
 ## 已驗證為已修或過期
 
@@ -55,3 +56,4 @@
 - `f49413e` 記錄 EA short-circuit 遙測失敗
 - `0a75d11` 記錄 Claude 成本節流檢查失敗
 - `5625032` 記錄 AI caller registry 匯入失敗
+- `1757837` 記錄 Observability fail-safe 區塊失敗

routes/admin_observability_routes.py

@@ -17,6 +17,7 @@ Operation Ollama-First v5.0 / Phase 27 — Admin Observability Dashboard
 - 不暴露 secret / prompt 原文
 """
 
+import logging
 from datetime import datetime, timedelta
 from flask import Blueprint, render_template, request, jsonify
 from sqlalchemy import text as sa_text
@@ -24,6 +25,8 @@ from sqlalchemy import text as sa_text
 from auth import login_required, get_current_user
 from database.manager import get_session
 
+logger = logging.getLogger(__name__)
+
 
 admin_observability_bp = Blueprint(
     'admin_observability',
@@ -1289,9 +1292,13 @@ def promotion_review_list():
                         for h in rag_result.hits[:3]
                     ]
                 except Exception:
-                    pass  # 單筆 RAG 失敗不影響其餘
+                    logger.debug(
+                        "Promotion review similar-insight lookup failed for episode_id=%s",
+                        ep.get('id'),
+                        exc_info=True,
+                    )
         except Exception:
-            pass  # rag_service import 失敗（feature flag OFF）→ 略過
+            logger.debug("Promotion review RAG service unavailable; skipping similar insights", exc_info=True)
 
         # Phase 47 K-4: 蒸餾池 status 分布（30d）
         ep_distribution = session.execute(
@@ -2192,7 +2199,7 @@ def ppt_audit_history():
         finally:
             session.close()
     except Exception:
-        pass  # 表可能尚未 migration，失敗安全
+        logger.debug("PPT audit history table unavailable; rendering empty audit history", exc_info=True)
 
     # PPT vision 啟用狀態
     try:
@@ -2374,7 +2381,7 @@ def host_health_dashboard():
             finally:
                 _session.close()
         except Exception:
-            pass  # DB 寫入失敗不影響頁面顯示
+            logger.warning("Failed to persist host health probe records", exc_info=True)
 
     # MCP server 健康
     mcp_status = {}
@@ -2520,7 +2527,7 @@ def host_health_dashboard():
         finally:
             _session2.close()
     except Exception:
-        pass  # 表可能尚未 migration，失敗安全
+        logger.debug("MCP calls table unavailable; rendering empty MCP 24h summary", exc_info=True)
 
     # Phase 47 K-1: incidents + heal_logs 詳細列表 + playbooks 排行 + backup + embed queue
     recent_incidents = []