add chache for last known good in sqlite..

src/fgai/monitor.py

@@ -12,7 +12,7 @@ from .event_context import build_event_context
 from .feedback import FeedbackStore
 from .graylog_mcp import GraylogMcpClient
 from .graylog_source import GraylogStreamSource
-from .history import HistoryStore
+from .history import HistoryStore, StatusSnapshotStore
 from .incidents import IncidentStore, build_incidents
 from .data_quality import assess_data_quality
 from .llm import ollama_dashboard_assessment, ollama_profile_advice
@@ -107,6 +107,7 @@ def build_status(
     config_path: str | None = None,
     history_path: str | None = None,
     incident_path: str | None = None,
+    status_cache_path: str | None = None,
 ) -> dict[str, object]:
     config_store = ConfigStore(config_path) if config_path else None
     config_exists = bool(config_store and config_store.path.exists())
@@ -342,9 +343,27 @@ def build_status(
         history = HistoryStore(history_path)
         history.record(status)
         status["history"] = history.recent()
+    if status_cache_path and not (isinstance(mcp_status, dict) and mcp_status.get("status") == "error"):
+        StatusSnapshotStore(status_cache_path).save("last_good", status)
     return status
 
 
+def cached_status_with_error(cache_path: str, error_status: dict[str, object]) -> dict[str, object] | None:
+    cached = StatusSnapshotStore(cache_path).load("last_good")
+    if not cached:
+        return None
+    cached["generated_at"] = int(time.time())
+    cached["stale"] = True
+    cached["stale_reason"] = "live_mcp_error"
+    capabilities = cached.setdefault("capabilities", {})
+    if isinstance(capabilities, dict):
+        capabilities["graylog_mcp"] = error_status
+    cached.setdefault("status_cache", {})
+    if isinstance(cached["status_cache"], dict):
+        cached["status_cache"].update({"served_from_cache": True, "reason": "live_mcp_error"})
+    return cached
+
+
 def add_llm_assessment(status: dict[str, object], *, previous: str | None = None, model: str | None = None, timeout: int | None = None) -> None:
     try:
         status["llm_assessment"] = {
@@ -385,6 +404,7 @@ def monitor_loop(
     baseline_path: str | None = None,
     config_path: str | None = None,
     history_path: str | None = None,
+    status_cache_path: str | None = None,
 ) -> None:
     print(f"Monitoring {log_path}")
     print(f"Writing status to {output}")
@@ -397,7 +417,13 @@ def monitor_loop(
         status = build_status(
             log_path, policy_path=policy_path, anomaly_limit=anomaly_limit,
             baseline_path=baseline_path, config_path=config_path, history_path=history_path,
+            status_cache_path=status_cache_path,
         )
+        mcp = status.get("capabilities", {}).get("graylog_mcp", {}) if isinstance(status.get("capabilities"), dict) else {}
+        if status_cache_path and isinstance(mcp, dict) and mcp.get("status") == "error":
+            cached = cached_status_with_error(status_cache_path, mcp)
+            if cached:
+                status = cached
         if effective_llm:
             now = int(time.time())
             if now - last_llm_at >= llm_interval: