debug

domain-api/app/services/runtime_control_service.py

@@ -5,6 +5,7 @@ import subprocess
 from pathlib import Path
 
 from app.core.config import settings
+from app.services.debug_event_service import push_debug_event
 from app.services.sync_push_service import pull_detect_task_batch_now, push_runtime_projection_now
 from app.services.runtime_settings_service import get_runtime_settings
 from app.services.worker_control_service import _run_systemctl, start_worker, stop_worker
@@ -62,6 +63,23 @@ def _run_systemd_action(service_name: str, action: str, *, no_block: bool = Fals
     return True, f"{service_name} {action} 命令已发送"
 
 
+def _emit_runtime_action_event(action: str, *, stage: str, ok: bool, message: str, data: dict | None = None) -> None:
+    try:
+        push_debug_event(
+            service="runtime-control",
+            event_type=f"runtime_action_{stage}",
+            level="info" if ok else "warning",
+            message=message,
+            payload={
+                "action": action,
+                "ok": ok,
+                **(data or {}),
+            },
+        )
+    except Exception:
+        pass
+
+
 def start_sync_agent() -> tuple[bool, str]:
     runtime = get_runtime_settings()
     worker_mode = runtime.get("worker_mode", settings.worker_mode)
@@ -81,39 +99,79 @@ def stop_sync_agent() -> tuple[bool, str]:
 
 
 def runtime_action(action: str) -> tuple[bool, str, dict]:
-    if action == "start_worker":
+    normalized_action = str(action or "").strip().lower().replace("-", "_")
+    if normalized_action != "push_debug_probe":
+        _emit_runtime_action_event(
+            normalized_action,
+            stage="requested",
+            ok=True,
+            message=f"收到运行时动作请求: {normalized_action}",
+        )
+    if normalized_action == "start_worker":
         ok, message = start_worker()
-        return ok, message, {"action": action, "poll_after_seconds": 2, "refresh_runtime": True}
-    if action == "stop_worker":
+        result = {"action": normalized_action, "poll_after_seconds": 2, "refresh_runtime": True}
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    if normalized_action == "stop_worker":
         ok, message = stop_worker()
-        return ok, message, {"action": action, "poll_after_seconds": 2, "refresh_runtime": True}
-    if action == "restart_api":
+        result = {"action": normalized_action, "poll_after_seconds": 2, "refresh_runtime": True}
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    if normalized_action == "restart_api":
         ok, message = restart_api()
-        return ok, message, {"action": action, "poll_after_seconds": 4, "refresh_runtime": True}
-    if action == "start_sync_agent":
+        result = {"action": normalized_action, "poll_after_seconds": 4, "refresh_runtime": True}
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    if normalized_action == "start_sync_agent":
         ok, message = start_sync_agent()
-        return ok, message, {"action": action, "poll_after_seconds": 2, "refresh_runtime": True}
-    if action == "stop_sync_agent":
+        result = {"action": normalized_action, "poll_after_seconds": 2, "refresh_runtime": True}
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    if normalized_action == "stop_sync_agent":
         ok, message = stop_sync_agent()
-        return ok, message, {"action": action, "poll_after_seconds": 2, "refresh_runtime": True}
-    if action == "push_sync":
+        result = {"action": normalized_action, "poll_after_seconds": 2, "refresh_runtime": True}
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    if normalized_action == "push_sync":
         ok, message, data = push_runtime_projection_now()
-        return ok, message, {
-            "action": action,
+        result = {
+            "action": normalized_action,
             "poll_after_seconds": 2,
             "refresh_runtime": True,
             **(data or {}),
         }
-    if action == "pull_tasks":
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    if normalized_action == "pull_tasks":
         ok, message, data = pull_detect_task_batch_now()
-        return ok, message, {
-            "action": action,
+        result = {
+            "action": normalized_action,
             "poll_after_seconds": 2,
             "refresh_runtime": True,
             **(data or {}),
         }
-    return False, f"不支持的运行时动作: {action}", {
-        "action": action,
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    if normalized_action == "push_debug_probe":
+        ok, message, data = push_debug_event(
+            service="domain-api",
+            event_type="debug_probe",
+            level="info",
+            message="manual debug probe",
+            payload={"node_code": settings.node_code, "node_region": settings.node_region},
+        )
+        result = {
+            "action": normalized_action,
+            "poll_after_seconds": 1,
+            "refresh_runtime": False,
+            **(data or {}),
+        }
+        _emit_runtime_action_event(normalized_action, stage="finished", ok=ok, message=message, data=result)
+        return ok, message, result
+    result = {
+        "action": normalized_action,
         "poll_after_seconds": 0,
         "refresh_runtime": False,
     }
+    _emit_runtime_action_event(normalized_action, stage="finished", ok=False, message=f"不支持的运行时动作: {action}", data=result)
+    return False, f"不支持的运行时动作: {action}", result