diff --git a/admin/dashboard/blueprints/system.py b/admin/dashboard/blueprints/system.py index 37db1ff..1a372d8 100644 --- a/admin/dashboard/blueprints/system.py +++ b/admin/dashboard/blueprints/system.py @@ -42,6 +42,203 @@ def _save_system_yaml(config_obj: dict) -> None: yaml.safe_dump(config_obj, f, allow_unicode=True, sort_keys=False) +def _safe_int(value, default: int = 0) -> int: + """把数据库 / Redis 返回的字符串数字安全转成整数。""" + try: + if value in (None, ""): + return default + return int(float(value)) + except (TypeError, ValueError): + return default + + +def _safe_float(value, default: float = 0.0) -> float: + """把数据库 / Redis 返回的值安全转成浮点数。""" + try: + if value in (None, ""): + return default + return float(value) + except (TypeError, ValueError): + return default + + +def _format_bytes_to_mb(value: int) -> float: + """把字节数转换为 MB,保留两位小数便于首页摘要展示。""" + return round((_safe_float(value, 0.0) / 1024 / 1024), 2) + + +def _extract_mysql_runtime_snapshot(db_manager) -> dict: + """采集 MySQL 运行态摘要。 + + 首页目标不是替代 DBA 工具,而是让管理员一眼判断: + 1. 数据库是不是活着; + 2. 当前连接压力高不高; + 3. 当前库规模是否已经明显变大; + 4. 有没有必要继续深入到更专业的监控页排查。 + """ + snapshot = { + "status": "healthy", + "summary": "连接正常", + "database": db_manager.get_mysql_database_name(), + "version": "", + "threads_connected": 0, + "threads_running": 0, + "max_connections": 0, + "connection_usage_percent": 0.0, + "questions_per_second": 0.0, + "uptime_seconds": 0, + "table_count": 0, + "schema_size_mb": 0.0, + "slow_query_threshold_ms": db_manager.get_slow_query_threshold_ms(), + } + + mysql_conn = db_manager.get_mysql_connection() + try: + with mysql_conn.cursor(dictionary=True) as cursor: + # 基础探活与版本识别: + # 1. SELECT VERSION() 成本极低; + # 2. 相比只做 SELECT 1,它还能顺便拿到版本信息; + # 3. 首页卡片里显示版本,方便线上排查“是不是某台库版本不一致”。 + cursor.execute("SELECT VERSION() AS version, DATABASE() AS database_name") + version_row = cursor.fetchone() or {} + snapshot["version"] = str(version_row.get("version") or "").strip() + snapshot["database"] = str(version_row.get("database_name") or snapshot["database"] or "").strip() + + cursor.execute( + """ + SHOW GLOBAL STATUS + WHERE Variable_name IN ('Threads_connected', 'Threads_running', 'Questions', 'Uptime') + """ + ) + status_rows = cursor.fetchall() or [] + status_map = { + str(row.get("Variable_name") or "").strip(): row.get("Value") + for row in status_rows + } + + cursor.execute( + """ + SHOW GLOBAL VARIABLES + WHERE Variable_name IN ('max_connections') + """ + ) + variable_rows = cursor.fetchall() or [] + variable_map = { + str(row.get("Variable_name") or "").strip(): row.get("Value") + for row in variable_rows + } + + # information_schema 聚合虽然比 SELECT 1 重一点,但仍属于轻量级元信息查询: + # 1. 只在首页 30 秒级刷新一次,成本可接受; + # 2. 能直接给出当前业务库表数量与体量变化; + # 3. 对判断“是不是消息表膨胀导致后台变慢”很有帮助。 + cursor.execute( + """ + SELECT + COUNT(*) AS table_count, + COALESCE(SUM(data_length + index_length), 0) AS schema_size_bytes + FROM information_schema.tables + WHERE table_schema = DATABASE() + """ + ) + schema_row = cursor.fetchone() or {} + + snapshot["threads_connected"] = _safe_int(status_map.get("Threads_connected")) + snapshot["threads_running"] = _safe_int(status_map.get("Threads_running")) + snapshot["max_connections"] = _safe_int(variable_map.get("max_connections")) + snapshot["uptime_seconds"] = _safe_int(status_map.get("Uptime")) + total_questions = _safe_int(status_map.get("Questions")) + if snapshot["uptime_seconds"] > 0: + snapshot["questions_per_second"] = round(total_questions / snapshot["uptime_seconds"], 2) + if snapshot["max_connections"] > 0: + snapshot["connection_usage_percent"] = round( + (snapshot["threads_connected"] / snapshot["max_connections"]) * 100, + 1, + ) + snapshot["table_count"] = _safe_int(schema_row.get("table_count")) + snapshot["schema_size_mb"] = _format_bytes_to_mb(schema_row.get("schema_size_bytes")) + + if snapshot["connection_usage_percent"] >= 80 or snapshot["threads_running"] >= 12: + snapshot["status"] = "warning" + snapshot["summary"] = ( + f"连接压力偏高:已连接 {snapshot['threads_connected']} / {snapshot['max_connections']}," + f"运行中线程 {snapshot['threads_running']}" + ) + else: + snapshot["summary"] = ( + f"连接正常:已连接 {snapshot['threads_connected']} / {snapshot['max_connections'] or '-'}," + f"QPS {snapshot['questions_per_second']}" + ) + return snapshot + except Exception as mysql_error: + snapshot["status"] = "danger" + snapshot["summary"] = f"MySQL 探测失败: {mysql_error}" + return snapshot + finally: + mysql_conn.close() + + +def _extract_redis_runtime_snapshot(db_manager) -> dict: + """采集 Redis 运行态摘要。""" + redis_config = getattr(db_manager, "redis_config", {}) or {} + snapshot = { + "status": "healthy", + "summary": "连接正常", + "db_index": _safe_int(redis_config.get("db", 0)), + "key_count": 0, + "connected_clients": 0, + "blocked_clients": 0, + "ops_per_sec": 0, + "used_memory_human": "", + "used_memory_peak_human": "", + "memory_usage_percent": 0.0, + "uptime_seconds": 0, + "hit_rate_percent": 0.0, + } + + try: + redis_conn = db_manager.get_redis_connection() + redis_conn.ping() + info = redis_conn.info() or {} + snapshot["key_count"] = _safe_int(redis_conn.dbsize()) + snapshot["connected_clients"] = _safe_int(info.get("connected_clients")) + snapshot["blocked_clients"] = _safe_int(info.get("blocked_clients")) + snapshot["ops_per_sec"] = _safe_int(info.get("instantaneous_ops_per_sec")) + snapshot["used_memory_human"] = str(info.get("used_memory_human") or "").strip() + snapshot["used_memory_peak_human"] = str(info.get("used_memory_peak_human") or "").strip() + snapshot["uptime_seconds"] = _safe_int(info.get("uptime_in_seconds")) + + maxmemory = _safe_int(info.get("maxmemory")) + used_memory = _safe_int(info.get("used_memory")) + if maxmemory > 0: + snapshot["memory_usage_percent"] = round((used_memory / maxmemory) * 100, 1) + + keyspace_hits = _safe_int(info.get("keyspace_hits")) + keyspace_misses = _safe_int(info.get("keyspace_misses")) + if (keyspace_hits + keyspace_misses) > 0: + snapshot["hit_rate_percent"] = round( + (keyspace_hits / (keyspace_hits + keyspace_misses)) * 100, + 1, + ) + + if snapshot["blocked_clients"] > 0 or snapshot["memory_usage_percent"] >= 80: + snapshot["status"] = "warning" + snapshot["summary"] = ( + f"缓存压力需关注:keys {snapshot['key_count']}," + f"clients {snapshot['connected_clients']},ops/s {snapshot['ops_per_sec']}" + ) + else: + snapshot["summary"] = ( + f"缓存正常:keys {snapshot['key_count']}," + f"clients {snapshot['connected_clients']},ops/s {snapshot['ops_per_sec']}" + ) + return snapshot + except Exception as redis_error: + snapshot["status"] = "danger" + snapshot["summary"] = f"Redis 探测失败: {redis_error}" + return snapshot + + def _legacy_llm_to_catalog(legacy_llm: dict) -> dict: """把旧 llm(backends/scenes) 结构转换为新目录结构(仅用于兜底展示)。 @@ -405,31 +602,11 @@ def api_system_health_summary(): _, recent_error_count = server.stats_db.get_error_logs(days=1, page=1, limit=1) # 基础设施健康: - # 1. MySQL 用最轻量的 SELECT 1 做可用性探测; - # 2. Redis 用 PING 验证连接池当前是否可拿到可用连接; + # 1. MySQL / Redis 都在这里做“首页摘要级”探测,而不是完整深度巡检; + # 2. 除了连通性,还补充少量负载指标,方便管理员快速判断是否需要继续下钻; # 3. 即使探测失败也只反馈到看板,不影响主接口整体返回。 - mysql_status = "healthy" - mysql_summary = "连接正常" - try: - mysql_conn = server.db_manager.get_mysql_connection() - try: - with mysql_conn.cursor() as cursor: - cursor.execute("SELECT 1") - cursor.fetchone() - finally: - mysql_conn.close() - except Exception as mysql_error: - mysql_status = "danger" - mysql_summary = f"MySQL 探测失败: {mysql_error}" - - redis_status = "healthy" - redis_summary = "连接正常" - try: - redis_conn = server.db_manager.get_redis_connection() - redis_conn.ping() - except Exception as redis_error: - redis_status = "danger" - redis_summary = f"Redis 探测失败: {redis_error}" + mysql_snapshot = _extract_mysql_runtime_snapshot(server.db_manager) + redis_snapshot = _extract_redis_runtime_snapshot(server.db_manager) # md2img 健康快照已经有现成实现,这里只做聚合,不主动预热运行时。 md2img_snapshot = get_md2img_health_snapshot(ensure_runtime=False) or {} @@ -524,20 +701,22 @@ def api_system_health_summary(): "summary": error_summary, }, "infrastructure": { - "status": "healthy" if mysql_status == "healthy" and redis_status == "healthy" else "danger", + "status": ( + "danger" + if "danger" in {mysql_snapshot.get("status"), redis_snapshot.get("status")} + else ("warning" if "warning" in {mysql_snapshot.get("status"), redis_snapshot.get("status")} else "healthy") + ), "summary": ( "MySQL / Redis 均正常" - if mysql_status == "healthy" and redis_status == "healthy" - else "存在基础设施连接异常" + if mysql_snapshot.get("status") == "healthy" and redis_snapshot.get("status") == "healthy" + else ( + "基础设施连接正常,但部分负载指标需要关注" + if mysql_snapshot.get("status") != "danger" and redis_snapshot.get("status") != "danger" + else "存在基础设施连接异常" + ) ), - "mysql": { - "status": mysql_status, - "summary": mysql_summary, - }, - "redis": { - "status": redis_status, - "summary": redis_summary, - }, + "mysql": mysql_snapshot, + "redis": redis_snapshot, }, "ai_runtime": { "status": ai_status, diff --git a/admin/dashboard/templates/index.html b/admin/dashboard/templates/index.html index d14657d..84344f1 100644 --- a/admin/dashboard/templates/index.html +++ b/admin/dashboard/templates/index.html @@ -148,6 +148,29 @@