feat: 数据面故障韧性 — 节点健康上报(控制面) + 客户端连通看门狗
盲区:节点 status 只看 agent 的 gRPC 在线,sing-box 数据面坏了(崩/配置坏/数据口不通) 而 agent 仍在线时,节点仍显示 up、connect 放行、客户端「已连接」但流量全失败,且不自愈。 控制面: - agent 每心跳探 sing-box clash_api /version(2s 超时,连续 2 次失败才报不健康,防抖), 经新增 HeartbeatRequest.data_plane_healthy 上报(手写 agentv1 契约 + proto 同步;JSON codec)。 - NodeLoad 加 DataPlaneHealthy(随 load 写 Redis;字段缺失默认 healthy 防滚动期误杀)。 - effectiveNodeStatus 扩为 (dbStatus, agentOnline, dataPlaneHealthy);ListNodes 与 ConnectNode 统一改用它 → 数据面坏的节点列表置灰 + connect 返回 404 拦截。 客户端(connection_provider 连通看门狗): - 连上后每 15s 经隧道 HTTP 探海外 generate_204(可注入),连续 3 次失败判当前节点不可用。 - 智能选择 → 自动切到其他最优可用节点重连;手动选定 → 断开并提示「节点异常」(尊重用户选择)。 - 新增 nodeUnhealthySwitched/nodeUnhealthyError 文案。 测试:agent 健康探测防抖、effectiveNodeStatus 真值表、dataPlaneHealthy 助手、看门狗 智能切/手动断/健康不触发;go test ./... 与 flutter test 全绿。 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
wangjia
@@ -6,12 +6,14 @@
|
||||
// - 严禁乐观翻转：VpnPhase.on 必须由 bridge.statusStream 确认后才置。
|
||||
// - 状态来源：bridge.statusStream（来自内核回调），非 Timer 模拟。
|
||||
import 'dart:async';
|
||||
import 'dart:io';
|
||||
|
||||
import 'package:flutter_riverpod/flutter_riverpod.dart';
|
||||
|
||||
import '../bridge/vpn_bridge.dart';
|
||||
import '../bridge/vpn_bridge_provider.dart';
|
||||
import '../l10n/app_text.dart';
|
||||
import '../models/node.dart';
|
||||
import '../services/api_config.dart';
|
||||
import '../services/connect_api.dart';
|
||||
import '../services/device_identity.dart';
|
||||
@@ -53,19 +55,59 @@ class ConnectionState {
|
||||
int get hashCode => Object.hash(phase, elapsed, error);
|
||||
}
|
||||
|
||||
// ── 连通看门狗 ─────────────────────────────────────────────────────
|
||||
//
|
||||
// 已连接(本地 TUN 已起)≠ 远端节点数据面可用:节点 sing-box 崩了/数据口不通而
|
||||
// agent 仍在线时,客户端会一直显示「已连接」但流量全失败。看门狗周期性经隧道做
|
||||
// 连通性探测,连续失败即判当前节点不可用 → 智能选择自动切节点 / 手动选定的只断开提示。
|
||||
|
||||
/// 连通探测接缝:true=经隧道能到外网。默认走海外 204 端点;测试可注入假实现。
|
||||
typedef HealthProber = Future<bool> Function();
|
||||
|
||||
const _kWatchdogInterval = Duration(seconds: 15);
|
||||
const _kHealthFailLimit = 3; // 连续 N 次失败(≈45s)才判定不可用,防抖
|
||||
|
||||
/// 默认探测:GET 海外 generate_204(全局 TUN 下走隧道 → 真实检验代理出网)。
|
||||
/// 任一端点 2xx/204 即视为通;多端点容错,避免单点误判。
|
||||
Future<bool> defaultHealthProbe() async {
|
||||
const urls = [
|
||||
'https://www.gstatic.com/generate_204',
|
||||
'https://cp.cloudflare.com/generate_204',
|
||||
];
|
||||
for (final u in urls) {
|
||||
final client = HttpClient()..connectionTimeout = const Duration(seconds: 4);
|
||||
try {
|
||||
final req = await client.getUrl(Uri.parse(u)).timeout(const Duration(seconds: 5));
|
||||
final resp = await req.close().timeout(const Duration(seconds: 5));
|
||||
await resp.drain<void>();
|
||||
if (resp.statusCode >= 200 && resp.statusCode < 300) return true;
|
||||
} catch (_) {
|
||||
// 该端点不通,试下一个
|
||||
} finally {
|
||||
client.close(force: true);
|
||||
}
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
// ── 状态机 ───────────────────────────────────────────────────────
|
||||
|
||||
class ConnectionController extends StateNotifier<ConnectionState> {
|
||||
ConnectionController(this._ref, this._bridge)
|
||||
: super(const ConnectionState(phase: VpnPhase.off)) {
|
||||
ConnectionController(this._ref, this._bridge, {HealthProber? prober})
|
||||
: _prober = prober ?? defaultHealthProbe,
|
||||
super(const ConnectionState(phase: VpnPhase.off)) {
|
||||
// 订阅桥状态流：状态由内核事件驱动，严禁 UI 乐观翻转。
|
||||
_statusSub = _bridge.statusStream.listen(_onKernelStatus);
|
||||
}
|
||||
|
||||
final Ref _ref;
|
||||
final VpnBridge _bridge;
|
||||
final HealthProber _prober;
|
||||
StreamSubscription<VpnStatus>? _statusSub;
|
||||
Timer? _elapsed;
|
||||
Timer? _watchdog;
|
||||
int _healthFails = 0;
|
||||
bool _probing = false;
|
||||
ConnectApi? _api;
|
||||
|
||||
// ── 公有 API ───────────────────────────────────────────────────
|
||||
@@ -167,16 +209,84 @@ class ConnectionController extends StateNotifier<ConnectionState> {
|
||||
case VpnStatus.on:
|
||||
state = state.copyWith(phase: VpnPhase.on);
|
||||
_startElapsed();
|
||||
_startWatchdog();
|
||||
case VpnStatus.connecting:
|
||||
state = state.copyWith(phase: VpnPhase.connecting);
|
||||
_stopElapsed();
|
||||
_stopWatchdog();
|
||||
case VpnStatus.off:
|
||||
case VpnStatus.error:
|
||||
state = const ConnectionState(phase: VpnPhase.off);
|
||||
_stopElapsed();
|
||||
_stopWatchdog();
|
||||
}
|
||||
}
|
||||
|
||||
// ── 连通看门狗 ───────────────────────────────────────────────
|
||||
void _startWatchdog() {
|
||||
_watchdog?.cancel();
|
||||
_healthFails = 0;
|
||||
_watchdog = Timer.periodic(_kWatchdogInterval, (_) => _checkHealth());
|
||||
}
|
||||
|
||||
void _stopWatchdog() {
|
||||
_watchdog?.cancel();
|
||||
_watchdog = null;
|
||||
_healthFails = 0;
|
||||
_probing = false;
|
||||
}
|
||||
|
||||
Future<void> _checkHealth() async {
|
||||
if (_probing || !mounted || state.phase != VpnPhase.on) return;
|
||||
_probing = true;
|
||||
final ok = await _prober();
|
||||
_probing = false;
|
||||
if (!mounted || state.phase != VpnPhase.on) return;
|
||||
if (ok) {
|
||||
_healthFails = 0;
|
||||
return;
|
||||
}
|
||||
_healthFails++;
|
||||
if (_healthFails >= _kHealthFailLimit) {
|
||||
_stopWatchdog();
|
||||
await _onNodeUnhealthy();
|
||||
}
|
||||
}
|
||||
|
||||
/// 当前节点连续探测失败:智能选择 → 切到其他最优可用节点重连;手动选定 → 断开并提示。
|
||||
Future<void> _onNodeUnhealthy() async {
|
||||
final t = _ref.read(appTextProvider);
|
||||
final smart = _ref.read(selectedNodeCodeProvider) == kSmartNodeCode;
|
||||
final altCode = smart ? _pickAlternativeCode(_ref.read(effectiveNodeProvider).code) : null;
|
||||
if (smart && altCode != null) {
|
||||
_ref.read(selectedNodeCodeProvider.notifier).state = altCode;
|
||||
await _disconnect();
|
||||
await _connect();
|
||||
// 重连进行中给出「已自动切换」提示(connecting 态显示,连上后随 copyWith 清除)。
|
||||
if (mounted && state.phase != VpnPhase.off) {
|
||||
state = ConnectionState(phase: state.phase, error: t.nodeUnhealthySwitched);
|
||||
}
|
||||
return;
|
||||
}
|
||||
// 手动选定节点(或智能模式无其他可用节点):断开并提示,尊重用户选择、不自动换。
|
||||
await _disconnect();
|
||||
if (mounted) state = ConnectionState(phase: VpnPhase.off, error: t.nodeUnhealthyError);
|
||||
}
|
||||
|
||||
/// 选延迟最优、可用(status up)、非当前节点的 code;无则 null。
|
||||
String? _pickAlternativeCode(String excludeCode) {
|
||||
final nodes = (_ref.read(nodesProvider).valueOrNull ?? const [])
|
||||
.where((n) => !n.isDown && n.uuid.isNotEmpty && n.code != excludeCode)
|
||||
.toList();
|
||||
if (nodes.isEmpty) return null;
|
||||
nodes.sort((a, b) {
|
||||
final pa = a.ping > 0 ? a.ping : 1 << 30;
|
||||
final pb = b.ping > 0 ? b.ping : 1 << 30;
|
||||
return pa.compareTo(pb);
|
||||
});
|
||||
return nodes.first.code;
|
||||
}
|
||||
|
||||
void _startElapsed() {
|
||||
_elapsed?.cancel();
|
||||
_elapsed = Timer.periodic(const Duration(seconds: 1), (_) {
|
||||
@@ -195,6 +305,7 @@ class ConnectionController extends StateNotifier<ConnectionState> {
|
||||
void dispose() {
|
||||
_statusSub?.cancel();
|
||||
_stopElapsed();
|
||||
_stopWatchdog();
|
||||
_api?.dispose();
|
||||
super.dispose();
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user