feat(admin): add system observability dashboard with circuit breaker monitoring

Backend: expose circuit breaker status via new AdminObservabilityController
(health, circuit-breakers, redis endpoints). Frontend: new observability
feature in admin-client with auto-refreshing status cards.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

packages/admin-client/src/App.tsx

@@ -11,6 +11,7 @@ import { ConversationsPage } from './features/conversations';
 import { SettingsPage } from './features/settings';
 import { TenantsPage } from './features/tenants';
 import { McpPage } from './features/mcp';
+import { ObservabilityPage } from './features/observability';
 
 function App() {
   return (
@@ -37,6 +38,7 @@ function App() {
         <Route path="conversations" element={<ConversationsPage />} />
         <Route path="tenants" element={<TenantsPage />} />
         <Route path="mcp" element={<McpPage />} />
+        <Route path="observability" element={<ObservabilityPage />} />
         <Route path="settings" element={<SettingsPage />} />
       </Route>
 

packages/admin-client/src/features/observability/application/index.ts

@@ -0,0 +1,6 @@
+export {
+  useSystemHealth,
+  useCircuitBreakers,
+  useRedisStatus,
+  OBSERVABILITY_KEYS,
+} from './useObservability';

packages/admin-client/src/features/observability/application/useObservability.ts

@@ -0,0 +1,33 @@
+import { useQuery } from '@tanstack/react-query';
+import { observabilityApi } from '../infrastructure/observability.api';
+
+export const OBSERVABILITY_KEYS = {
+  all: ['observability'] as const,
+  health: ['observability', 'health'] as const,
+  circuitBreakers: ['observability', 'circuit-breakers'] as const,
+  redis: ['observability', 'redis'] as const,
+};
+
+export function useSystemHealth() {
+  return useQuery({
+    queryKey: OBSERVABILITY_KEYS.health,
+    queryFn: () => observabilityApi.getHealth(),
+    refetchInterval: 30_000,
+  });
+}
+
+export function useCircuitBreakers() {
+  return useQuery({
+    queryKey: OBSERVABILITY_KEYS.circuitBreakers,
+    queryFn: () => observabilityApi.getCircuitBreakers(),
+    refetchInterval: 10_000,
+  });
+}
+
+export function useRedisStatus() {
+  return useQuery({
+    queryKey: OBSERVABILITY_KEYS.redis,
+    queryFn: () => observabilityApi.getRedisStatus(),
+    refetchInterval: 30_000,
+  });
+}

packages/admin-client/src/features/observability/index.ts

@@ -0,0 +1,3 @@
+export { ObservabilityPage } from './presentation/pages/ObservabilityPage';
+export * from './application';
+export type { CircuitBreakerStatus, SystemHealth, RedisStatus } from './infrastructure';

packages/admin-client/src/features/observability/infrastructure/index.ts

@@ -0,0 +1,2 @@
+export { observabilityApi } from './observability.api';
+export type { CircuitBreakerStatus, SystemHealth, RedisStatus } from './observability.api';

packages/admin-client/src/features/observability/infrastructure/observability.api.ts

@@ -0,0 +1,40 @@
+import api from '../../../shared/utils/api';
+
+export interface CircuitBreakerStatus {
+  name: string;
+  state: 'CLOSED' | 'OPEN' | 'HALF_OPEN';
+  failureCount: number;
+  config: {
+    failureThreshold: number;
+    resetTimeoutMs: number;
+  };
+}
+
+export interface SystemHealth {
+  status: 'healthy' | 'degraded';
+  redis: { available: boolean };
+  circuitBreakers: CircuitBreakerStatus[];
+  timestamp: string;
+}
+
+export interface RedisStatus {
+  available: boolean;
+  purpose: string;
+}
+
+export const observabilityApi = {
+  getHealth: async (): Promise<SystemHealth> => {
+    const response = await api.get('/conversations/admin/observability/health');
+    return response.data.data;
+  },
+
+  getCircuitBreakers: async (): Promise<CircuitBreakerStatus[]> => {
+    const response = await api.get('/conversations/admin/observability/circuit-breakers');
+    return response.data.data;
+  },
+
+  getRedisStatus: async (): Promise<RedisStatus> => {
+    const response = await api.get('/conversations/admin/observability/redis');
+    return response.data.data;
+  },
+};

packages/admin-client/src/features/observability/presentation/pages/ObservabilityPage.tsx

@@ -0,0 +1,155 @@
+import { Card, Row, Col, Tag, Typography, Spin, Progress, Descriptions, Badge, Button } from 'antd';
+import {
+  CheckCircleOutlined,
+  CloseCircleOutlined,
+  ExclamationCircleOutlined,
+  ReloadOutlined,
+  DatabaseOutlined,
+  CloudServerOutlined,
+} from '@ant-design/icons';
+import { useQueryClient } from '@tanstack/react-query';
+import { useSystemHealth, useCircuitBreakers, OBSERVABILITY_KEYS } from '../../application';
+import type { CircuitBreakerStatus } from '../../infrastructure';
+
+const { Title, Text } = Typography;
+
+const STATE_CONFIG: Record<string, { color: string; icon: React.ReactNode; label: string }> = {
+  CLOSED: { color: 'success', icon: <CheckCircleOutlined />, label: '正常' },
+  OPEN: { color: 'error', icon: <CloseCircleOutlined />, label: '熔断中' },
+  HALF_OPEN: { color: 'warning', icon: <ExclamationCircleOutlined />, label: '探测恢复' },
+};
+
+function CircuitBreakerCard({ breaker }: { breaker: CircuitBreakerStatus }) {
+  const config = STATE_CONFIG[breaker.state] || STATE_CONFIG.CLOSED;
+  const failurePercent = Math.round((breaker.failureCount / breaker.config.failureThreshold) * 100);
+
+  return (
+    <Card
+      title={
+        <span>
+          <CloudServerOutlined className="mr-2" />
+          {breaker.name}
+        </span>
+      }
+      extra={<Tag color={config.color} icon={config.icon}>{config.label}</Tag>}
+    >
+      <div className="mb-4">
+        <Text type="secondary">失败次数 / 阈值</Text>
+        <Progress
+          percent={failurePercent}
+          format={() => `${breaker.failureCount} / ${breaker.config.failureThreshold}`}
+          status={breaker.state === 'OPEN' ? 'exception' : breaker.state === 'HALF_OPEN' ? 'active' : 'normal'}
+        />
+      </div>
+      <Descriptions size="small" column={1}>
+        <Descriptions.Item label="熔断阈值">{breaker.config.failureThreshold} 次连续失败</Descriptions.Item>
+        <Descriptions.Item label="重置超时">{breaker.config.resetTimeoutMs / 1000}s</Descriptions.Item>
+        <Descriptions.Item label="当前状态">
+          <Badge status={breaker.state === 'CLOSED' ? 'success' : breaker.state === 'OPEN' ? 'error' : 'warning'} text={config.label} />
+        </Descriptions.Item>
+      </Descriptions>
+    </Card>
+  );
+}
+
+export function ObservabilityPage() {
+  const queryClient = useQueryClient();
+  const { data: health, isLoading: loadingHealth } = useSystemHealth();
+  const { data: breakers, isLoading: loadingBreakers } = useCircuitBreakers();
+
+  const handleRefresh = () => {
+    queryClient.invalidateQueries({ queryKey: OBSERVABILITY_KEYS.all });
+  };
+
+  return (
+    <div className="p-6">
+      <div className="flex justify-between items-center mb-6">
+        <Title level={4} className="!mb-0">系统监控</Title>
+        <Button icon={<ReloadOutlined />} onClick={handleRefresh}>
+          刷新
+        </Button>
+      </div>
+
+      {/* Health Overview */}
+      <Spin spinning={loadingHealth}>
+        <Row gutter={[16, 16]} className="mb-6">
+          <Col xs={24} sm={8}>
+            <Card>
+              <div className="text-center">
+                <div className="text-lg mb-2">系统状态</div>
+                {health?.status === 'healthy' ? (
+                  <Tag color="success" icon={<CheckCircleOutlined />} className="text-base px-4 py-1">
+                    健康
+                  </Tag>
+                ) : (
+                  <Tag color="error" icon={<ExclamationCircleOutlined />} className="text-base px-4 py-1">
+                    降级
+                  </Tag>
+                )}
+              </div>
+            </Card>
+          </Col>
+          <Col xs={24} sm={8}>
+            <Card>
+              <div className="text-center">
+                <div className="text-lg mb-2">
+                  <DatabaseOutlined className="mr-1" />
+                  Redis
+                </div>
+                {health?.redis.available ? (
+                  <Tag color="success" icon={<CheckCircleOutlined />} className="text-base px-4 py-1">
+                    已连接
+                  </Tag>
+                ) : (
+                  <Tag color="default" icon={<CloseCircleOutlined />} className="text-base px-4 py-1">
+                    未连接
+                  </Tag>
+                )}
+              </div>
+            </Card>
+          </Col>
+          <Col xs={24} sm={8}>
+            <Card>
+              <div className="text-center">
+                <div className="text-lg mb-2">熔断器</div>
+                <Text>
+                  {health?.circuitBreakers.filter(cb => cb.state === 'CLOSED').length || 0}
+                  {' / '}
+                  {health?.circuitBreakers.length || 0}
+                  {' '}
+                  正常
+                </Text>
+              </div>
+            </Card>
+          </Col>
+        </Row>
+      </Spin>
+
+      {/* Circuit Breaker Details */}
+      <Title level={5} className="mb-4">熔断器状态</Title>
+      <Spin spinning={loadingBreakers}>
+        <Row gutter={[16, 16]}>
+          {breakers?.map((breaker) => (
+            <Col xs={24} md={12} key={breaker.name}>
+              <CircuitBreakerCard breaker={breaker} />
+            </Col>
+          ))}
+          {breakers?.length === 0 && (
+            <Col span={24}>
+              <Card>
+                <Text type="secondary">暂无熔断器数据</Text>
+              </Card>
+            </Col>
+          )}
+        </Row>
+      </Spin>
+
+      {/* Auto-refresh indicator */}
+      <div className="mt-4 text-right">
+        <Text type="secondary" className="text-xs">
+          数据每 10 秒自动刷新 | 最后更新: {health?.timestamp ? new Date(health.timestamp).toLocaleTimeString() : '-'}
+        </Text>
+      </div>
+    </div>
+  );
+}

packages/admin-client/src/shared/components/MainLayout.tsx

@@ -18,6 +18,7 @@ import {
   MessageOutlined,
   ClusterOutlined,
   ApiOutlined,
+  MonitorOutlined,
 } from '@ant-design/icons';
 import { useAuth } from '../hooks/useAuth';
 
@@ -82,6 +83,11 @@ const menuItems: MenuProps['items'] = [
     icon: <ApiOutlined />,
     label: 'MCP 服务器',
   },
+  {
+    key: '/observability',
+    icon: <MonitorOutlined />,
+    label: '系统监控',
+  },
   {
     key: '/settings',
     icon: <SettingOutlined />,

packages/services/conversation-service/src/adapters/inbound/admin-observability.controller.ts

@@ -0,0 +1,99 @@
+import {
+  Controller,
+  Get,
+  Headers,
+  UnauthorizedException,
+} from '@nestjs/common';
+import * as jwt from 'jsonwebtoken';
+import { KnowledgeClientService } from '../../infrastructure/knowledge/knowledge-client.service';
+import { PaymentClientService } from '../../infrastructure/payment/payment-client.service';
+import { RedisClientService } from '../../infrastructure/cache/redis-client.service';
+
+interface AdminPayload {
+  id: string;
+  username: string;
+  role: string;
+}
+
+/**
+ * 管理员可观测性 API — 系统健康、熔断器状态、Redis 状态
+ */
+@Controller('conversations/admin/observability')
+export class AdminObservabilityController {
+  constructor(
+    private readonly knowledgeClient: KnowledgeClientService,
+    private readonly paymentClient: PaymentClientService,
+    private readonly redisClient: RedisClientService,
+  ) {}
+
+  private verifyAdmin(authorization: string): AdminPayload {
+    const token = authorization?.replace('Bearer ', '');
+    if (!token) {
+      throw new UnauthorizedException('Missing token');
+    }
+    try {
+      const secret = process.env.JWT_SECRET || 'your-jwt-secret-key';
+      return jwt.verify(token, secret) as AdminPayload;
+    } catch {
+      throw new UnauthorizedException('Invalid token');
+    }
+  }
+
+  /**
+   * 综合健康检查
+   */
+  @Get('health')
+  getHealth(@Headers('authorization') auth: string) {
+    this.verifyAdmin(auth);
+
+    const circuitBreakers = [
+      this.knowledgeClient.getCircuitBreakerStatus(),
+      this.paymentClient.getCircuitBreakerStatus(),
+    ];
+
+    const allHealthy = circuitBreakers.every(cb => cb.state === 'CLOSED');
+    const redisAvailable = this.redisClient.isAvailable();
+
+    return {
+      success: true,
+      data: {
+        status: allHealthy && redisAvailable ? 'healthy' : 'degraded',
+        redis: { available: redisAvailable },
+        circuitBreakers,
+        timestamp: new Date().toISOString(),
+      },
+    };
+  }
+
+  /**
+   * 熔断器详细状态
+   */
+  @Get('circuit-breakers')
+  getCircuitBreakers(@Headers('authorization') auth: string) {
+    this.verifyAdmin(auth);
+
+    return {
+      success: true,
+      data: [
+        this.knowledgeClient.getCircuitBreakerStatus(),
+        this.paymentClient.getCircuitBreakerStatus(),
+      ],
+    };
+  }
+
+  /**
+   * Redis 连接状态
+   */
+  @Get('redis')
+  getRedisStatus(@Headers('authorization') auth: string) {
+    this.verifyAdmin(auth);
+
+    return {
+      success: true,
+      data: {
+        available: this.redisClient.isAvailable(),
+        purpose: 'Agent loop checkpoint persistence',
+      },
+    };
+  }
+}

packages/services/conversation-service/src/conversation/conversation.module.ts

@@ -16,11 +16,13 @@ import { InternalConversationController } from '../adapters/inbound/internal.con
 import { AdminConversationController } from '../adapters/inbound/admin-conversation.controller';
 import { AdminMcpController } from '../adapters/inbound/admin-mcp.controller';
 import { AdminEvaluationRuleController } from '../adapters/inbound/admin-evaluation-rule.controller';
+import { AdminObservabilityController } from '../adapters/inbound/admin-observability.controller';
 import { ConversationGateway } from '../adapters/inbound/conversation.gateway';
+import { PaymentModule } from '../infrastructure/payment/payment.module';
 
 @Module({
-  imports: [TypeOrmModule.forFeature([ConversationORM, MessageORM, TokenUsageORM, AgentExecutionORM])],
+  imports: [TypeOrmModule.forFeature([ConversationORM, MessageORM, TokenUsageORM, AgentExecutionORM]), PaymentModule],
-  controllers: [ConversationController, InternalConversationController, AdminMcpController, AdminEvaluationRuleController, AdminConversationController],
+  controllers: [ConversationController, InternalConversationController, AdminMcpController, AdminEvaluationRuleController, AdminConversationController, AdminObservabilityController],
   providers: [
     ConversationService,
     ConversationGateway,

packages/services/conversation-service/src/infrastructure/knowledge/knowledge-client.service.ts

@@ -282,4 +282,16 @@ export class KnowledgeClientService implements OnModuleInit {
     if (memories.length === 0) return '';
     return memories.map(m => `[${m.memoryType}] (重要度:${m.importance}) ${m.content}`).join('\n');
   }
+
+  /**
+   * 获取熔断器状态 — 供 Admin API 使用
+   */
+  getCircuitBreakerStatus() {
+    return {
+      name: 'knowledge-service',
+      state: this.circuitBreaker.getState(),
+      failureCount: this.circuitBreaker.getFailureCount(),
+      config: { failureThreshold: 5, resetTimeoutMs: 60_000 },
+    };
+  }
 }

packages/services/conversation-service/src/infrastructure/payment/payment-client.service.ts

@@ -210,4 +210,16 @@ export class PaymentClientService implements OnModuleInit {
     }>;
     return data.success ? data.data : null;
   }
+
+  /**
+   * 获取熔断器状态 — 供 Admin API 使用
+   */
+  getCircuitBreakerStatus() {
+    return {
+      name: 'payment-service',
+      state: this.circuitBreaker.getState(),
+      failureCount: this.circuitBreaker.getFailureCount(),
+      config: { failureThreshold: 3, resetTimeoutMs: 30_000 },
+    };
+  }
 }