HIHUO HANDBOOK服务访问清单
HHiHuo约 7 分钟
本文目录
AIForge 平台 (新)集群访问节点 SSHWeb 服务Ingress 路由集群内部服务存储Gitea SSH (Git 推送)已部署组件清单ML Platform端口分配表所有对外暴露的服务端口和凭据，集中管理。
| 项目 | 值 |
|---|
| 前端控制台 | http://107.148.176.193:30801 |
| Gateway API | http://107.148.176.193:30800 |
| 注册测试用户 | admin@aiforge.dev / aiforge123 |
| OpenAI 兼容 | http://107.148.176.193:30800/v1/chat/completions |
| Playground | http://107.148.176.193:30801/playground |
| 注册 | POST /api/v1/auth/register |
| 登录 | POST /api/v1/auth/login |
| 项目 | 值 |
|---|
| API Server | https://107.148.176.193:6443 |
| kubeconfig (本机) | ~/.kube/config-k8s-lab |
| 使用方式 | export KUBECONFIG=~/.kube/config-k8s-lab && kubectl get nodes |
| 角色 | IP | SSH | WireGuard IP |
|---|
| Master | 107.148.176.193 | ssh root@107.148.176.193 | 10.10.0.1 |
| Worker-1 | 107.148.164.118 | ssh root@107.148.164.118 | 10.10.0.2 |
| Worker-2 | 154.9.27.60 | ssh root@154.9.27.60 | 10.10.0.3 |
| Worker-3 | 38.76.221.17 | ssh root@38.76.221.17 | 10.10.0.4 |
| Worker-4 | 154.219.104.66 | ssh root@154.219.104.66 | 10.10.0.5 |
通过任意节点的公网 IP + NodePort 访问。推荐用 Master IP 107.148.176.193。
| 服务 | URL | 用户名 | 密码 | 备注 |
|---|
| Grafana | http://107.148.176.193:30300 | admin | XXALH7sLZeubHVTofib3E3U9n6VgWTv0fbBq7KZ1 | 监控仪表盘 |
| Harbor | http://107.148.176.193:30180 | admin | Harbor12345 | 镜像仓库 |
| ArgoCD | http://107.148.176.193:30880 | admin | oJAKbz26bMzRCfTN | GitOps CI/CD |
| Gitea | http://107.148.176.193:30500 | admin | Gitea12345 | 自托管 Git (NFS I/O 慢，PostgreSQL 需迁移到 local PV) |
| Ingress HTTP | http://154.9.27.60:30080 | — | — | 需要 Host header |
| Ingress HTTPS | https://154.9.27.60:30443 | — | — | 需要 Host header |
| Nginx Demo | http://107.148.176.193:31281 | — | — | Phase 1 练习 |
curl -H "Host: app.k8s-lab.local" http://154.9.27.60:30080/
只能从集群内部（Pod 中或节点上）访问：
| 服务 | ClusterIP | DNS |
|---|
| Prometheus | 10.x.x.x:9090 | monitoring-kube-prometheus-prometheus.monitoring.svc:9090 |
| Alertmanager | 10.x.x.x:9093 | monitoring-kube-prometheus-alertmanager.monitoring.svc:9093 |
| CoreDNS | 10.96.0.10:53 | — |
| Kubernetes API | 10.96.0.1:443 | kubernetes.default.svc |
| 项目 | 值 |
|---|
| NFS Server | Worker-4 (10.10.0.5 / 154.219.104.66) |
| NFS 路径 | /srv/nfs/k8s |
| StorageClass | nfs-dynamic（默认） |
| 查看 PVC | kubectl get pvc --all-namespaces |
git remote add gitea ssh://git@107.148.176.193:30022/admin/my-app.git
| 组件 | Namespace | 状态 | 存储 |
|---|
| Calico CNI | kube-system | Running | — |
| metrics-server | kube-system | Running | — |
| NFS Provisioner | kube-system | Running | Worker-4 NFS |
| Prometheus + Alertmanager | monitoring | Running | 10Gi + 2Gi NFS |
| Grafana | monitoring | Running | 5Gi NFS |
| Loki + Promtail | monitoring | Running | 10Gi NFS |
| node-exporter | monitoring | DaemonSet 5 节点 | — |
| kube-state-metrics | monitoring | Running | — |
| nginx-ingress | ingress-nginx | Running | — |
| Harbor | harbor | Running | 20Gi+ NFS |
| ArgoCD | argocd | Running | — |
| Gitea | gitea | Starting | 5Gi NFS |
| 项目 | 值 |
|---|
| 推理 API | kubectl -n ml-platform port-forward svc/housing-model-v1-svc 8080:8080 |
| 预测请求 | curl -X POST http://localhost:8080/predict -d '{"features":[8.3,20,6.0,1.0,1500,3.0,37.78,-122.42]}' |
| 模型信息 | curl http://localhost:8080/model/info |
| Prometheus 指标 | curl http://localhost:8080/metrics |
| 查看 MLModel | kubectl -n ml-platform get mlmodel |
| 训练日志 | kubectl -n ml-platform logs job/ml-training-v1 |
| Operator 日志 | kubectl -n ml-platform logs deploy/ml-operator |
| 端口 | 服务 | 协议 |
|---|
| 6443 | K8s API Server | HTTPS |
| 30080 | Ingress HTTP | HTTP |
| 30180 | Harbor | HTTP |
| 30300 | Grafana | HTTP |
| 30443 | Ingress HTTPS | HTTPS |
| 30500 | Gitea Web | HTTP |
| 30022 | Gitea SSH | SSH |
| 30880 | ArgoCD HTTP | HTTP |
| 30881 | ArgoCD HTTPS | HTTPS |
| 31281 | Nginx Demo | HTTP |
| 51820 | WireGuard | UDP |
HiHuo