添加qiming-rcoder模块
This commit is contained in:
Codex
959
qiming-rcoder/specs/docker-pod-info/0001-docker-pod-info.md
Normal file
959
qiming-rcoder/specs/docker-pod-info/0001-docker-pod-info.md
Normal file
@@ -0,0 +1,959 @@
|
||||
# Docker Pod 信息接口设计文档
|
||||
|
||||
> **文档版本**: v1.0.1
|
||||
> **创建日期**: 2024-12-16
|
||||
> **更新日期**: 2024-12-16
|
||||
> **状态**: 草案
|
||||
> **相关模块**: `crates/rcoder/src/router.rs`, `crates/docker_manager`, `crates/shared_types`
|
||||
|
||||
---
|
||||
|
||||
## 1. 需求概述
|
||||
|
||||
### 1.1 背景
|
||||
|
||||
RCoder 系统采用动态容器管理模式，可以根据 `user_id` 和 `project_id` 创建和管理 Docker 容器（特别是 `ComputerAgentRunner` 类型的容器）。为了便于系统监控和前端用户使用 noVNC 远程虚拟桌面，需要新增三个接口：
|
||||
|
||||
1. **获取当前容器数量接口** - 用于监控系统中已创建的容器总数
|
||||
2. **启动容器接口** - 根据 `user_id` 和 `project_id` 启动（或获取已存在的）容器，**仅启动容器本身，不启动 Agent 服务**，以便用户可以通过 noVNC 访问远程虚拟桌面
|
||||
3. **容器保活接口** - 刷新容器的最后活动时间，防止容器被定时清理任务销毁
|
||||
|
||||
> [!IMPORTANT]
|
||||
> 所有接口响应统一使用 `shared_types::HttpResult<T>` 结构进行包装，保持与系统其他 API 的一致性。
|
||||
|
||||
### 1.2 目标
|
||||
|
||||
| 目标 | 描述 |
|
||||
|------|------|
|
||||
| **监控能力** | 提供容器数量统计，便于系统运维监控 |
|
||||
| **用户体验** | 简化容器启动流程，自动化处理容器存在性检查 |
|
||||
| **简洁响应** | 响应只包含容器基本信息，VNC 访问通过独立接口获取 |
|
||||
| **轻量启动** | 仅启动容器，不启动 Agent 服务，减少资源占用 |
|
||||
| **容器保活** | 支持刷新容器活动时间，防止被自动清理 |
|
||||
|
||||
---
|
||||
|
||||
## 2. HttpResult 响应包装
|
||||
|
||||
### 2.1 HttpResult 结构 (来自 `shared_types`)
|
||||
|
||||
所有接口响应均使用 `HttpResult<T>` 包装：
|
||||
|
||||
```rust
|
||||
// 来自 crates/shared_types/src/model/http_result.rs
|
||||
|
||||
#[derive(Debug, Deserialize, ToSchema)]
|
||||
pub struct HttpResult<T> {
|
||||
/// 响应码: "0000" 表示成功
|
||||
pub code: String,
|
||||
|
||||
/// 响应消息
|
||||
pub message: String,
|
||||
|
||||
/// 响应数据 (成功时有值)
|
||||
pub data: Option<T>,
|
||||
|
||||
/// OpenTelemetry Trace ID (用于链路追踪)
|
||||
pub tid: Option<String>,
|
||||
|
||||
/// 是否成功 (code == "0000")
|
||||
#[serde(skip)]
|
||||
pub success: bool,
|
||||
}
|
||||
|
||||
impl<T> HttpResult<T> {
|
||||
/// 创建成功响应
|
||||
pub fn success(data: T) -> Self;
|
||||
|
||||
/// 创建错误响应
|
||||
pub fn error(code: &str, message: &str) -> Self;
|
||||
|
||||
/// 创建内部错误响应 (code: "5000")
|
||||
pub fn internal_error(message: &str) -> Self;
|
||||
}
|
||||
```
|
||||
|
||||
### 2.2 响应格式示例
|
||||
|
||||
**成功响应:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "0000",
|
||||
"message": "成功",
|
||||
"data": { ... },
|
||||
"tid": "abc123def456...",
|
||||
"success": true
|
||||
}
|
||||
```
|
||||
|
||||
**错误响应:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "5000",
|
||||
"message": "获取 DockerManager 失败",
|
||||
"data": null,
|
||||
"tid": "abc123def456...",
|
||||
"success": false
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 3. 接口设计
|
||||
|
||||
### 3.1 接口一：获取当前容器数量
|
||||
|
||||
#### 3.1.1 接口规格
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **路径** | `GET /computer/pod/count` |
|
||||
| **方法** | GET |
|
||||
| **入参** | 无 |
|
||||
| **响应类型** | `HttpResult<PodCountResponse>` |
|
||||
| **标签** | `pod` (Pod 容器管理接口) |
|
||||
|
||||
#### 3.1.2 请求示例
|
||||
|
||||
```bash
|
||||
curl -X GET http://localhost:8087/computer/pod/count
|
||||
```
|
||||
|
||||
#### 3.1.3 响应数据结构 (data 字段)
|
||||
|
||||
```rust
|
||||
/// 容器数量响应结构
|
||||
#[derive(Debug, Clone, Serialize, Deserialize, ToSchema)]
|
||||
pub struct PodCountResponse {
|
||||
/// 当前运行的容器总数
|
||||
#[schema(example = 5)]
|
||||
pub total_count: u32,
|
||||
|
||||
/// 按服务类型分类的容器数量
|
||||
pub by_service_type: PodCountByServiceType,
|
||||
|
||||
/// 统计时间戳 (Unix 毫秒)
|
||||
#[schema(example = 1702700000000)]
|
||||
pub timestamp: u64,
|
||||
}
|
||||
|
||||
/// 按服务类型分类的容器数量
|
||||
#[derive(Debug, Clone, Serialize, Deserialize, ToSchema)]
|
||||
pub struct PodCountByServiceType {
|
||||
/// RCoder 类型容器数量
|
||||
#[schema(example = 2)]
|
||||
pub rcoder: u32,
|
||||
|
||||
/// ComputerAgentRunner 类型容器数量
|
||||
#[schema(example = 3)]
|
||||
pub computer_agent_runner: u32,
|
||||
}
|
||||
```
|
||||
|
||||
#### 3.1.4 完整响应示例
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "0000",
|
||||
"message": "成功",
|
||||
"data": {
|
||||
"total_count": 5,
|
||||
"by_service_type": {
|
||||
"rcoder": 2,
|
||||
"computer_agent_runner": 3
|
||||
},
|
||||
"timestamp": 1702700000000
|
||||
},
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": true
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
### 3.2 接口二：启动容器 (Ensure Container)
|
||||
|
||||
#### 3.2.1 接口规格
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **路径** | `POST /computer/pod/ensure` |
|
||||
| **方法** | POST |
|
||||
| **入参** | `user_id` (必填), `project_id` (必填) |
|
||||
| **响应类型** | `HttpResult<EnsurePodResponse>` |
|
||||
| **标签** | `pod` (Pod 容器管理接口) |
|
||||
|
||||
> [!NOTE]
|
||||
> 此接口采用 **幂等设计**：如果对应的容器已经存在且处于运行状态，则直接返回现有容器信息，不会创建新容器。
|
||||
|
||||
> [!IMPORTANT]
|
||||
> **此接口仅启动容器，不启动 Agent 服务。** 容器启动后，用户可直接通过 noVNC 访问虚拟桌面，无需等待 Agent 服务初始化。如需使用 AI Agent 功能，请调用 `/computer/chat` 接口。
|
||||
|
||||
#### 3.2.2 请求结构
|
||||
|
||||
```rust
|
||||
/// 启动容器请求结构
|
||||
#[derive(Debug, Clone, Deserialize, ToSchema)]
|
||||
pub struct EnsurePodRequest {
|
||||
/// 用户唯一标识符
|
||||
#[schema(example = "user_123")]
|
||||
pub user_id: String,
|
||||
|
||||
/// 项目唯一标识符
|
||||
#[schema(example = "proj_456")]
|
||||
pub project_id: String,
|
||||
|
||||
/// 可选的资源限制配置
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub resource_limits: Option<PodResourceLimits>,
|
||||
}
|
||||
|
||||
/// Pod 资源限制配置
|
||||
#[derive(Debug, Clone, Deserialize, Serialize, ToSchema)]
|
||||
pub struct PodResourceLimits {
|
||||
/// 内存限制 (bytes), 例如 4GB = 4294967296
|
||||
#[schema(example = 4294967296)]
|
||||
pub memory: Option<u64>,
|
||||
|
||||
/// CPU 份额 (1024 = 1 核)
|
||||
#[schema(example = 2048)]
|
||||
pub cpu_shares: Option<u64>,
|
||||
}
|
||||
```
|
||||
|
||||
#### 3.2.3 请求示例
|
||||
|
||||
```bash
|
||||
curl -X POST http://localhost:8087/computer/pod/ensure \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"user_id": "user_123",
|
||||
"project_id": "proj_456"
|
||||
}'
|
||||
```
|
||||
|
||||
#### 3.2.4 响应数据结构 (data 字段)
|
||||
|
||||
```rust
|
||||
/// 启动容器响应结构
|
||||
#[derive(Debug, Clone, Serialize, ToSchema)]
|
||||
pub struct EnsurePodResponse {
|
||||
/// 容器是否为新创建 (false 表示已存在)
|
||||
pub created: bool,
|
||||
|
||||
/// 容器基本信息
|
||||
pub container_info: PodContainerInfo,
|
||||
|
||||
/// 提示消息
|
||||
#[schema(example = "容器已就绪")]
|
||||
pub message: String,
|
||||
}
|
||||
|
||||
/// 容器基本信息
|
||||
#[derive(Debug, Clone, Serialize, ToSchema)]
|
||||
pub struct PodContainerInfo {
|
||||
/// 容器 ID
|
||||
#[schema(example = "abc123def456")]
|
||||
pub container_id: String,
|
||||
|
||||
/// 容器名称
|
||||
#[schema(example = "computer-agent-runner-user_123")]
|
||||
pub container_name: String,
|
||||
|
||||
/// 容器 IP 地址 (内部网络)
|
||||
#[schema(example = "172.17.0.5")]
|
||||
pub container_ip: String,
|
||||
|
||||
/// 服务 URL
|
||||
#[schema(example = "http://172.17.0.5:8086")]
|
||||
pub service_url: String,
|
||||
|
||||
/// 容器状态
|
||||
#[schema(example = "running")]
|
||||
pub status: String,
|
||||
}
|
||||
```
|
||||
|
||||
#### 3.2.5 完整响应示例
|
||||
|
||||
**成功响应（新创建容器）:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "0000",
|
||||
"message": "成功",
|
||||
"data": {
|
||||
"created": true,
|
||||
"container_info": {
|
||||
"container_id": "abc123def456...",
|
||||
"container_name": "computer-agent-runner-user_123",
|
||||
"container_ip": "172.17.0.5",
|
||||
"service_url": "http://172.17.0.5:8086",
|
||||
"status": "running"
|
||||
},
|
||||
"message": "容器创建成功（Agent 服务未启动）"
|
||||
},
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": true
|
||||
}
|
||||
```
|
||||
|
||||
**成功响应（容器已存在）:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "0000",
|
||||
"message": "成功",
|
||||
"data": {
|
||||
"created": false,
|
||||
"container_info": {
|
||||
"container_id": "abc123def456...",
|
||||
"container_name": "computer-agent-runner-user_123",
|
||||
"container_ip": "172.17.0.5",
|
||||
"service_url": "http://172.17.0.5:8086",
|
||||
"status": "running"
|
||||
},
|
||||
"message": "容器已存在"
|
||||
},
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": true
|
||||
}
|
||||
```
|
||||
|
||||
**错误响应:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "5000",
|
||||
"message": "启动容器失败: Docker 连接超时",
|
||||
"data": null,
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": false
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
### 3.3 接口三：容器保活 (Keepalive)
|
||||
|
||||
#### 3.3.1 接口规格
|
||||
|
||||
| 属性 | 值 |
|
||||
|------|-----|
|
||||
| **路径** | `POST /computer/pod/keepalive` |
|
||||
| **方法** | POST |
|
||||
| **入参** | `user_id` (必填), `project_id` (必填) |
|
||||
| **响应类型** | `HttpResult<KeepalivePodResponse>` |
|
||||
| **标签** | `pod` (Pod 容器管理接口) |
|
||||
|
||||
> [!NOTE]
|
||||
> 此接口用于防止容器被定时清理任务销毁。系统会定期（默认每 5 分钟）检查容器的最后活动时间，闲置超过 30 分钟的容器将被自动清理。
|
||||
|
||||
> [!TIP]
|
||||
> **使用场景**：前端在用户使用 noVNC 远程桌面时，应定期调用此接口（建议每 5-10 分钟一次）来保持容器活跃状态。
|
||||
|
||||
#### 3.3.2 请求结构
|
||||
|
||||
```rust
|
||||
/// 容器保活请求结构
|
||||
#[derive(Debug, Clone, Deserialize, ToSchema)]
|
||||
pub struct KeepalivePodRequest {
|
||||
/// 用户唯一标识符
|
||||
#[schema(example = "user_123")]
|
||||
pub user_id: String,
|
||||
|
||||
/// 项目唯一标识符
|
||||
#[schema(example = "proj_456")]
|
||||
pub project_id: String,
|
||||
}
|
||||
```
|
||||
|
||||
#### 3.3.3 请求示例
|
||||
|
||||
```bash
|
||||
curl -X POST http://localhost:8087/computer/pod/keepalive \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"user_id": "user_123",
|
||||
"project_id": "proj_456"
|
||||
}'
|
||||
```
|
||||
|
||||
#### 3.3.4 响应数据结构 (data 字段)
|
||||
|
||||
```rust
|
||||
/// 容器保活响应结构
|
||||
#[derive(Debug, Clone, Serialize, ToSchema)]
|
||||
pub struct KeepalivePodResponse {
|
||||
/// 容器是否存在 (true=已存在, false=新创建)
|
||||
pub existed: bool,
|
||||
|
||||
/// 容器是否为新创建 (当 existed=false 时为 true)
|
||||
pub created: bool,
|
||||
|
||||
/// 容器基本信息
|
||||
pub container_info: PodContainerInfo,
|
||||
|
||||
/// 上次活动时间 (Unix 毫秒时间戳, 更新前)
|
||||
#[schema(example = 1702700000000)]
|
||||
pub previous_activity_time: u64,
|
||||
|
||||
/// 当前活动时间 (Unix 毫秒时间戳, 更新后)
|
||||
#[schema(example = 1702700600000)]
|
||||
pub current_activity_time: u64,
|
||||
|
||||
/// 距离下次清理的剩余时间 (秒)
|
||||
/// 基于默认 30 分钟闲置超时计算
|
||||
#[schema(example = 1800)]
|
||||
pub time_until_cleanup: u64,
|
||||
|
||||
/// 提示消息
|
||||
#[schema(example = "容器活动时间已刷新")]
|
||||
pub message: String,
|
||||
}
|
||||
```
|
||||
|
||||
#### 3.3.5 完整响应示例
|
||||
|
||||
**成功响应（容器已存在，刷新活动时间）:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "0000",
|
||||
"message": "成功",
|
||||
"data": {
|
||||
"existed": true,
|
||||
"created": false,
|
||||
"container_info": {
|
||||
"container_id": "abc123def456...",
|
||||
"container_name": "computer-agent-runner-user_123",
|
||||
"container_ip": "172.17.0.5",
|
||||
"service_url": "http://172.17.0.5:8086",
|
||||
"status": "running"
|
||||
},
|
||||
"previous_activity_time": 1702700000000,
|
||||
"current_activity_time": 1702700600000,
|
||||
"time_until_cleanup": 1800,
|
||||
"message": "容器活动时间已刷新，距离自动清理还有 30 分钟"
|
||||
},
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": true
|
||||
}
|
||||
```
|
||||
|
||||
**成功响应（容器不存在，自动创建）:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "0000",
|
||||
"message": "成功",
|
||||
"data": {
|
||||
"existed": false,
|
||||
"created": true,
|
||||
"container_info": {
|
||||
"container_id": "xyz789...",
|
||||
"container_name": "computer-agent-runner-user_123",
|
||||
"container_ip": "172.17.0.6",
|
||||
"service_url": "http://172.17.0.6:8086",
|
||||
"status": "running"
|
||||
},
|
||||
"previous_activity_time": 0,
|
||||
"current_activity_time": 1702700600000,
|
||||
"time_until_cleanup": 1800,
|
||||
"message": "容器已自动创建，距离自动清理还有 30 分钟"
|
||||
},
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": true
|
||||
}
|
||||
```
|
||||
|
||||
**错误响应:**
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "5000",
|
||||
"message": "刷新活动时间失败: 容器状态异常",
|
||||
"data": null,
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": false
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 4. 架构设计
|
||||
|
||||
### 4.1 整体架构
|
||||
|
||||
```
|
||||
┌─────────────────────────────────────────────────────────────────┐
|
||||
│ 外部客户端 │
|
||||
└────────────────────────────┬────────────────────────────────────┘
|
||||
│ HTTP
|
||||
▼
|
||||
┌─────────────────────────────────────────────────────────────────┐
|
||||
│ rcoder (主服务) │
|
||||
│ ┌─────────────────────────────────────────────────────────┐ │
|
||||
│ │ router.rs │ │
|
||||
│ │ /computer/pod/count ─────▶ pod_count_handler │ │
|
||||
│ │ /computer/pod/ensure ─────▶ pod_ensure_handler │ │
|
||||
│ │ /computer/pod/keepalive─────▶ pod_keepalive_handler │ │
|
||||
│ └───────────────────────────┬─────────────────────────────┘ │
|
||||
│ │ │
|
||||
│ ┌───────────────────────────▼─────────────────────────────┐ │
|
||||
│ │ service/pod_service.rs │ │
|
||||
│ │ ┌─────────────────┐ ┌──────────────────────────────┐ │ │
|
||||
│ │ │ PodService │ │ PodServiceTrait (trait) │ │ │
|
||||
│ │ │ - get_count() │ │ - get_pod_count() │ │ │
|
||||
│ │ │ - ensure_pod() │ │ - ensure_pod() │ │ │
|
||||
│ │ └─────────────────┘ └──────────────────────────────┘ │ │
|
||||
│ └───────────────────────────┬─────────────────────────────┘ │
|
||||
│ │ │
|
||||
│ ┌───────────────────────────▼─────────────────────────────┐ │
|
||||
│ │ docker_manager (crate) │ │
|
||||
│ │ DockerManager │ │
|
||||
│ │ - list_containers() │ │
|
||||
│ │ - get_agent_info() │ │
|
||||
│ │ - start_agent_container() │ │
|
||||
│ └───────────────────────────┬─────────────────────────────┘ │
|
||||
└──────────────────────────────┼──────────────────────────────────┘
|
||||
│ Docker API
|
||||
▼
|
||||
┌─────────────────────────────────────────────────────────────────┐
|
||||
│ Docker Daemon │
|
||||
│ ┌────────────────┐ ┌────────────────┐ ┌────────────────┐ │
|
||||
│ │ container-1 │ │ container-2 │ │ container-3 │ │
|
||||
│ │ (user_123) │ │ (user_456) │ │ (project_xyz) │ │
|
||||
│ └────────────────┘ └────────────────┘ └────────────────┘ │
|
||||
└─────────────────────────────────────────────────────────────────┘
|
||||
```
|
||||
|
||||
### 4.2 模块职责
|
||||
|
||||
| 模块 | 职责 |
|
||||
|------|------|
|
||||
| `handler/pod_handler.rs` | 处理 HTTP 请求，参数解析与验证 |
|
||||
| `service/pod_service.rs` | 核心业务逻辑，调用 DockerManager |
|
||||
| `docker_manager` | Docker API 交互，容器生命周期管理 |
|
||||
|
||||
---
|
||||
|
||||
## 5. Trait 与结构体定义
|
||||
|
||||
### 5.1 Service Trait
|
||||
|
||||
```rust
|
||||
// crates/rcoder/src/service/pod_service.rs
|
||||
|
||||
use async_trait::async_trait;
|
||||
use crate::AppError;
|
||||
|
||||
/// Pod 服务 Trait
|
||||
///
|
||||
/// 定义 Pod 容器管理的核心能力接口
|
||||
#[async_trait]
|
||||
pub trait PodServiceTrait: Send + Sync {
|
||||
/// 获取当前容器数量统计
|
||||
///
|
||||
/// # 返回
|
||||
/// 容器数量统计信息
|
||||
async fn get_pod_count(&self) -> Result<PodCountResponse, AppError>;
|
||||
|
||||
/// 确保容器存在（幂等操作）
|
||||
///
|
||||
/// 如果容器已存在则返回现有容器信息，否则创建新容器
|
||||
/// 仅启动容器，不启动 Agent 服务
|
||||
///
|
||||
/// # 参数
|
||||
/// - `request`: 启动容器请求
|
||||
///
|
||||
/// # 返回
|
||||
/// 容器信息和 VNC 访问信息
|
||||
async fn ensure_pod(&self, request: EnsurePodRequest) -> Result<EnsurePodResponse, AppError>;
|
||||
|
||||
/// 容器保活（刷新活动时间）
|
||||
///
|
||||
/// 如果容器已存在则刷新其最后活动时间，防止被清理任务销毁
|
||||
/// 如果容器不存在则自动创建
|
||||
///
|
||||
/// # 参数
|
||||
/// - `request`: 保活请求（包含 user_id 和 project_id）
|
||||
///
|
||||
/// # 返回
|
||||
/// 保活结果，包含活动时间信息和距离清理的剩余时间
|
||||
async fn keepalive_pod(&self, request: KeepalivePodRequest) -> Result<KeepalivePodResponse, AppError>;
|
||||
}
|
||||
```
|
||||
|
||||
### 5.2 请求/响应结构体
|
||||
|
||||
```rust
|
||||
// crates/rcoder/src/handler/pod_handler.rs
|
||||
|
||||
use serde::{Deserialize, Serialize};
|
||||
use utoipa::ToSchema;
|
||||
|
||||
// ============================================================================
|
||||
// 接口一：获取容器数量
|
||||
// ============================================================================
|
||||
|
||||
/// 容器数量响应
|
||||
#[derive(Debug, Clone, Serialize, Deserialize, ToSchema)]
|
||||
pub struct PodCountResponse {
|
||||
/// 当前运行的容器总数
|
||||
pub total_count: u32,
|
||||
|
||||
/// 按服务类型分类的容器数量
|
||||
pub by_service_type: PodCountByServiceType,
|
||||
|
||||
/// 统计时间戳 (Unix 毫秒)
|
||||
pub timestamp: u64,
|
||||
}
|
||||
|
||||
/// 按服务类型分类的容器数量
|
||||
#[derive(Debug, Clone, Serialize, Deserialize, ToSchema)]
|
||||
pub struct PodCountByServiceType {
|
||||
/// RCoder 类型容器数量
|
||||
pub rcoder: u32,
|
||||
|
||||
/// ComputerAgentRunner 类型容器数量
|
||||
pub computer_agent_runner: u32,
|
||||
}
|
||||
|
||||
// ============================================================================
|
||||
// 接口二：启动容器
|
||||
// ============================================================================
|
||||
|
||||
/// 启动容器请求
|
||||
#[derive(Debug, Clone, Deserialize, ToSchema)]
|
||||
pub struct EnsurePodRequest {
|
||||
/// 用户唯一标识符 (必填)
|
||||
#[schema(example = "user_123")]
|
||||
pub user_id: String,
|
||||
|
||||
/// 项目唯一标识符 (必填)
|
||||
#[schema(example = "proj_456")]
|
||||
pub project_id: String,
|
||||
|
||||
/// 可选的资源限制配置
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub resource_limits: Option<PodResourceLimits>,
|
||||
}
|
||||
|
||||
/// Pod 资源限制配置
|
||||
#[derive(Debug, Clone, Deserialize, Serialize, ToSchema)]
|
||||
pub struct PodResourceLimits {
|
||||
/// 内存限制 (bytes)
|
||||
pub memory: Option<u64>,
|
||||
|
||||
/// CPU 份额
|
||||
pub cpu_shares: Option<u64>,
|
||||
}
|
||||
|
||||
/// 启动容器响应
|
||||
///
|
||||
/// 注意: 此结构体作为 HttpResult<EnsurePodResponse> 的 data 字段返回
|
||||
/// 成功状态由外层 HttpResult 的 code 字段表示
|
||||
#[derive(Debug, Clone, Serialize, ToSchema)]
|
||||
pub struct EnsurePodResponse {
|
||||
/// 容器是否为新创建 (false 表示已存在)
|
||||
pub created: bool,
|
||||
|
||||
/// 容器基本信息
|
||||
pub container_info: PodContainerInfo,
|
||||
|
||||
/// 提示消息
|
||||
pub message: String,
|
||||
}
|
||||
|
||||
/// 容器基本信息
|
||||
#[derive(Debug, Clone, Serialize, ToSchema)]
|
||||
pub struct PodContainerInfo {
|
||||
/// 容器 ID
|
||||
pub container_id: String,
|
||||
|
||||
/// 容器名称
|
||||
pub container_name: String,
|
||||
|
||||
/// 容器 IP 地址
|
||||
pub container_ip: String,
|
||||
|
||||
/// 服务 URL
|
||||
pub service_url: String,
|
||||
|
||||
/// 容器状态
|
||||
pub status: String,
|
||||
}
|
||||
|
||||
// ============================================================================
|
||||
// 接口三：容器保活
|
||||
// ============================================================================
|
||||
|
||||
/// 容器保活请求
|
||||
#[derive(Debug, Clone, Deserialize, ToSchema)]
|
||||
pub struct KeepalivePodRequest {
|
||||
/// 用户唯一标识符
|
||||
#[schema(example = "user_123")]
|
||||
pub user_id: String,
|
||||
|
||||
/// 项目唯一标识符
|
||||
#[schema(example = "proj_456")]
|
||||
pub project_id: String,
|
||||
}
|
||||
|
||||
/// 容器保活响应
|
||||
///
|
||||
/// 注意: 此结构体作为 HttpResult<KeepalivePodResponse> 的 data 字段返回
|
||||
#[derive(Debug, Clone, Serialize, ToSchema)]
|
||||
pub struct KeepalivePodResponse {
|
||||
/// 容器是否已存在
|
||||
pub existed: bool,
|
||||
|
||||
/// 容器是否为新创建 (当 existed=false 时为 true)
|
||||
pub created: bool,
|
||||
|
||||
/// 容器基本信息
|
||||
pub container_info: PodContainerInfo,
|
||||
|
||||
/// 上次活动时间 (Unix 毫秒时间戳, 更新前)
|
||||
pub previous_activity_time: u64,
|
||||
|
||||
/// 当前活动时间 (Unix 毫秒时间戳, 更新后)
|
||||
pub current_activity_time: u64,
|
||||
|
||||
/// 距离下次清理的剩余时间 (秒)
|
||||
pub time_until_cleanup: u64,
|
||||
|
||||
/// 提示消息
|
||||
pub message: String,
|
||||
}
|
||||
```
|
||||
|
||||
### 5.3 Service 实现结构
|
||||
|
||||
```rust
|
||||
// crates/rcoder/src/service/pod_service.rs
|
||||
|
||||
use std::sync::Arc;
|
||||
|
||||
/// Pod 服务实现
|
||||
///
|
||||
/// 封装 Docker 容器管理的业务逻辑
|
||||
pub struct PodService {
|
||||
/// Docker 管理器
|
||||
docker_manager: Arc<docker_manager::DockerManager>,
|
||||
}
|
||||
|
||||
impl PodService {
|
||||
/// 创建 PodService 实例
|
||||
pub fn new(docker_manager: Arc<docker_manager::DockerManager>) -> Self {
|
||||
Self { docker_manager }
|
||||
}
|
||||
}
|
||||
|
||||
#[async_trait]
|
||||
impl PodServiceTrait for PodService {
|
||||
async fn get_pod_count(&self) -> Result<PodCountResponse, AppError> {
|
||||
// 具体实现...
|
||||
todo!()
|
||||
}
|
||||
|
||||
async fn ensure_pod(&self, request: EnsurePodRequest) -> Result<EnsurePodResponse, AppError> {
|
||||
// 具体实现...
|
||||
todo!()
|
||||
}
|
||||
|
||||
async fn keepalive_pod(&self, request: KeepalivePodRequest) -> Result<KeepalivePodResponse, AppError> {
|
||||
// 具体实现...
|
||||
// 1. 查询容器是否存在
|
||||
// 2. 不存在则创建容器
|
||||
// 3. 存在则刷新 last_activity 时间
|
||||
// 4. 计算距离清理的剩余时间
|
||||
todo!()
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 6. 路由配置
|
||||
|
||||
### 6.1 新增路由
|
||||
|
||||
在 `crates/rcoder/src/router.rs` 中添加以下路由：
|
||||
|
||||
```rust
|
||||
// 在 create_router 函数中添加
|
||||
|
||||
// Pod 容器管理路由 (统一使用 /computer 前缀)
|
||||
.route("/computer/pod/count", get(handler::pod_count))
|
||||
.route("/computer/pod/ensure", post(handler::pod_ensure))
|
||||
.route("/computer/pod/keepalive", post(handler::pod_keepalive))
|
||||
```
|
||||
|
||||
### 6.2 OpenAPI 配置更新
|
||||
|
||||
```rust
|
||||
// 在 ApiDoc 结构体中添加
|
||||
|
||||
#[openapi(
|
||||
paths(
|
||||
// ... 现有路径
|
||||
handler::pod_count,
|
||||
handler::pod_ensure,
|
||||
handler::pod_keepalive,
|
||||
),
|
||||
components(
|
||||
schemas(
|
||||
// ... 现有 schemas
|
||||
handler::PodCountResponse,
|
||||
handler::PodCountByServiceType,
|
||||
handler::EnsurePodRequest,
|
||||
handler::PodResourceLimits,
|
||||
handler::EnsurePodResponse,
|
||||
handler::PodContainerInfo,
|
||||
handler::KeepalivePodRequest,
|
||||
handler::KeepalivePodResponse,
|
||||
)
|
||||
),
|
||||
tags(
|
||||
// ... 现有 tags
|
||||
(name = "pod", description = "Pod 容器管理接口，支持容器监控和启动"),
|
||||
),
|
||||
)]
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 7. 错误处理
|
||||
|
||||
### 7.1 错误类型
|
||||
|
||||
| 错误码 | HTTP 状态码 | 描述 |
|
||||
|--------|-------------|------|
|
||||
| `DOCKER_MANAGER_ERROR` | 500 | Docker Manager 获取失败 |
|
||||
| `CONTAINER_CREATE_FAILED` | 500 | 容器创建失败 |
|
||||
| `CONTAINER_NOT_FOUND` | 404 | 容器未找到（仅查询场景） |
|
||||
| `INVALID_REQUEST` | 400 | 请求参数无效 |
|
||||
|
||||
### 7.2 错误响应结构（HttpResult 统一格式）
|
||||
|
||||
错误响应统一使用 `HttpResult<T>` 格式，`data` 字段为 `null`：
|
||||
|
||||
```json
|
||||
{
|
||||
"code": "5000",
|
||||
"message": "启动容器失败: Docker 连接超时",
|
||||
"data": null,
|
||||
"tid": "4bf92f3577b34da6a3ce929d0e0e4736",
|
||||
"success": false
|
||||
}
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 8. 与现有功能的关系
|
||||
|
||||
### 8.1 复用现有组件
|
||||
|
||||
| 组件 | 复用方式 |
|
||||
|------|----------|
|
||||
| `docker_manager::global::get_global_docker_manager()` | 获取全局 Docker 管理器实例 |
|
||||
| `ComputerContainerManager` | 复用用户容器创建逻辑 |
|
||||
| `ContainerBasicInfo` | 复用容器信息结构 |
|
||||
|
||||
### 8.2 与 noVNC 集成
|
||||
|
||||
VNC 访问信息通过独立接口获取：
|
||||
|
||||
- **VNC 桌面访问接口**: `GET /computer/desktop/{user_id}/{project_id}`
|
||||
|
||||
Pingora 已配置路由规则 `/computer/vnc/{user_id}/{project_id}/{*path}`，会自动代理到对应容器的 6080 端口。
|
||||
|
||||
---
|
||||
|
||||
## 9. 实现清单
|
||||
|
||||
### 9.1 新增文件
|
||||
|
||||
| 文件路径 | 描述 |
|
||||
|----------|------|
|
||||
| `crates/rcoder/src/handler/pod_handler.rs` | Pod 相关 HTTP 处理器 |
|
||||
| `crates/rcoder/src/service/pod_service.rs` | Pod 服务业务逻辑 |
|
||||
|
||||
### 9.2 修改文件
|
||||
|
||||
| 文件路径 | 修改内容 |
|
||||
|----------|----------|
|
||||
| `crates/rcoder/src/router.rs` | 添加 `/computer/pod/*` 路由 |
|
||||
| `crates/rcoder/src/handler/mod.rs` | 导出 `pod_handler` 模块 |
|
||||
| `crates/rcoder/src/service/mod.rs` | 导出 `pod_service` 模块 |
|
||||
|
||||
---
|
||||
|
||||
## 10. 验证计划
|
||||
|
||||
### 10.1 自动化测试
|
||||
|
||||
```bash
|
||||
# 1. 编译检查
|
||||
cargo build --workspace
|
||||
|
||||
# 2. 单元测试
|
||||
cargo test --workspace
|
||||
|
||||
# 3. Clippy 检查
|
||||
cargo clippy --workspace --all-targets
|
||||
```
|
||||
|
||||
### 10.2 手动验证
|
||||
|
||||
```bash
|
||||
# 1. 启动服务
|
||||
make dev-restart
|
||||
|
||||
# 2. 测试获取容器数量
|
||||
curl -X GET http://localhost:8087/computer/pod/count | jq
|
||||
|
||||
# 3. 测试启动容器
|
||||
curl -X POST http://localhost:8087/computer/pod/ensure \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{"user_id": "test_user", "project_id": "test_project"}' | jq
|
||||
|
||||
# 4. 验证 VNC 访问
|
||||
# 使用返回的 proxy_vnc_url 在浏览器中访问
|
||||
```
|
||||
|
||||
### 10.3 OpenAPI 文档验证
|
||||
|
||||
启动服务后访问 Swagger UI：`http://localhost:8087/swagger-ui/` 确认新接口已正确注册。
|
||||
|
||||
---
|
||||
|
||||
## 11. 附录
|
||||
|
||||
### 11.1 参考文件
|
||||
|
||||
- [computer_container_manager.rs](file:///Volumes/soddygo/git_work/rcoder/crates/rcoder/src/service/computer_container_manager.rs) - 用户容器管理逻辑
|
||||
- [computer_desktop_handler.rs](file:///Volumes/soddygo/git_work/rcoder/crates/rcoder/src/handler/computer_desktop_handler.rs) - VNC 桌面处理器
|
||||
- [docker_manager/manager.rs](file:///Volumes/soddygo/git_work/rcoder/crates/docker_manager/src/manager.rs) - Docker 容器管理核心
|
||||
|
||||
### 11.2 相关接口
|
||||
|
||||
| 接口 | 描述 |
|
||||
|------|------|
|
||||
| `GET /computer/desktop/{user_id}/{project_id}` | 获取 VNC 桌面访问信息 |
|
||||
| `POST /computer/chat` | Computer Agent 聊天接口 |
|
||||
| `/computer/vnc/{user_id}/{project_id}/*` | Pingora VNC 代理路由 |
|
||||
|
||||
---
|
||||
|
||||
## 12. 变更记录
|
||||
|
||||
| 日期 | 版本 | 变更内容 | 作者 |
|
||||
|------|------|----------|------|
|
||||
| 2024-12-16 | v1.0.0 | 初始版本 | AI Assistant |
|
||||
| 2024-12-16 | v1.0.1 | 使用 HttpResult 包装响应；明确不启动 Agent 服务 | AI Assistant |
|
||||
| 2024-12-16 | v1.1.0 | 新增容器保活接口 (keepalive)，支持刷新容器活动时间防止被自动清理 | AI Assistant |
|
||||
13
qiming-rcoder/specs/docker-pod-info/docker-pod-info.md
Normal file
13
qiming-rcoder/specs/docker-pod-info/docker-pod-info.md
Normal file
@@ -0,0 +1,13 @@
|
||||
# Instruction
|
||||
|
||||
## Project Alpha
|
||||
|
||||
@crates/rcoder/src/router.rs
|
||||
我需要新增接口:
|
||||
1. 获取当前容器数量,这个接口不需要入参
|
||||
2. 入参给 user_id, project_id , 启动容器(如果对应容器已经启动,则do nothing),不需要启动agent服务,方便我使用 noVNC 来远程虚拟桌面使用.
|
||||
3. 增加一个容器保活的keepalive接口,入参给 user_id, project_id , 如果对应容器存在,则do nothing,如果不存在,则启动容器,另外自动刷新容器的最后更新使用时间,防止容器被定时清理任务,给清理销毁掉
|
||||
|
||||
这2个新接口,返回结构要用 HttpResult 来包装, @crates/shared_types/src/model/http_result.rs
|
||||
|
||||
按照这个想法，帮我生成详细的需求和设计文档，输出为中文。涉及到代码部分，不要写详细的具体实现，只需要写 trait 和 结构体 struct
|
||||
Reference in New Issue
Block a user