功能: 修复并实现多节点集群部署 (#38)

基础修复:
- 新增节点离线检测:每 15s 扫描,超 45s 未心跳的远程节点自动置离线
- 节点删除前检查关联任务,避免孤立备份任务
- BackupTaskRepository 新增 CountByNodeID/ListByNodeID

Master 端 Agent 协议:
- 新增 AgentCommand 模型与命令队列仓储(pending/dispatched/succeeded/failed/timeout)
- 新增 AgentService:任务下发、命令轮询、结果回收、超时扫描
- 新增专用 Agent HTTP API(X-Agent-Token 认证):
  /api/agent/heartbeat
  /api/agent/commands/poll
  /api/agent/commands/:id/result
  /api/agent/tasks/:id
  /api/agent/records/:id
- BackupExecutionService 支持 node 路由:task.NodeID 指向远程节点时自动入队派发

Agent CLI(backupx agent 子命令):
- 配置:YAML 文件 / 环境变量 / CLI 参数,优先级 CLI > 文件 > 环境
- 心跳循环 + 命令轮询循环 + 优雅退出
- 本地复用 BackupRunner 与 storage registry 执行备份并直接上传
- 支持 run_task 和 list_dir 两种命令

远程目录浏览:
- NodeService 支持通过 Agent RPC 列出远程节点目录(15s 超时)

前端:
- NodesPage 添加节点后展示 Agent 启动命令和环境变量配置

文档:
- README 中英文重写"多节点集群"章节,含架构图、步骤、限制、CLI 参考
This commit is contained in:
Wu Qing
2026-04-17 12:29:08 +08:00
committed by GitHub
parent e04774ff68
commit 757b0fa5ed
27 changed files with 2224 additions and 24 deletions

View File

@@ -0,0 +1,156 @@
package http
import (
stdhttp "net/http"
"strconv"
"strings"
"backupx/server/internal/service"
"backupx/server/pkg/response"
"github.com/gin-gonic/gin"
)
// AgentHandler 实现 Agent 调用 Master 的 HTTP API。
// 全部端点通过 X-Agent-Token 头做节点认证,不使用 JWT。
type AgentHandler struct {
agentService *service.AgentService
nodeService *service.NodeService
}
func NewAgentHandler(agentService *service.AgentService, nodeService *service.NodeService) *AgentHandler {
return &AgentHandler{agentService: agentService, nodeService: nodeService}
}
// extractToken 从请求头或 JSON body 中提取 Agent Token。
func extractToken(c *gin.Context) string {
if t := strings.TrimSpace(c.GetHeader("X-Agent-Token")); t != "" {
return t
}
// Authorization: Bearer <token>
if auth := c.GetHeader("Authorization"); strings.HasPrefix(auth, "Bearer ") {
return strings.TrimSpace(strings.TrimPrefix(auth, "Bearer "))
}
return ""
}
// Heartbeat 扩展原有 heartbeat除上报状态外返回节点 ID 给 Agent 做后续调用。
func (h *AgentHandler) Heartbeat(c *gin.Context) {
var input struct {
Token string `json:"token"`
Hostname string `json:"hostname"`
IPAddress string `json:"ipAddress"`
AgentVersion string `json:"agentVersion"`
OS string `json:"os"`
Arch string `json:"arch"`
}
_ = c.ShouldBindJSON(&input)
// token 优先走 body向后兼容否则从 header 读
token := input.Token
if token == "" {
token = extractToken(c)
}
if token == "" {
c.JSON(stdhttp.StatusBadRequest, gin.H{"code": "INVALID_INPUT", "message": "missing token"})
return
}
if err := h.nodeService.Heartbeat(c.Request.Context(), token, input.Hostname, input.IPAddress, input.AgentVersion, input.OS, input.Arch); err != nil {
response.Error(c, err)
return
}
// 返回节点元信息给 Agentnode_id 用于后续 API 路径)
node, err := h.agentService.AuthenticatedNode(c.Request.Context(), token)
if err != nil {
response.Error(c, err)
return
}
response.Success(c, gin.H{
"status": "ok",
"nodeId": node.ID,
"name": node.Name,
})
}
// Poll Agent 长轮询获取下一条待执行命令。
// 无命令时返回 {command: null}。
func (h *AgentHandler) Poll(c *gin.Context) {
node, err := h.agentService.AuthenticatedNode(c.Request.Context(), extractToken(c))
if err != nil {
response.Error(c, err)
return
}
cmd, err := h.agentService.PollCommand(c.Request.Context(), node)
if err != nil {
response.Error(c, err)
return
}
response.Success(c, gin.H{"command": cmd})
}
// SubmitCommandResult Agent 上报命令执行结果。
func (h *AgentHandler) SubmitCommandResult(c *gin.Context) {
node, err := h.agentService.AuthenticatedNode(c.Request.Context(), extractToken(c))
if err != nil {
response.Error(c, err)
return
}
id, err := strconv.ParseUint(c.Param("id"), 10, 32)
if err != nil {
response.Error(c, err)
return
}
var input service.AgentCommandResult
if err := c.ShouldBindJSON(&input); err != nil {
c.JSON(stdhttp.StatusBadRequest, gin.H{"code": "INVALID_INPUT", "message": err.Error()})
return
}
if err := h.agentService.SubmitCommandResult(c.Request.Context(), node, uint(id), input); err != nil {
response.Error(c, err)
return
}
response.Success(c, gin.H{"status": "ok"})
}
// GetTaskSpec Agent 拉取任务规格(含解密后的存储配置)。
func (h *AgentHandler) GetTaskSpec(c *gin.Context) {
node, err := h.agentService.AuthenticatedNode(c.Request.Context(), extractToken(c))
if err != nil {
response.Error(c, err)
return
}
id, err := strconv.ParseUint(c.Param("id"), 10, 32)
if err != nil {
response.Error(c, err)
return
}
spec, err := h.agentService.GetTaskSpec(c.Request.Context(), node, uint(id))
if err != nil {
response.Error(c, err)
return
}
response.Success(c, spec)
}
// UpdateRecord Agent 更新备份记录(进度/完成状态/日志)。
func (h *AgentHandler) UpdateRecord(c *gin.Context) {
node, err := h.agentService.AuthenticatedNode(c.Request.Context(), extractToken(c))
if err != nil {
response.Error(c, err)
return
}
id, err := strconv.ParseUint(c.Param("id"), 10, 32)
if err != nil {
response.Error(c, err)
return
}
var input service.AgentRecordUpdate
if err := c.ShouldBindJSON(&input); err != nil {
c.JSON(stdhttp.StatusBadRequest, gin.H{"code": "INVALID_INPUT", "message": err.Error()})
return
}
if err := h.agentService.UpdateRecord(c.Request.Context(), node, uint(id), input); err != nil {
response.Error(c, err)
return
}
response.Success(c, gin.H{"status": "ok"})
}

View File

@@ -28,6 +28,7 @@ type RouterDependencies struct {
DashboardService *service.DashboardService
SettingsService *service.SettingsService
NodeService *service.NodeService
AgentService *service.AgentService
DatabaseDiscoveryService *service.DatabaseDiscoveryService
AuditService *service.AuditService
JWTManager *security.JWTManager
@@ -150,8 +151,19 @@ func NewRouter(deps RouterDependencies) *gin.Engine {
nodes.DELETE("/:id", nodeHandler.Delete)
nodes.GET("/:id/fs/list", nodeHandler.ListDirectory)
// Agent heartbeat (public, token-authenticated)
api.POST("/agent/heartbeat", nodeHandler.Heartbeat)
// Agent APItoken 认证,无需 JWT
if deps.AgentService != nil {
agentHandler := NewAgentHandler(deps.AgentService, deps.NodeService)
agent := api.Group("/agent")
agent.POST("/heartbeat", agentHandler.Heartbeat)
agent.POST("/commands/poll", agentHandler.Poll)
agent.POST("/commands/:id/result", agentHandler.SubmitCommandResult)
agent.GET("/tasks/:id", agentHandler.GetTaskSpec)
agent.POST("/records/:id", agentHandler.UpdateRecord)
} else {
// 未启用 Agent 服务时,保留原有 heartbeat 端点以兼容
api.POST("/agent/heartbeat", nodeHandler.Heartbeat)
}
}
engine.NoRoute(func(c *gin.Context) {