opsx
37e2f0febc
=== 1. 每日日志量必须落库（核心问题：Loki 只留 24 小时）===
实测：请求"近 7 天每日量"，Loki 只返回 27 个**小时**桶，全部在最近 24h 内 ——
因为 retention_period=24h，更早的 chunk 已经被删了。
也就是说「固定近 7 天」这张卡片，过去一直是"6 天空白 + 1 天有数据"，
用户看着像坏了，实际是数据物理不存在。
方案：平台自己留存历史。
- models.DailyVolumeSnapshot{Date(本地YYYY-MM-DD), SourceID, Bytes, Entries, Partial}
- services/daily_volume.go
CaptureDailyVolume 按天抓 index/stats 落库（幂等覆盖）
StartDailyVolumeCapture 启动 20s 后回补 + 每 30 分钟检查一次
GetDailyVolumeHistory 读库返回最近 N 天（今天会实时补抓）
- 保留 400 天（够看年度趋势），按 source+date 去重
- 前端可选 7 / 30 / 90 / 365 天，能看多长取决于平台积累多久
为什么用"每 30 分钟检查昨天有没有落库"而不是固定 00:10 定时：
平台可能整晚没运行，固定时点的定时器会直接错过；检查式回补不依赖平台在线时刻。
Partial 标记（关键，否则会误读）：
- 当天未结束 → partial=true
- 待抓取日期的起点已超出 24h 保留期 → partial=true（Loki 里就没那么多数据）
实测：10-27 32.5GB/2.65亿条 partial=true（保留期外），
10-28 348GB/28.56亿条 partial=true（当天进行中）
=== 2. 修一个 Go 副本 bug（自己写的，靠实测发现）===
第一版用 `for _, snap := range db.DailyVolumeSnapshots` 后改 snap 字段，
range 拿到的是**副本**，改动不会写回切片 → 同一天永远保持首次抓取的值，
partial 也永不更新（实测发现 10-27 该是 true 却显示 false）。
改为 `for i := range ... { snap := &db.DailyVolumeSnapshots[i] }`。
=== 3. 趋势图可读性 ===
从"裸折线"换成"渐变面积图"，并在图上方给出峰值 / 均值：
- 面积图能一眼看出"这一段整体比上一段高"，折线只能靠对齐格子数
- 峰值/均值避免用户自己对刻度
- Y 轴与 tooltip 统一 formatBytes，"每桶体积"写进 subtitle
顺带确认：趋势图只画体积（字节），不再混入 error（条数）——单位不同不能同轴。
=== 4. 饼图：命名空间分布 → 应用日志量分布 ===
本环境的命名空间只有 astralx / log 两个，饼图没有信息量；
改成按应用看谁在刷日志（前 8 名 + 其他），才是运维真正要的答案。
=== 5. 告警页拆成三个 Tab ===
「告警规则 / 通知通道 / 发送历史」独立成分页（用 ui/tabs）。
之前通道是折叠块、历史堆在规则下面，规则一多就要滚很久才看得到。
发送历史增加空状态文案，不再是"空页面不知道有没有在发"。
=== 6. 内置规则关联默认通道 + 首次部署 webhook 留空的评估 ===
结论：**webhook 可以留空，但通道必须是禁用状态**。
- 若种子给的是"已启用 + 空地址"通道，规则一命中就必然投递失败，
发送历史堆满红色失败，反而让人以为平台坏了
- 所以首次部署建一个**禁用**的占位通道「默认通知通道（待配置 Webhook）」，
内置规则全部挂到它；管理员填好 Webhook 再启用即可，不用逐条关联
- 保存时加校验：启用则必须有非空 Webhook（返回明确原因，不是裸 500）
- schemaVersion v5→v6：把已有的"无通道"规则补挂到默认通道
实测：8 条规则全部已关联通道
=== 实测（v11-1028-2252 / 187062644c0f）===
/logs/daily-volume?days=7 → 2 天数据，partial 标记正确
规则 8 条全部有 channelIds
平台 ready、健康 ok
282 lines
11 KiB
Go
282 lines
11 KiB
Go
package routers
|
||
|
||
import (
|
||
"net/http"
|
||
"os"
|
||
"path/filepath"
|
||
"strings"
|
||
"time"
|
||
|
||
"lokistack/controllers"
|
||
"lokistack/middleware"
|
||
"lokistack/pkg/metrics"
|
||
|
||
"github.com/gin-contrib/cors"
|
||
"github.com/gin-gonic/gin"
|
||
)
|
||
|
||
func SetupRouter() *gin.Engine {
|
||
// 用 gin.New() 而不是 gin.Default()：Default 自带文本格式的 gin.Logger()，
|
||
// 这里换成结构化 JSON 访问日志（middleware.AccessLog），便于 Promtail→Loki
|
||
// 按字段过滤，同时跳过健康检查/静态资源，避免探针轮询把日志量刷满。
|
||
// Recovery 必须显式加回来（Default 里包含它），否则 panic 会打挂进程。
|
||
r := gin.New()
|
||
// AuditLog 放在最外层：它依赖的 userId/username 由 group 内的 JWTAuth 写入，
|
||
// 但中间件的后置逻辑在 c.Next() 返回后才执行，那时已就绪（见 middleware/audit.go）。
|
||
r.Use(middleware.AccessLog(), middleware.AuditLog(), middleware.Metrics(), gin.Recovery())
|
||
|
||
// CORS
|
||
cfg := cors.DefaultConfig()
|
||
cfg.AllowAllOrigins = true
|
||
cfg.AllowHeaders = []string{"Origin", "Content-Type", "Accept", "Authorization", "X-Requested-With"}
|
||
cfg.AllowMethods = []string{"GET", "POST", "PUT", "DELETE", "OPTIONS", "PATCH"}
|
||
r.Use(cors.New(cfg))
|
||
|
||
// Controllers
|
||
authCtrl := controllers.NewAuthController()
|
||
clusterCtrl := controllers.NewClusterController()
|
||
hostCtrl := controllers.NewHostController()
|
||
storageCtrl := controllers.NewStorageController()
|
||
deploymentCtrl := controllers.NewDeploymentController()
|
||
collectorCtrl := controllers.NewCollectorController()
|
||
userCtrl := controllers.NewUserController()
|
||
ldapCtrl := controllers.NewLdapController()
|
||
logsCtrl := controllers.NewLogsController()
|
||
alertsCtrl := controllers.NewAlertsController()
|
||
auditCtrl := controllers.NewAuditController()
|
||
systemCtrl := controllers.NewSystemController()
|
||
registryCtrl := controllers.NewImageRegistryController()
|
||
|
||
// API v1
|
||
api := r.Group("/api/v1")
|
||
|
||
// 健康检查（公开访问，用于负载均衡和监控）
|
||
api.GET("/health", func(c *gin.Context) {
|
||
c.JSON(http.StatusOK, gin.H{
|
||
"status": "ok",
|
||
"timestamp": time.Now().Unix(),
|
||
"version": "2.0.0",
|
||
})
|
||
})
|
||
|
||
// Prometheus metrics 端点（真实指标，见 services/metrics.go）
|
||
// 原实现只返回一行注释却声称"可被 Prometheus 抓取"，平台自身等于不可观测。
|
||
api.GET("/metrics", gin.WrapH(metrics.Handler()))
|
||
|
||
// Public（登录接口挂 IP 限流防爆破：20 次/分钟）
|
||
api.POST("/auth/login", middleware.LoginRateLimit(20, time.Minute), authCtrl.Login)
|
||
api.POST("/auth/refresh", authCtrl.RefreshToken)
|
||
|
||
// Auth required
|
||
authorized := api.Group("")
|
||
authorized.Use(middleware.JWTAuth())
|
||
|
||
// Auth routes
|
||
authorized.GET("/auth/user", authCtrl.GetUser)
|
||
authorized.POST("/auth/password", authCtrl.ChangePassword)
|
||
|
||
// 日志查询:数据源管理 + Loki 代理查询
|
||
logs := authorized.Group("/logs")
|
||
{
|
||
logs.GET("/sources", logsCtrl.ListSources)
|
||
logs.POST("/sources", middleware.AdminAuth(), logsCtrl.CreateSource)
|
||
logs.PUT("/sources/:id", middleware.AdminAuth(), logsCtrl.UpdateSource)
|
||
logs.DELETE("/sources/:id", middleware.AdminAuth(), logsCtrl.DeleteSource)
|
||
logs.POST("/sources/test", middleware.AdminAuth(), logsCtrl.TestSource)
|
||
logs.GET("/policy-groups", middleware.AdminAuth(), logsCtrl.ListPolicyGroups)
|
||
logs.POST("/policy-groups", middleware.AdminAuth(), logsCtrl.CreatePolicyGroup)
|
||
logs.PUT("/policy-groups/:id", middleware.AdminAuth(), logsCtrl.UpdatePolicyGroup)
|
||
logs.DELETE("/policy-groups/:id", middleware.AdminAuth(), logsCtrl.DeletePolicyGroup)
|
||
logs.POST("/query", logsCtrl.Query)
|
||
logs.POST("/context", logsCtrl.Context)
|
||
logs.GET("/labels", logsCtrl.Labels)
|
||
logs.GET("/label/:name/values", logsCtrl.LabelValues)
|
||
logs.GET("/my-scope", logsCtrl.MyScope)
|
||
logs.GET("/histogram", logsCtrl.Histogram)
|
||
logs.GET("/daily-volume", logsCtrl.DailyVolume)
|
||
// 总量统计（index/stats）：只读索引，用于算总页数
|
||
logs.POST("/stats", logsCtrl.Stats)
|
||
logs.POST("/metrics", middleware.AdminAuth(), logsCtrl.Metrics)
|
||
logs.GET("/health", logsCtrl.Health)
|
||
}
|
||
|
||
// 镜像源(国内加速/海外/私有仓库)
|
||
authorized.GET("/settings/image-registry", registryCtrl.Get)
|
||
authorized.PUT("/settings/image-registry", middleware.AdminAuth(), registryCtrl.Save)
|
||
|
||
// 告警:飞书通知配置 + 告警规则
|
||
alertNotify := authorized.Group("/alerts")
|
||
{
|
||
alertNotify.GET("/notify-config", alertsCtrl.GetNotifyConfig)
|
||
alertNotify.PUT("/notify-config", middleware.AdminAuth(), alertsCtrl.UpdateNotifyConfig)
|
||
alertNotify.POST("/notify-config/test", middleware.AdminAuth(), alertsCtrl.TestNotifyConfig)
|
||
alertNotify.GET("/rules", alertsCtrl.ListRules)
|
||
alertNotify.POST("/rules", middleware.AdminAuth(), alertsCtrl.CreateRule)
|
||
alertNotify.PUT("/rules/:id", middleware.AdminAuth(), alertsCtrl.UpdateRule)
|
||
alertNotify.DELETE("/rules/:id", middleware.AdminAuth(), alertsCtrl.DeleteRule)
|
||
alertNotify.POST("/check", middleware.AdminAuth(), alertsCtrl.CheckNow)
|
||
alertNotify.GET("/deliveries", alertsCtrl.Deliveries)
|
||
alertNotify.GET("/channels", alertsCtrl.ListChannels)
|
||
alertNotify.POST("/channels", middleware.AdminAuth(), alertsCtrl.SaveChannel)
|
||
alertNotify.PUT("/channels/:id", middleware.AdminAuth(), alertsCtrl.SaveChannel)
|
||
alertNotify.DELETE("/channels/:id", middleware.AdminAuth(), alertsCtrl.DeleteChannel)
|
||
alertNotify.POST("/channels/test", middleware.AdminAuth(), alertsCtrl.TestChannel)
|
||
}
|
||
|
||
// LDAP 配置(admin 读写;查询供用户页判断导入按钮显隐,登录用户即可读)
|
||
authorized.GET("/settings/ldap", middleware.AdminAuth(), ldapCtrl.GetConfig)
|
||
authorized.PUT("/settings/ldap", middleware.AdminAuth(), ldapCtrl.UpdateConfig)
|
||
authorized.POST("/settings/ldap/test", middleware.AdminAuth(), ldapCtrl.TestConnection)
|
||
authorized.POST("/settings/ldap/users", middleware.AdminAuth(), ldapCtrl.ListUsers)
|
||
|
||
// Users（admin 专用）
|
||
// 系统运维：备份 / 恢复 / 一键自检（仅管理员）
|
||
system := authorized.Group("/system")
|
||
system.Use(middleware.AdminAuth())
|
||
{
|
||
system.GET("/backup", systemCtrl.Backup)
|
||
system.POST("/restore", systemCtrl.Restore)
|
||
system.POST("/restore/apply", systemCtrl.RestoreApply)
|
||
system.GET("/backups", systemCtrl.Backups)
|
||
system.GET("/self-check", systemCtrl.Health)
|
||
}
|
||
|
||
// 审计日志：仅管理员可查（含他人操作记录）
|
||
audit := authorized.Group("/audit-logs")
|
||
audit.Use(middleware.AdminAuth())
|
||
{
|
||
audit.GET("", auditCtrl.List)
|
||
audit.GET("/stats", auditCtrl.Stats)
|
||
audit.GET("/:id", auditCtrl.Get)
|
||
}
|
||
|
||
users := authorized.Group("/users")
|
||
users.Use(middleware.AdminAuth())
|
||
{
|
||
users.GET("", userCtrl.List)
|
||
users.POST("", userCtrl.Create)
|
||
users.DELETE("/:id", userCtrl.Delete)
|
||
users.POST("/batch-delete", userCtrl.BatchDelete)
|
||
users.POST("/:id/reset-password", userCtrl.ResetPassword)
|
||
users.POST("/batch-import-ldap", userCtrl.BatchImportLDAP)
|
||
}
|
||
// 普通用户仅可访问日志查询及自己的账号接口；管理资源统一在服务端保护。
|
||
authorized.Use(middleware.AdminAuth())
|
||
|
||
// Clusters
|
||
authorized.GET("/clusters", clusterCtrl.List)
|
||
authorized.GET("/clusters/:id", clusterCtrl.Get)
|
||
authorized.POST("/clusters", clusterCtrl.Create)
|
||
authorized.PUT("/clusters/:id", clusterCtrl.Update)
|
||
authorized.DELETE("/clusters/:id", clusterCtrl.Delete)
|
||
authorized.POST("/clusters/:id/validate", clusterCtrl.Validate)
|
||
authorized.GET("/clusters/:id/info", clusterCtrl.GetInfo)
|
||
authorized.GET("/clusters/:id/namespaces", clusterCtrl.Namespaces)
|
||
authorized.GET("/clusters/:id/namespaces/:namespace/deployments", clusterCtrl.Deployments)
|
||
|
||
// Hosts
|
||
authorized.GET("/hosts", hostCtrl.List)
|
||
authorized.GET("/hosts/:id", hostCtrl.Get)
|
||
authorized.POST("/hosts", hostCtrl.Create)
|
||
authorized.PUT("/hosts/:id", hostCtrl.Update)
|
||
authorized.DELETE("/hosts/:id", hostCtrl.Delete)
|
||
authorized.POST("/hosts/:id/validate", hostCtrl.Validate)
|
||
authorized.GET("/hosts/:id/info", hostCtrl.GetInfo)
|
||
|
||
// Storage
|
||
authorized.GET("/storages", storageCtrl.List)
|
||
authorized.GET("/storages/:id", storageCtrl.Get)
|
||
authorized.POST("/storages", storageCtrl.Create)
|
||
authorized.POST("/storages/test", storageCtrl.TestConnection)
|
||
authorized.PUT("/storages/:id", storageCtrl.Update)
|
||
authorized.DELETE("/storages/:id", storageCtrl.Delete)
|
||
authorized.GET("/storages/default", storageCtrl.GetDefault)
|
||
|
||
// Deployments
|
||
authorized.GET("/deployments", deploymentCtrl.List)
|
||
authorized.GET("/deployments/:id", deploymentCtrl.Get)
|
||
authorized.POST("/deployments", deploymentCtrl.Create)
|
||
authorized.PUT("/deployments/:id", deploymentCtrl.Update)
|
||
authorized.POST("/deployments/:id/retention", deploymentCtrl.UpdateRetention)
|
||
authorized.GET("/deployments/:id/config", middleware.AdminAuth(), deploymentCtrl.GetLiveConfig)
|
||
authorized.DELETE("/deployments/:id", deploymentCtrl.Delete)
|
||
authorized.POST("/deployments/:id/deploy", deploymentCtrl.Deploy)
|
||
authorized.POST("/deployments/:id/stop", deploymentCtrl.Stop)
|
||
authorized.POST("/deployments/:id/restart", deploymentCtrl.Restart)
|
||
authorized.GET("/deployments/:id/logs", deploymentCtrl.GetLogs)
|
||
// 资源用量（CPU/内存/限流），页面用来判断是否要扩容
|
||
authorized.GET("/deployments/:id/usage", deploymentCtrl.GetUsage)
|
||
authorized.GET("/deployments/stats", deploymentCtrl.GetStats)
|
||
authorized.GET("/deployments/preflight", deploymentCtrl.Preflight)
|
||
authorized.GET("/deployments/:id/ready", deploymentCtrl.Ready)
|
||
authorized.GET("/deployments/versions", deploymentCtrl.GetVersions)
|
||
|
||
// Collectors
|
||
authorized.GET("/collectors", collectorCtrl.List)
|
||
authorized.GET("/collectors/:id", collectorCtrl.Get)
|
||
authorized.POST("/collectors", collectorCtrl.Create)
|
||
authorized.PUT("/collectors/:id", collectorCtrl.Update)
|
||
authorized.DELETE("/collectors/:id", collectorCtrl.Delete)
|
||
authorized.POST("/collectors/:id/install", collectorCtrl.Install)
|
||
authorized.POST("/collectors/:id/uninstall", collectorCtrl.Uninstall)
|
||
authorized.POST("/collectors/:id/preview", collectorCtrl.Preview)
|
||
authorized.GET("/collectors/:id/logs", collectorCtrl.Logs)
|
||
|
||
// Static files - use executable directory as base
|
||
distPath := findDistPath()
|
||
|
||
if distPath != "" {
|
||
r.Static("/assets", filepath.Join(distPath, "assets"))
|
||
|
||
// favicon：NoRoute 会把所有非 /api/ 请求回退到 index.html，
|
||
// 故此处显式注册 /favicon.svg 与 /favicon.ico，否则浏览器请求图标只会拿到 HTML。
|
||
faviconSVG := filepath.Join(distPath, "favicon.svg")
|
||
serveFavicon := func(c *gin.Context) {
|
||
if _, err := os.Stat(faviconSVG); err != nil {
|
||
c.Status(404)
|
||
return
|
||
}
|
||
c.Header("Content-Type", "image/svg+xml")
|
||
c.File(faviconSVG)
|
||
}
|
||
r.GET("/favicon.svg", serveFavicon)
|
||
r.GET("/favicon.ico", serveFavicon)
|
||
|
||
r.NoRoute(func(c *gin.Context) {
|
||
if strings.HasPrefix(c.Request.URL.Path, "/api/") {
|
||
c.JSON(404, gin.H{"code": 404, "message": "API endpoint not found"})
|
||
return
|
||
}
|
||
c.File(filepath.Join(distPath, "index.html"))
|
||
})
|
||
}
|
||
|
||
return r
|
||
}
|
||
|
||
// findDistPath finds the frontend dist directory
|
||
func findDistPath() string {
|
||
// Try relative to executable
|
||
if ex, err := os.Executable(); err == nil {
|
||
execDir := filepath.Dir(ex)
|
||
// If running via "go run", executable is in temp dir - fall through
|
||
if !strings.Contains(execDir, "/tmp/go-build") {
|
||
for _, rel := range []string{"dist", "../dist", "../../dist"} {
|
||
p := filepath.Join(execDir, rel)
|
||
if _, err := os.Stat(filepath.Join(p, "index.html")); err == nil {
|
||
return p
|
||
}
|
||
}
|
||
}
|
||
}
|
||
// Try working directory relative paths
|
||
for _, p := range []string{"./dist", "../dist", "/app/dist"} {
|
||
if abs, err := filepath.Abs(p); err == nil {
|
||
if _, err := os.Stat(filepath.Join(abs, "index.html")); err == nil {
|
||
return abs
|
||
}
|
||
}
|
||
}
|
||
return ""
|
||
}
|