feat: 监控告警体系统一重构 + 自动化任务调度引擎 + 监控大屏拆分
告警中心 - 新增统一告警(alert_unified)、监控告警规则(monitor_alert_rule)、指标目录(alert_metric_catalog) - 通知策略与消息分发(alert_notify_policy/dispatch/message)、静默规则 - 告警引擎按阈值/SNMP 拆分(alert_engine_monitor/snmp) 监控采集 - SNMP 常量/模板与历史指标(snmp templates/constants、snmp_metrics_history、snmp_collect) - 中间件监控 Redis 采集与种子(middleware_monitor_redis/seed) - 站点探测 pkg(siteprobe) + 站点视图(site_view) 自动化任务 - cron 调度器(cronscheduler)、脚本运行器(scriptrunner)、系统任务(systask)及运行记录(cron_job_run、system_task) 审计与凭证 - 审计设置(audit_setting) + 历史清理(auditcleaner) 前端 - 监控页面(host/middleware/site/snmp)拆分为子模块并大屏化 - 告警中心多 Tab(规则/指标目录/通知策略/静默)、任务管理组件重构、Agent CLI Markdown 渲染 - 移除旧脚本卡片/详情/日志组件，统一为 ScriptTable/ScriptCodeEditor/TerminalOutput Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
LoganandClaude committed
1 parent
092d0e4b9e
commit
471353e3d7
190 files changed
+16640
-3484
No files matched your search
@@ -0,0 +1,226 @@
|
||||
// 内置系统定时任务注册表 — 告警引擎 / 数据清理等 // Logan.Li
|
||||
package systask
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"sync"
|
||||
"time"
|
||||
|
||||
"ezops/model"
|
||||
"ezops/pkg/auditcleaner"
|
||||
"ezops/pkg/database"
|
||||
snmppkg "ezops/pkg/snmp"
|
||||
|
||||
"gorm.io/gorm"
|
||||
)
|
||||
|
||||
// Def 系统任务元数据 // Logan.Li
|
||||
type Def struct {
|
||||
Code string `json:"code"`
|
||||
Name string `json:"name"`
|
||||
Description string `json:"description"`
|
||||
Schedule string `json:"schedule"`
|
||||
Category string `json:"category"` // monitor / cleanup
|
||||
Runnable bool `json:"runnable"`
|
||||
Enabled bool `json:"enabled"`
|
||||
}
|
||||
|
||||
// Item API 列表项（元数据 + 运行状态） // Logan.Li
|
||||
type Item struct {
|
||||
Def
|
||||
LastRunAt *time.Time `json:"last_run_at,omitempty"`
|
||||
LastStatus string `json:"last_status,omitempty"`
|
||||
LastMessage string `json:"last_message,omitempty"`
|
||||
RunCount int64 `json:"run_count"`
|
||||
LastDurationMs int `json:"last_duration_ms"`
|
||||
Extra any `json:"extra,omitempty"`
|
||||
}
|
||||
|
||||
var (
|
||||
mu sync.RWMutex
|
||||
db *gorm.DB
|
||||
|
||||
catalog = []Def{
|
||||
{
|
||||
Code: "alert_engine", Name: "告警引擎扫描",
|
||||
Description: "扫描 Agent 指标阈值、主机离线、端口策略与日志告警规则，触发通知与自动恢复",
|
||||
Schedule: "每 30 秒", Category: "monitor", Runnable: false, Enabled: true,
|
||||
},
|
||||
{
|
||||
Code: "middleware_monitor", Name: "中间件监控采集",
|
||||
Description: "直连 Redis/MySQL 等中间件目标采集指标并写入监控库",
|
||||
Schedule: "每 30 秒", Category: "monitor", Runnable: false, Enabled: true,
|
||||
},
|
||||
{
|
||||
Code: "agent_report_cleanup", Name: "Agent 上报清理",
|
||||
Description: "清理 7 天前的 Agent 指标上报原始数据，防止表膨胀",
|
||||
Schedule: "每 24 小时", Category: "cleanup", Runnable: true, Enabled: true,
|
||||
},
|
||||
{
|
||||
Code: "audit_log_cleanup", Name: "审计日志清理",
|
||||
Description: "按系统设置中的保留天数清理操作审计、登录日志、终端会话与命令记录",
|
||||
Schedule: "每 24 小时", Category: "cleanup", Runnable: true, Enabled: true,
|
||||
},
|
||||
{
|
||||
Code: "snmp_metrics_cleanup", Name: "SNMP 指标清理",
|
||||
Description: "清理 7 天前的 SNMP 采集指标历史，防止表膨胀",
|
||||
Schedule: "每 24 小时", Category: "cleanup", Runnable: true, Enabled: true,
|
||||
},
|
||||
}
|
||||
)
|
||||
|
||||
// Init 注入数据库连接，并确保状态行存在 // Logan.Li
|
||||
func Init(gdb *gorm.DB) {
|
||||
mu.Lock()
|
||||
db = gdb
|
||||
mu.Unlock()
|
||||
ensureStatusRows(gdb)
|
||||
}
|
||||
|
||||
func ensureStatusRows(gdb *gorm.DB) {
|
||||
for _, d := range catalog {
|
||||
var row model.SystemTaskStatus
|
||||
if gdb.Where("code = ?", d.Code).First(&row).Error == gorm.ErrRecordNotFound {
|
||||
gdb.Create(&model.SystemTaskStatus{Code: d.Code, LastStatus: "pending"})
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Catalog 返回任务定义列表 // Logan.Li
|
||||
func Catalog() []Def {
|
||||
out := make([]Def, len(catalog))
|
||||
copy(out, catalog)
|
||||
return out
|
||||
}
|
||||
|
||||
// Record 记录任务执行结果 // Logan.Li
|
||||
func Record(code, status, message string, duration time.Duration) {
|
||||
mu.RLock()
|
||||
gdb := db
|
||||
mu.RUnlock()
|
||||
if gdb == nil {
|
||||
return
|
||||
}
|
||||
now := time.Now()
|
||||
updates := map[string]interface{}{
|
||||
"last_run_at": now,
|
||||
"last_status": status,
|
||||
"last_message": truncate(message, 480),
|
||||
"last_duration_ms": int(duration.Milliseconds()),
|
||||
"run_count": gorm.Expr("run_count + 1"),
|
||||
}
|
||||
gdb.Model(&model.SystemTaskStatus{}).Where("code = ?", code).Updates(updates)
|
||||
}
|
||||
|
||||
func truncate(s string, max int) string {
|
||||
if len(s) <= max {
|
||||
return s
|
||||
}
|
||||
return s[:max]
|
||||
}
|
||||
|
||||
// List 合并元数据与库中状态 // Logan.Li
|
||||
func List(gdb *gorm.DB, extras map[string]any) []Item {
|
||||
ensureStatusRows(gdb)
|
||||
var rows []model.SystemTaskStatus
|
||||
gdb.Find(&rows)
|
||||
statusMap := make(map[string]model.SystemTaskStatus, len(rows))
|
||||
for _, r := range rows {
|
||||
statusMap[r.Code] = r
|
||||
}
|
||||
|
||||
items := make([]Item, 0, len(catalog))
|
||||
for _, d := range catalog {
|
||||
item := Item{Def: d}
|
||||
if st, ok := statusMap[d.Code]; ok {
|
||||
item.LastRunAt = st.LastRunAt
|
||||
item.LastStatus = st.LastStatus
|
||||
item.LastMessage = st.LastMessage
|
||||
item.RunCount = st.RunCount
|
||||
item.LastDurationMs = st.LastDurationMs
|
||||
}
|
||||
if extras != nil {
|
||||
if ex, ok := extras[d.Code]; ok {
|
||||
item.Extra = ex
|
||||
}
|
||||
}
|
||||
// 审计清理任务补充保留天数 // Logan.Li
|
||||
if d.Code == "audit_log_cleanup" {
|
||||
days := auditcleaner.GetRetentionDays(gdb)
|
||||
if item.Extra == nil {
|
||||
item.Extra = map[string]any{"retention_days": days}
|
||||
}
|
||||
}
|
||||
if d.Code == "snmp_metrics_cleanup" && item.Extra == nil {
|
||||
item.Extra = map[string]any{"retention_days": snmppkg.MetricsRetentionDays}
|
||||
}
|
||||
items = append(items, item)
|
||||
}
|
||||
return items
|
||||
}
|
||||
|
||||
// RunManual 手动触发可运行的系统任务 // Logan.Li
|
||||
func RunManual(gdb *gorm.DB, driver, code string) (Item, error) {
|
||||
def, ok := findDef(code)
|
||||
if !ok {
|
||||
return Item{}, fmt.Errorf("未知系统任务: %s", code)
|
||||
}
|
||||
if !def.Runnable {
|
||||
return Item{}, fmt.Errorf("任务 %s 不支持手动执行", def.Name)
|
||||
}
|
||||
|
||||
start := time.Now()
|
||||
var msg string
|
||||
var err error
|
||||
|
||||
switch code {
|
||||
case "agent_report_cleanup":
|
||||
sql := database.CleanupAgentReportsSQL(driver)
|
||||
r := gdb.Exec(sql)
|
||||
err = r.Error
|
||||
msg = fmt.Sprintf("清理 %d 条 Agent 上报记录", r.RowsAffected)
|
||||
case "audit_log_cleanup":
|
||||
var result auditcleaner.Result
|
||||
result, err = auditcleaner.RunFromSettings(gdb)
|
||||
if err == nil {
|
||||
msg = fmt.Sprintf("清理 %d 条超期记录（保留 %d 天）", result.Total, result.RetentionDays)
|
||||
now := time.Now()
|
||||
gdb.Model(&model.AuditSettings{}).Where("id > 0").Updates(map[string]interface{}{
|
||||
"last_cleanup_at": now,
|
||||
"last_cleanup_summary": result.SummaryJSON(),
|
||||
})
|
||||
}
|
||||
case "snmp_metrics_cleanup":
|
||||
sql := database.CleanupSNMPMetricsSQL(driver, snmppkg.MetricsRetentionDays)
|
||||
r := gdb.Exec(sql)
|
||||
err = r.Error
|
||||
msg = fmt.Sprintf("清理 %d 条 SNMP 指标（保留 %d 天）", r.RowsAffected, snmppkg.MetricsRetentionDays)
|
||||
default:
|
||||
return Item{}, fmt.Errorf("任务 %s 未实现手动执行", code)
|
||||
}
|
||||
|
||||
duration := time.Since(start)
|
||||
status := "success"
|
||||
if err != nil {
|
||||
status = "failed"
|
||||
msg = err.Error()
|
||||
}
|
||||
Record(code, status, msg, duration)
|
||||
|
||||
items := List(gdb, nil)
|
||||
for _, it := range items {
|
||||
if it.Code == code {
|
||||
return it, err
|
||||
}
|
||||
}
|
||||
return Item{Def: def}, err
|
||||
}
|
||||
|
||||
func findDef(code string) (Def, bool) {
|
||||
for _, d := range catalog {
|
||||
if d.Code == code {
|
||||
return d, true
|
||||
}
|
||||
}
|
||||
return Def{}, false
|
||||
}
|
||||
Reference in new issue
Block a user