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 2026-07-01 23:44:45 +08:00
1 parent 092d0e4b9e
commit 471353e3d7
190 files changed
+16640 -3484

No files matched your search

+159
View File
@@ -0,0 +1,159 @@
// 脚本运行时探测与本地执行 — 适配 ./aiops 单二进制 / Docker 容器环境 // Logan.Li
package scriptrunner
import (
"bytes"
"context"
"fmt"
"os"
"os/exec"
"runtime"
"time"
)
// RuntimeInfo 当前主机可用的脚本解释器
type RuntimeInfo struct {
Shell bool `json:"shell"`
ShellPath string `json:"shell_path,omitempty"`
Python bool `json:"python"`
PythonPath string `json:"python_path,omitempty"`
PowerShell bool `json:"powershell"`
PowerShellPath string `json:"powershell_path,omitempty"`
OS string `json:"os"`
Arch string `json:"arch"`
EmbedMode bool `json:"embed_mode"` // 单二进制 embed 部署
}
// DetectRuntime 探测 bash/sh、python3、powershell 是否可用 // Logan.Li
func DetectRuntime() RuntimeInfo {
info := RuntimeInfo{
OS: runtime.GOOS,
Arch: runtime.GOARCH,
EmbedMode: true,
}
for _, p := range []string{"/bin/bash", "/usr/bin/bash", "bash"} {
if path, ok := lookExecutable(p); ok {
info.Shell = true
info.ShellPath = path
break
}
}
if !info.Shell {
for _, p := range []string{"/bin/sh", "/usr/bin/sh", "sh"} {
if path, ok := lookExecutable(p); ok {
info.Shell = true
info.ShellPath = path
break
}
}
}
for _, p := range []string{"/usr/bin/python3", "/usr/local/bin/python3", "python3"} {
if path, ok := lookExecutable(p); ok {
info.Python = true
info.PythonPath = path
break
}
}
for _, p := range []string{"/usr/bin/pwsh", "/usr/local/bin/pwsh", "pwsh", "powershell"} {
if path, ok := lookExecutable(p); ok {
info.PowerShell = true
info.PowerShellPath = path
break
}
}
return info
}
func lookExecutable(name string) (string, bool) {
path, err := exec.LookPath(name)
if err != nil {
return "", false
}
return path, true
}
// CanRun 判断指定脚本类型是否可在本机执行 // Logan.Li
func CanRun(rt RuntimeInfo, scriptType string) (bool, string) {
switch scriptType {
case "shell", "bash", "command", "":
if rt.Shell {
return true, ""
}
return false, "当前环境未找到 bash/sh,无法执行 Shell 脚本(Docker 镜像需安装 bash)"
case "python":
if rt.Python {
return true, ""
}
return false, "当前环境未找到 python3,无法执行 Python 脚本(请在镜像中安装 python3)"
case "powershell":
if rt.PowerShell {
return true, ""
}
return false, "当前环境未找到 PowerShell (pwsh),无法执行 PowerShell 脚本"
default:
return false, fmt.Sprintf("不支持的脚本类型: %s", scriptType)
}
}
// Run 在 AIOps 本机执行脚本,返回合并输出、退出码、耗时(ms) // Logan.Li
func Run(scriptType, content string, timeoutSec int) (output string, exitCode int, durationMs int, err error) {
rt := DetectRuntime()
ok, reason := CanRun(rt, scriptType)
if !ok {
return reason + "\n", -1, 0, fmt.Errorf("%s", reason)
}
start := time.Now()
timeout := time.Duration(timeoutSec) * time.Second
if timeout <= 0 {
timeout = 300 * time.Second
}
ctx, cancel := context.WithTimeout(context.Background(), timeout)
defer cancel()
var cmd *exec.Cmd
switch scriptType {
case "python":
py := rt.PythonPath
if py == "" {
py = "python3"
}
cmd = exec.CommandContext(ctx, py, "-c", content)
case "powershell":
pwsh := rt.PowerShellPath
if pwsh == "" {
pwsh = "pwsh"
}
cmd = exec.CommandContext(ctx, pwsh, "-NoProfile", "-NonInteractive", "-Command", content)
default:
sh := rt.ShellPath
if sh == "" {
sh = "bash"
}
cmd = exec.CommandContext(ctx, sh, "-c", content)
}
var out bytes.Buffer
cmd.Stdout = &out
cmd.Stderr = &out
cmd.Env = os.Environ()
runErr := cmd.Run()
durationMs = int(time.Since(start).Milliseconds())
output = out.String()
if ctx.Err() == context.DeadlineExceeded {
output += fmt.Sprintf("\n[超时] 执行超过 %d 秒被终止", timeoutSec)
return output, -1, durationMs, ctx.Err()
}
if runErr != nil {
if exitError, ok := runErr.(*exec.ExitError); ok {
exitCode = exitError.ExitCode()
} else {
exitCode = -1
output += "\n" + runErr.Error()
}
return output, exitCode, durationMs, runErr
}
return output, 0, durationMs, nil
}