CI / verify (push) Canceled after 0s
- 自动备份:每日一份(按最新备份是否在 24h 内判定,重启不重复)、保留份数可配、列表/下载/恢复/删除,恢复前先校验归档;顺手修 writeArchive 逐个检查 tar/gzip/文件 Close()(审查报告 §二) - 健康持续化:只存状态变更(up/down/unknown)并推导可用率与不可用区间;公开状态页默认关闭、访问码存 SHA-256、按 IP 限流、只暴露域名与可用率 - 隧道判定修复:RouteHealthCandidate 新增 tunnel_known,frps 管理接口不可用时不再把「判不出来」当成「不通」(此前会误报「隧道连续不通」);汇总与徽标同样区分「无法判定」 - 指标历史:metrics_samples 纵向点表(本机 CPU/内存/磁盘 + 每规则上下行,30 天保留、查询按桶平均、不做降采样表);仪表盘趋势卡支持 1h/24h/7d - 仪表盘布局:运行状态并入顶部统计区(运行时长补进「服务」卡);性能并入系统资源那一行,整行 4 列 - 测试:backup/metrics/frp/api/app 新增用例;gofmt、go build/vet/test、vue-tsc、vite build 全绿 - agent 无改动,AgentVersion 保持 0.12.2
191 lines
5.6 KiB
Go
191 lines
5.6 KiB
Go
package app
|
||
|
||
import (
|
||
"context"
|
||
"fmt"
|
||
"log/slog"
|
||
|
||
"github.com/fonu/fonu/internal/frp"
|
||
"github.com/fonu/fonu/internal/monitor"
|
||
"github.com/fonu/fonu/internal/traffic"
|
||
)
|
||
|
||
// monitorSource 把 frp 服务适配成 monitor.Source:monitor 只关心「状态快照」,
|
||
// 怎么从 frp / agent 取是这里的事,两边因此互不依赖。
|
||
type monitorSource struct {
|
||
frp *frp.Service
|
||
logger *slog.Logger
|
||
traffic *traffic.Collector
|
||
}
|
||
|
||
func (s monitorSource) Routes(ctx context.Context) ([]monitor.RouteStatus, error) {
|
||
servers, err := s.enabledServers(ctx)
|
||
if err != nil {
|
||
return nil, err
|
||
}
|
||
routes := make([]monitor.RouteStatus, 0, len(servers))
|
||
for _, server := range servers {
|
||
// 单台服务端探测失败不影响其它服务端:它本轮跳过,等下一轮
|
||
health, err := s.frp.RouteHealth(ctx, server.ID)
|
||
if err != nil {
|
||
s.logger.Warn("巡检跳过服务端隧道", "module", "MONITOR", "server_id", server.ID, "error", err.Error())
|
||
continue
|
||
}
|
||
for _, item := range health {
|
||
routes = append(routes, monitor.RouteStatus{
|
||
ServerID: server.ID,
|
||
Domain: item.Domain,
|
||
TunnelOK: item.TunnelOK,
|
||
DNSOK: item.DNSOK,
|
||
ServiceOK: item.ServiceOK,
|
||
Known: item.TunnelKnown,
|
||
Detail: item.TunnelDetail,
|
||
})
|
||
}
|
||
}
|
||
return routes, nil
|
||
}
|
||
|
||
func (s monitorSource) Servers(ctx context.Context) ([]monitor.ServerStatus, error) {
|
||
servers, err := s.enabledServers(ctx)
|
||
if err != nil {
|
||
return nil, err
|
||
}
|
||
out := make([]monitor.ServerStatus, 0, len(servers))
|
||
for _, server := range servers {
|
||
out = append(out, monitor.ServerStatus{
|
||
ID: server.ID,
|
||
Name: server.Name,
|
||
DesiredState: server.DesiredState,
|
||
ProcessState: server.ProcessState,
|
||
ConnectionState: server.ConnectionState,
|
||
Detail: server.LastError,
|
||
})
|
||
}
|
||
return out, nil
|
||
}
|
||
|
||
func (s monitorSource) Agents(ctx context.Context) ([]monitor.AgentStatus, error) {
|
||
servers, err := s.enabledServers(ctx)
|
||
if err != nil {
|
||
return nil, err
|
||
}
|
||
out := make([]monitor.AgentStatus, 0, len(servers))
|
||
for _, server := range servers {
|
||
// AgentClient 只在「地址与令牌都可用」时才返回 ok:没配 agent 的服务端不参与可达性判定
|
||
client, ok := s.frp.AgentClient(ctx, server.ID)
|
||
if !ok {
|
||
continue
|
||
}
|
||
status := monitor.AgentStatus{ID: server.ID, Name: server.Name, Reachable: true}
|
||
if err := client.Ping(ctx); err != nil {
|
||
status.Reachable = false
|
||
status.Detail = err.Error()
|
||
}
|
||
out = append(out, status)
|
||
}
|
||
return out, nil
|
||
}
|
||
|
||
func (s monitorSource) Drift(ctx context.Context) ([]monitor.DriftStatus, error) {
|
||
servers, err := s.enabledServers(ctx)
|
||
if err != nil {
|
||
return nil, err
|
||
}
|
||
out := make([]monitor.DriftStatus, 0, len(servers))
|
||
for _, server := range servers {
|
||
report, err := s.frp.RouteDrift(ctx, server.ID)
|
||
if err != nil {
|
||
s.logger.Warn("巡检跳过漂移检测", "module", "MONITOR", "server_id", server.ID, "error", err.Error())
|
||
continue
|
||
}
|
||
status := monitor.DriftStatus{ID: server.ID, Name: server.Name, Count: len(report.Items)}
|
||
for _, item := range report.Items {
|
||
status.Items = append(status.Items, fmt.Sprintf("%s %s", item.Domain, driftKindText(item.Kind)))
|
||
}
|
||
out = append(out, status)
|
||
}
|
||
return out, nil
|
||
}
|
||
|
||
// Traffic 返回若干域名的当前上下行速率:流量采集器内部按 host:port 累计,这里按域名合并。
|
||
func (s monitorSource) Traffic(ctx context.Context, domains []string) (map[string]monitor.TrafficRate, error) {
|
||
if s.traffic == nil {
|
||
return nil, nil
|
||
}
|
||
snapshot := s.traffic.SnapshotByHost(domains)
|
||
rates := make(map[string]monitor.TrafficRate, len(snapshot))
|
||
for host, item := range snapshot {
|
||
rates[host] = monitor.TrafficRate{Upload: item.UploadRate, Download: item.DownloadRate}
|
||
}
|
||
return rates, nil
|
||
}
|
||
|
||
func (s monitorSource) enabledServers(ctx context.Context) ([]frp.Server, error) {
|
||
servers, err := s.frp.ListServers(ctx)
|
||
if err != nil {
|
||
return nil, err
|
||
}
|
||
enabled := make([]frp.Server, 0, len(servers))
|
||
for _, server := range servers {
|
||
// 停用中的服务端是用户主动关掉的,不该参与任何告警判定
|
||
if server.Enabled {
|
||
enabled = append(enabled, server)
|
||
}
|
||
}
|
||
return enabled, nil
|
||
}
|
||
|
||
// driftKindText 与公网反代页漂移清单的文案保持一致,通知里才不用回页面去对照。
|
||
func driftKindText(kind string) string {
|
||
switch kind {
|
||
case frp.RouteDriftMissing:
|
||
return "未部署"
|
||
case frp.RouteDriftOrphan:
|
||
return "孤儿配置"
|
||
case frp.RouteDriftNotLoaded:
|
||
return "未生效"
|
||
case frp.RouteDriftContentMismatch:
|
||
return "内容不一致"
|
||
default:
|
||
return kind
|
||
}
|
||
}
|
||
|
||
// healthRecorder 把 monitor 的健康结论翻成 frp 侧样本并落库(
|
||
// 两侧的状态字面量由 healthStateValue 显式翻译,改动一侧会在这里编译期报错)。
|
||
type healthRecorder struct {
|
||
frp *frp.Service
|
||
}
|
||
|
||
func (r healthRecorder) RecordRouteHealth(ctx context.Context, samples []monitor.HealthSample) error {
|
||
converted := make([]frp.RouteHealthSample, 0, len(samples))
|
||
for _, sample := range samples {
|
||
converted = append(converted, frp.RouteHealthSample{
|
||
ServerID: sample.ServerID,
|
||
Domain: sample.Domain,
|
||
State: healthStateValue(sample.State),
|
||
Reason: sample.Reason,
|
||
Detail: sample.Detail,
|
||
})
|
||
}
|
||
_, err := r.frp.RecordRouteHealth(ctx, converted)
|
||
return err
|
||
}
|
||
|
||
func (r healthRecorder) PruneRouteHealth(ctx context.Context) error {
|
||
_, err := r.frp.PruneRouteHealth(ctx)
|
||
return err
|
||
}
|
||
|
||
func healthStateValue(state string) string {
|
||
switch state {
|
||
case monitor.HealthUp:
|
||
return frp.RouteHealthUp
|
||
case monitor.HealthDown:
|
||
return frp.RouteHealthDown
|
||
default:
|
||
return frp.RouteHealthUnknown
|
||
}
|
||
}
|