feat: add metricsServer for health checks, metrics and pprof (#5578)

Add a separate HTTP listener for observability, configured through
metricsServer in both frps and frpc. It serves /healthz and, when
pprofEnable is set, the pprof handlers except /debug/pprof/cmdline. It
never serves the dashboard, static assets or the admin API, and it does
not touch the global assets filesystem.

frps also serves /metrics on this listener with both runtime and business
metrics, enabled by metricsServer.port alone. The in-memory collector
still starts only with the dashboard, and the Prometheus collector starts
either with metricsServer or with enablePrometheus plus webServer, so a
configuration that collected nothing before still collects nothing.

The /healthz, /metrics and pprof endpoints on webServer, and
enablePrometheus, keep working but are deprecated and log a hint at
startup. frpc does not serve /metrics yet.
This commit is contained in:
fatedier
2026-10-19 20:08:39 +08:00
committed by GitHub
parent 00e4fc4dd3
commit f13fdd58d4
20 changed files with 858 additions and 27 deletions
+32 -3
View File
@@ -123,7 +123,8 @@ type Service struct {
auth *auth.ClientAuth
// web server for admin UI and apis
webServer *httppkg.Server
webServer *httppkg.Server
metricsServer *httppkg.Server
vnetController *vnet.Controller
@@ -188,8 +189,8 @@ func NewService(options ServiceOptions) (*Service, error) {
proxyCfgs = config.CompleteProxyConfigurers(proxyCfgs)
visitorCfgs = config.CompleteVisitorConfigurers(visitorCfgs)
// Create the web server after all fallible steps so its listener is not
// leaked when an earlier error causes NewService to return.
// Create HTTP servers after all other fallible steps, rolling back the web
// listener if the metrics server cannot be created.
var webServer *httppkg.Server
if options.Common.WebServer.Port > 0 {
ws, err := httppkg.NewServer(options.Common.WebServer)
@@ -198,11 +199,19 @@ func NewService(options ServiceOptions) (*Service, error) {
}
webServer = ws
}
metricsServer, err := httppkg.NewMetricsServer(options.Common.MetricsServer, nil)
if err != nil {
if webServer != nil {
_ = webServer.Close()
}
return nil, err
}
s := &Service{
ctx: context.Background(),
auth: authRuntime,
webServer: webServer,
metricsServer: metricsServer,
common: options.Common,
reloadCommon: options.Common,
configFilePath: options.ConfigFilePath,
@@ -263,6 +272,12 @@ func (svr *Service) Run(ctx context.Context) error {
}
if svr.webServer != nil {
if svr.common.WebServer.PprofEnable {
log.Warnf("webServer pprof endpoints are deprecated; use metricsServer.pprofEnable instead")
}
if svr.common.MetricsServer.Port == 0 {
log.Infof("webServer /healthz is deprecated; health probes can migrate to metricsServer.port")
}
webServer := svr.webServer
go func() {
log.Infof("admin server listen on %s", webServer.Address())
@@ -272,6 +287,16 @@ func (svr *Service) Run(ctx context.Context) error {
}()
}
if svr.metricsServer != nil {
metricsServer := svr.metricsServer
go func() {
log.Infof("metrics server listen on %s", metricsServer.Address())
if err := metricsServer.Run(); err != nil && !errors.Is(err, http.ErrServerClosed) {
log.Warnf("metrics server exit with error: %v", err)
}
}()
}
// first login to frps
svr.loopLoginUntilSuccess(10*time.Second, lo.FromPtr(svr.common.LoginFailExit))
if svr.ctl == nil {
@@ -497,6 +522,10 @@ func (svr *Service) stop() {
svr.webServer.Close()
svr.webServer = nil
}
if svr.metricsServer != nil {
svr.metricsServer.Close()
svr.metricsServer = nil
}
if svr.vnetController != nil {
_ = svr.vnetController.Stop()
svr.vnetController = nil