[client] add per-phase timing metrics for sync processing (#6533)

* Adds metrics sync phases time split to monitor costs

* Address review fixes

* Increment README.md with description on usage with debug bundles
This commit is contained in:
Riccardo Manfrin
2026-06-29 11:02:02 +02:00
committed by GitHub
parent 1409a1325a
commit 5711f0e38c
7 changed files with 468 additions and 34 deletions
+86 -33
View File
@@ -895,6 +895,16 @@ func (e *Engine) handleAutoUpdateVersion(autoUpdateSettings *mgmProto.AutoUpdate
e.updateManager.SetVersion(autoUpdateSettings.Version, autoUpdateSettings.AlwaysUpdate)
}
// phase times a sync sub-phase: it returns a function that records the elapsed
// duration when called. Starting the timer at the call site keeps inter-phase
// glue code out of the measurement.
func (e *Engine) phase(name string) func() {
start := time.Now()
return func() {
e.clientMetrics.RecordSyncPhase(e.ctx, name, time.Since(start))
}
}
func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
started := time.Now()
defer func() {
@@ -914,7 +924,10 @@ func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
e.handleAutoUpdateVersion(update.NetworkMap.PeerConfig.AutoUpdate)
}
if err := e.updateNetbirdConfig(update.GetNetbirdConfig()); err != nil {
done := e.phase("netbird_config")
err := e.updateNetbirdConfig(update.GetNetbirdConfig())
done()
if err != nil {
return err
}
@@ -928,11 +941,16 @@ func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
return nil
}
if err := e.updateChecksIfNew(update.Checks); err != nil {
done = e.phase("checks")
err = e.updateChecksIfNew(update.Checks)
done()
if err != nil {
return err
}
done = e.phase("persist")
e.persistSyncResponse(update)
done()
// only apply new changes and ignore old ones
if err := e.updateNetworkMap(nm); err != nil {
@@ -1371,13 +1389,16 @@ func (e *Engine) updateNetworkMap(networkMap *mgmProto.NetworkMap) error {
dnsConfig := toDNSConfig(protoDNSConfig, e.wgInterface.Address())
done := e.phase("dns_server")
if err := e.dnsServer.UpdateDNSServer(serial, dnsConfig); err != nil {
log.Errorf("failed to update dns server, err: %v", err)
}
done()
e.routeManager.SetDNSForwarderPort(dnsConfig.ForwarderPort)
// apply routes first, route related actions might depend on routing being enabled
done = e.phase("routes_classify")
routes := toRoutes(networkMap.GetRoutes())
serverRoutes, clientRoutes := e.routeManager.ClassifyRoutes(routes)
@@ -1386,29 +1407,60 @@ func (e *Engine) updateNetworkMap(networkMap *mgmProto.NetworkMap) error {
e.connMgr.UpdateRouteHAMap(clientRoutes)
log.Debugf("updated lazy connection manager with %d HA groups", len(clientRoutes))
}
done()
done = e.phase("routes_apply")
dnsRouteFeatureFlag := toDNSFeatureFlag(networkMap)
if err := e.routeManager.UpdateRoutes(serial, serverRoutes, clientRoutes, dnsRouteFeatureFlag); err != nil {
log.Errorf("failed to update routes: %v", err)
}
done()
done = e.phase("filtering")
if e.acl != nil {
e.acl.ApplyFiltering(networkMap, dnsRouteFeatureFlag)
}
done()
done = e.phase("dns_forwarder")
fwdEntries := toRouteDomains(e.config.WgPrivateKey.PublicKey().String(), routes)
e.updateDNSForwarder(dnsRouteFeatureFlag, fwdEntries)
done()
// Ingress forward rules
done = e.phase("forward_rules")
forwardingRules, err := e.updateForwardRules(networkMap.GetForwardingRules())
if err != nil {
log.Errorf("failed to update forward rules, err: %v", err)
}
done()
log.Debugf("got peers update from Management Service, total peers to connect to = %d", len(networkMap.GetRemotePeers()))
done = e.phase("offline_peers")
e.updateOfflinePeers(networkMap.GetOfflinePeers())
done()
remotePeers, err := e.reconcilePeers(networkMap)
if err != nil {
return err
}
// must set the exclude list after the peers are added. Without it the manager can not figure out the peers parameters from the store
done = e.phase("lazy_exclude")
excludedLazyPeers := e.toExcludedLazyPeers(forwardingRules, remotePeers)
e.connMgr.SetExcludeList(e.ctx, excludedLazyPeers)
done()
e.networkSerial = serial
return nil
}
// reconcilePeers applies the remote peer list from the network map (removing,
// modifying and adding peers, then updating SSH config) and returns the remote
// peers with our own peer filtered out, for use by later sync steps.
func (e *Engine) reconcilePeers(networkMap *mgmProto.NetworkMap) ([]*mgmProto.RemotePeerConfig, error) {
// Filter out own peer from the remote peers list
localPubKey := e.config.WgPrivateKey.PublicKey().String()
remotePeers := make([]*mgmProto.RemotePeerConfig, 0, len(networkMap.GetRemotePeers()))
@@ -1423,42 +1475,43 @@ func (e *Engine) updateNetworkMap(networkMap *mgmProto.NetworkMap) error {
err := e.removeAllPeers()
e.statusRecorder.FinishPeerListModifications()
if err != nil {
return err
return nil, err
}
} else {
err := e.removePeers(remotePeers)
if err != nil {
return err
}
err = e.modifyPeers(remotePeers)
if err != nil {
return err
}
err = e.addNewPeers(remotePeers)
if err != nil {
return err
}
e.statusRecorder.FinishPeerListModifications()
e.updatePeerSSHHostKeys(remotePeers)
if err := e.updateSSHClientConfig(remotePeers); err != nil {
log.Warnf("failed to update SSH client config: %v", err)
}
e.updateSSHServerAuth(networkMap.GetSshAuth())
return remotePeers, nil
}
// must set the exclude list after the peers are added. Without it the manager can not figure out the peers parameters from the store
excludedLazyPeers := e.toExcludedLazyPeers(forwardingRules, remotePeers)
e.connMgr.SetExcludeList(e.ctx, excludedLazyPeers)
done := e.phase("removed_peers")
err := e.removePeers(remotePeers)
done()
if err != nil {
return nil, err
}
e.networkSerial = serial
done = e.phase("modified_peers")
err = e.modifyPeers(remotePeers)
done()
if err != nil {
return nil, err
}
return nil
done = e.phase("added_peers")
err = e.addNewPeers(remotePeers)
done()
if err != nil {
return nil, err
}
e.statusRecorder.FinishPeerListModifications()
e.updatePeerSSHHostKeys(remotePeers)
if err := e.updateSSHClientConfig(remotePeers); err != nil {
log.Warnf("failed to update SSH client config: %v", err)
}
e.updateSSHServerAuth(networkMap.GetSshAuth())
return remotePeers, nil
}
func toDNSFeatureFlag(networkMap *mgmProto.NetworkMap) bool {