Fix race condition in tunnel server startup

Several places in the code used a 5-second retry loop to wait on
Runtime.Core to be set. This caused a race condition where OnChange
handlers could be added after the Wrangler shared informers were already
started. When this happened, the handlers were never called because the
shared informers they relied upon were not started.

Fix that by requiring anything that waits on Runtime.Core to run from a
cluster controller startup hook that is guaranteed to be called before
the shared informers are started, instead of just firing it off in a
goroutine that retries until it is set.

Signed-off-by: Brad Davidson <[email protected]>

pkg/cluster/managed.go

@@ -82,7 +82,7 @@ func (c *Cluster) start(ctx context.Context) error {
 	if _, err := os.Stat(resetFile); err == nil {
 		// Before removing reset file we need to delete the node passwd secret in case the node
 		// password from the previously restored snapshot differs from the current password on disk.
-		go c.deleteNodePasswdSecret(ctx)
+		c.config.Runtime.ClusterControllerStarts["node-password-secret-cleanup"] = c.deleteNodePasswdSecret
 		os.Remove(resetFile)
 	}
 
@@ -176,30 +176,13 @@ func (c *Cluster) setupEtcdProxy(ctx context.Context, etcdProxy etcd.Proxy) {
 
 // deleteNodePasswdSecret wipes out the node password secret after restoration
 func (c *Cluster) deleteNodePasswdSecret(ctx context.Context) {
-	t := time.NewTicker(5 * time.Second)
-	defer t.Stop()
-	for range t.C {
-		nodeName := os.Getenv("NODE_NAME")
-		if nodeName == "" {
-			logrus.Infof("waiting for node name to be set")
-			continue
+	nodeName := os.Getenv("NODE_NAME")
+	secretsClient := c.config.Runtime.Core.Core().V1().Secret()
+	if err := nodepassword.Delete(secretsClient, nodeName); err != nil {
+		if apierrors.IsNotFound(err) {
+			logrus.Debugf("node password secret is not found for node %s", nodeName)
+			return
 		}
-		// the core factory may not yet be initialized so we
-		// want to wait until it is so not to evoke a panic.
-		if c.config.Runtime.Core == nil {
-			logrus.Infof("runtime is not yet initialized")
-			continue
-		}
-		secretsClient := c.config.Runtime.Core.Core().V1().Secret()
-		if err := nodepassword.Delete(secretsClient, nodeName); err != nil {
-			if apierrors.IsNotFound(err) {
-				logrus.Debugf("node password secret is not found for node %s", nodeName)
-				return
-			}
-			logrus.Warnf("failed to delete old node password secret: %v", err)
-			continue
-		}
-		return
+		logrus.Warnf("failed to delete old node password secret: %v", err)
 	}
-
 }

pkg/daemons/control/tunnel.go

@@ -10,7 +10,6 @@ import (
 	"strconv"
 	"strings"
 	"sync"
-	"time"
 
 	"github.com/k3s-io/k3s/pkg/daemons/config"
 	"github.com/k3s-io/k3s/pkg/daemons/control/proxy"
@@ -45,7 +44,7 @@ func setupTunnel(ctx context.Context, cfg *config.Control) (http.Handler, error)
 		server: remotedialer.New(authorizer, loggingErrorWriter),
 		egress: map[string]bool{},
 	}
-	go tunnel.watch(ctx)
+	cfg.Runtime.ClusterControllerStarts["tunnel-server"] = tunnel.watch
 	return tunnel, nil
 }
 
@@ -112,17 +111,10 @@ func (t *TunnelServer) watch(ctx context.Context) {
 		return
 	}
 
-	for {
-		if t.config.Runtime.Core != nil {
-			t.config.Runtime.Core.Core().V1().Node().OnChange(ctx, version.Program+"-tunnel-server", t.onChangeNode)
-			switch t.config.EgressSelectorMode {
-			case config.EgressSelectorModeCluster, config.EgressSelectorModePod:
-				t.config.Runtime.Core.Core().V1().Pod().OnChange(ctx, version.Program+"-tunnel-server", t.onChangePod)
-			}
-			return
-		}
-		logrus.Infof("Tunnel server egress proxy waiting for runtime core to become available")
-		time.Sleep(5 * time.Second)
+	t.config.Runtime.Core.Core().V1().Node().OnChange(ctx, version.Program+"-tunnel-server", t.onChangeNode)
+	switch t.config.EgressSelectorMode {
+	case config.EgressSelectorModeCluster, config.EgressSelectorModePod:
+		t.config.Runtime.Core.Core().V1().Pod().OnChange(ctx, version.Program+"-tunnel-server", t.onChangePod)
 	}
 }
 
@@ -173,7 +165,6 @@ func (t *TunnelServer) onChangePod(podName string, pod *v1.Pod) (*v1.Pod, error)
 		}
 	}
 	return pod, nil
-
 }
 
 // serveConnect attempts to handle the HTTP CONNECT request by dialing

pkg/server/server.go

@@ -168,8 +168,11 @@ func apiserverControllers(ctx context.Context, sc *Context, config *Config) {
 			panic(errors.Wrapf(err, "failed to start %s leader controller", util.GetFunctionName(controller)))
 		}
 	}
+
+	// Re-run context startup after core and leader-elected controllers have started. Additional
+	// informer caches may need to start for the newly added OnChange callbacks.
 	if err := sc.Start(ctx); err != nil {
-		panic(err)
+		panic(errors.Wrap(err, "failed to start wranger controllers"))
 	}
 }