fix(daemon): handle OVS bridge creation race condition with exchangeLinkName (#6374)

When exchangeLinkName=true, a race condition exists between OVS vswitchd
processing the netlink rename notification and the bridge creation: OVS's
netdev cache still holds a reference to the old NIC name, causing the
bridge's kernel interface to never appear. Additionally, the retry path
fails because changeProviderNicName cannot detect the already-renamed
state.

Fix by:
1. Making changeProviderNicName idempotent - when the source link is not
   found, check if the target already exists as a non-OVS interface,
   indicating a prior successful rename.
2. Adding waitForBridgeInterface to poll for the kernel interface after
   bridge creation. If it doesn't appear within 5s, delete the stale
   OVSDB bridge record so the reconciliation loop can retry cleanly.

Signed-off-by: Mengxin Liu <liumengxinfly@gmail.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: oilbeater

pkg/daemon/init_linux.go

@@ -52,6 +52,12 @@ func (c *Controller) changeProviderNicName(current, target string) (bool, error)
 	link, err := netlink.LinkByName(current)
 	if err != nil {
 		if _, ok := err.(netlink.LinkNotFoundError); ok {
+			// Check if the NIC was already renamed from a previous attempt
+			targetLink, targetErr := netlink.LinkByName(target)
+			if targetErr == nil && targetLink.Type() != "openvswitch" {
+				klog.Infof("link %s already renamed to %s, skip rename", current, target)
+				return true, nil
+			}
 			klog.Infof("link %s not found, skip", current)
 			return false, nil
 		}

pkg/daemon/ovs.go

@@ -259,6 +259,19 @@ func (c *Controller) configExternalBridge(provider, bridge, nic string, exchange
 	if err != nil {
 		return fmt.Errorf("failed to create OVS bridge %s, %w: %q", bridge, err, output)
 	}
+
+	if exchangeLinkName {
+		if err := c.waitForBridgeInterface(bridge, 5*time.Second); err != nil {
+			// Bridge created in OVSDB but kernel interface not available.
+			// Delete the stale bridge to allow a clean retry.
+			klog.Warningf("OVS bridge %s interface not ready, cleaning up: %v", bridge, err)
+			if output, delErr := ovs.Exec(ovs.IfExists, "del-br", bridge); delErr != nil {
+				klog.Errorf("failed to delete stale bridge %s: %v, %q", bridge, delErr, output)
+			}
+			return err
+		}
+	}
+
 	if output, err = ovs.Exec("list-ports", bridge); err != nil {
 		return fmt.Errorf("failed to list ports of OVS bridge %s, %w: %q", bridge, err, output)
 	}
@@ -307,6 +320,17 @@ func (c *Controller) configExternalBridge(provider, bridge, nic string, exchange
 	return nil
 }
 
+func (c *Controller) waitForBridgeInterface(bridge string, timeout time.Duration) error {
+	deadline := time.Now().Add(timeout)
+	for time.Now().Before(deadline) {
+		if ok, _ := linkExists(bridge); ok {
+			return nil
+		}
+		time.Sleep(100 * time.Millisecond)
+	}
+	return fmt.Errorf("timed out waiting for OVS bridge %s kernel interface", bridge)
+}
+
 func initProviderChassisMac(provider string) error {
 	if err := addOvnMapping("ovn-chassis-mac-mappings", provider, util.GenerateMac(), false); err != nil {
 		klog.Error(err)