feat: nocloud prefer mac address

Use MAC address over network interface name.

Signed-off-by: Serge Logvinov <serge.logvinov@sinextra.dev>
Signed-off-by: Andrey Smirnov <andrey.smirnov@siderolabs.com>

Author: sergelogvinov

internal/app/machined/pkg/runtime/v1alpha1/platform/internal/netutils/netutils.go

@@ -7,9 +7,12 @@ package netutils
 
 import (
 	"context"
+	"fmt"
 	"log"
 	"time"
 
+	"github.com/cenkalti/backoff/v4"
+	"github.com/cosi-project/runtime/pkg/safe"
 	"github.com/cosi-project/runtime/pkg/state"
 	"github.com/siderolabs/go-retry/retry"
 
@@ -24,6 +27,36 @@ func Wait(ctx context.Context, r state.State) error {
 	return network.NewReadyCondition(r, network.AddressReady).Wait(ctx)
 }
 
+// WaitInterfaces for the interfaces to be up to interact with platform metadata services.
+func WaitInterfaces(ctx context.Context, r state.State) error {
+	backoff := backoff.NewExponentialBackOff()
+	backoff.MaxInterval = 2 * time.Second
+	backoff.MaxElapsedTime = 30 * time.Second
+
+	for ctx.Err() == nil {
+		hostInterfaces, err := safe.StateListAll[*network.LinkStatus](ctx, r)
+		if err != nil {
+			return fmt.Errorf("error listing host interfaces: %w", err)
+		}
+
+		if hostInterfaces.Len() != 0 {
+			return nil
+		}
+
+		log.Printf("waiting for network interface appearse...")
+
+		interval := backoff.NextBackOff()
+
+		select {
+		case <-ctx.Done():
+			return nil
+		case <-time.After(interval):
+		}
+	}
+
+	return nil
+}
+
 // RetryFetch retries fetching from metadata service.
 func RetryFetch(ctx context.Context, f func(ctx context.Context) (string, error)) (string, error) {
 	var (

internal/app/machined/pkg/runtime/v1alpha1/platform/nocloud/metadata.go

@@ -270,8 +270,19 @@ func (n *Nocloud) acquireConfig(ctx context.Context, r state.State) (metadataCon
 	return metadataConfigDl, metadataNetworkConfigDl, machineConfigDl, metadata, err
 }
 
-//nolint:gocyclo
-func (n *Nocloud) applyNetworkConfigV1(config *NetworkConfig, networkConfig *runtime.PlatformNetworkConfig) error {
+//nolint:gocyclo,cyclop
+func (n *Nocloud) applyNetworkConfigV1(config *NetworkConfig, st state.State, networkConfig *runtime.PlatformNetworkConfig) error {
+	ctx := context.TODO()
+
+	if err := netutils.WaitInterfaces(ctx, st); err != nil {
+		return err
+	}
+
+	hostInterfaces, err := safe.StateListAll[*network.LinkStatus](ctx, st)
+	if err != nil {
+		return fmt.Errorf("error listing host interfaces: %w", err)
+	}
+
 	for _, ntwrk := range config.Config {
 		switch ntwrk.Type {
 		case "nameserver":
@@ -290,8 +301,29 @@ func (n *Nocloud) applyNetworkConfigV1(config *NetworkConfig, networkConfig *run
 				ConfigLayer: network.ConfigPlatform,
 			})
 		case "physical":
+			name := ntwrk.Interfaces
+
+			if ntwrk.Mac != "" {
+				macAddressMatched := false
+				hostInterfaceIter := hostInterfaces.Iterator()
+
+				for hostInterfaceIter.Next() {
+					macAddress := hostInterfaceIter.Value().TypedSpec().PermanentAddr.String()
+					if macAddress == ntwrk.Mac {
+						name = hostInterfaceIter.Value().Metadata().ID()
+						macAddressMatched = true
+
+						break
+					}
+				}
+
+				if !macAddressMatched {
+					log.Printf("nocloud: no link with matching MAC address %q, defaulted to use name %s instead", ntwrk.Mac, name)
+				}
+			}
+
 			networkConfig.Links = append(networkConfig.Links, network.LinkSpecSpec{
-				Name:        ntwrk.Interfaces,
+				Name:        name,
 				Up:          true,
 				ConfigLayer: network.ConfigPlatform,
 			})
@@ -301,7 +333,7 @@ func (n *Nocloud) applyNetworkConfigV1(config *NetworkConfig, networkConfig *run
 				case "dhcp", "dhcp4":
 					networkConfig.Operators = append(networkConfig.Operators, network.OperatorSpecSpec{
 						Operator:  network.OperatorDHCP4,
-						LinkName:  ntwrk.Interfaces,
+						LinkName:  name,
 						RequireUp: true,
 						DHCP4: network.DHCP4OperatorSpec{
 							RouteMetric: 1024,
@@ -335,7 +367,7 @@ func (n *Nocloud) applyNetworkConfigV1(config *NetworkConfig, networkConfig *run
 					networkConfig.Addresses = append(networkConfig.Addresses,
 						network.AddressSpecSpec{
 							ConfigLayer: network.ConfigPlatform,
-							LinkName:    ntwrk.Interfaces,
+							LinkName:    name,
 							Address:     ipPrefix,
 							Scope:       nethelpers.ScopeGlobal,
 							Flags:       nethelpers.AddressFlags(nethelpers.AddressPermanent),
@@ -352,7 +384,7 @@ func (n *Nocloud) applyNetworkConfigV1(config *NetworkConfig, networkConfig *run
 						route := network.RouteSpecSpec{
 							ConfigLayer: network.ConfigPlatform,
 							Gateway:     gw,
-							OutLinkName: ntwrk.Interfaces,
+							OutLinkName: name,
 							Table:       nethelpers.TableMain,
 							Protocol:    nethelpers.ProtocolStatic,
 							Type:        nethelpers.TypeUnicast,
@@ -371,7 +403,7 @@ func (n *Nocloud) applyNetworkConfigV1(config *NetworkConfig, networkConfig *run
 				case "ipv6_dhcpv6-stateful":
 					networkConfig.Operators = append(networkConfig.Operators, network.OperatorSpecSpec{
 						Operator:  network.OperatorDHCP6,
-						LinkName:  ntwrk.Interfaces,
+						LinkName:  name,
 						RequireUp: true,
 						DHCP6: network.DHCP6OperatorSpec{
 							RouteMetric: 1024,

internal/app/machined/pkg/runtime/v1alpha1/platform/nocloud/nocloud.go

@@ -52,7 +52,7 @@ func (n *Nocloud) ParseMetadata(unmarshalledNetworkConfig *NetworkConfig, st sta
 
 	switch unmarshalledNetworkConfig.Version {
 	case 1:
-		if err := n.applyNetworkConfigV1(unmarshalledNetworkConfig, networkConfig); err != nil {
+		if err := n.applyNetworkConfigV1(unmarshalledNetworkConfig, st, networkConfig); err != nil {
 			return nil, err
 		}
 	case 2:

internal/app/machined/pkg/runtime/v1alpha1/platform/nocloud/nocloud_test.go

@@ -58,6 +58,10 @@ func TestParseMetadata(t *testing.T) {
 
 			st := state.WrapCore(namespaced.NewState(inmem.Build))
 
+			eth0 := network.NewLinkStatus(network.NamespaceName, "eth0")
+			eth0.TypedSpec().PermanentAddr = nethelpers.HardwareAddr{0x68, 0x05, 0xca, 0xb8, 0xf1, 0xf7}
+			require.NoError(t, st.Create(context.TODO(), eth0))
+
 			eth1 := network.NewLinkStatus(network.NamespaceName, "eth1")
 			eth1.TypedSpec().PermanentAddr = nethelpers.HardwareAddr{0x68, 0x05, 0xca, 0xb8, 0xf1, 0xf8}
 			require.NoError(t, st.Create(context.TODO(), eth1))

internal/app/machined/pkg/runtime/v1alpha1/platform/nocloud/testdata/expected-v1.yaml

@@ -5,12 +5,24 @@ addresses:
       scope: global
       flags: permanent
       layer: platform
-    - address: 2001:2:3:4:5:6:7:8/64
+    - address: 2001:2:3:4:5:6:7:f7/64
       linkName: eth0
       family: inet6
       scope: global
       flags: permanent
       layer: platform
+    - address: 192.168.2.11/24
+      linkName: eth2
+      family: inet4
+      scope: global
+      flags: permanent
+      layer: platform
+    - address: 2001:2:3:4:5:6:7:f9/64
+      linkName: eth2
+      family: inet6
+      scope: global
+      flags: permanent
+      layer: platform
 links:
     - name: eth0
       logical: false
@@ -19,6 +31,13 @@ links:
       kind: ""
       type: netrom
       layer: platform
+    - name: eth2
+      logical: false
+      up: true
+      mtu: 0
+      kind: ""
+      type: netrom
+      layer: platform
 routes:
     - family: inet4
       dst: ""
@@ -44,6 +63,30 @@ routes:
       flags: ""
       protocol: static
       layer: platform
+    - family: inet4
+      dst: ""
+      src: ""
+      gateway: 192.168.2.1
+      outLinkName: eth2
+      table: main
+      priority: 1024
+      scope: global
+      type: unicast
+      flags: ""
+      protocol: static
+      layer: platform
+    - family: inet6
+      dst: ""
+      src: ""
+      gateway: fe80::2
+      outLinkName: eth2
+      table: main
+      priority: 2048
+      scope: global
+      type: unicast
+      flags: ""
+      protocol: static
+      layer: platform
 hostnames:
     - hostname: talos
       domainname: fqdn

internal/app/machined/pkg/runtime/v1alpha1/platform/nocloud/testdata/metadata-v1.yaml

@@ -2,15 +2,26 @@ version: 1
 config:
   - type: physical
     name: eth0
-    mac_address: 'ae:71:9e:61:d0:ad'
+    mac_address: '68:05:ca:b8:f1:f7'
     subnets:
     - type: static
       address: '192.168.1.11'
       netmask: '255.255.255.0'
       gateway: '192.168.1.1'
     - type: static6
-      address: '2001:2:3:4:5:6:7:8/64'
+      address: '2001:2:3:4:5:6:7:f7/64'
       gateway: 'fe80::1'
+  - type: physical
+    name: eth1
+    mac_address: '68:05:ca:b8:f1:f9'
+    subnets:
+    - type: static
+      address: '192.168.2.11'
+      netmask: '255.255.255.0'
+      gateway: '192.168.2.1'
+    - type: static6
+      address: '2001:2:3:4:5:6:7:f9/64'
+      gateway: 'fe80::2'
   - type: nameserver
     address:
     - '192.168.1.1'