Add keepalive to wesher

(cherry picked from commit 982cac3b1a69d5759eaf477785301a564b551bc5)

README.md

@@ -145,6 +145,7 @@ All options can be passed either as command-line flags or environment variables:
 | `--interface DEV` | WESHER_INTERFACE | name of the wireguard interface to create and manage | `wgoverlay` |
 | `--no-etc-hosts` | WESHER_NO_ETC_HOSTS | whether to skip writing hosts entries for each node in mesh | `false` |
 | `--log-level LEVEL` | WESHER_LOG_LEVEL | set the verbosity (one of debug/info/warn/error) | `warn` |
+| `--keepalive-interval INTERVAL` | WESHER_KEEPALIVE_INTERVAL | interval for which to send keepalive packets | `30s` |
 
 ## Running multiple clusters
 

config.go

@@ -11,21 +11,22 @@ import (
 )
 
 type config struct {
-	ClusterKey    []byte   `id:"cluster-key" desc:"shared key for cluster membership; must be 32 bytes base64 encoded; will be generated if not provided"`
+	ClusterKey        []byte   `id:"cluster-key" desc:"shared key for cluster membership; must be 32 bytes base64 encoded; will be generated if not provided"`
-	Join          []string `desc:"comma separated list of hostnames or IP addresses to existing cluster members; if not provided, will attempt resuming any known state or otherwise wait for further members."`
+	Join              []string `desc:"comma separated list of hostnames or IP addresses to existing cluster members; if not provided, will attempt resuming any known state or otherwise wait for further members."`
-	Rejoin        int      `desc:"interval at which join nodes are joined again if away, 0 disables rejoining altogether" default:"0"`
+	Rejoin            int      `desc:"interval at which join nodes are joined again if away, 0 disables rejoining altogether" default:"0"`
-	Init          bool     `desc:"whether to explicitly (re)initialize the cluster; any known state from previous runs will be forgotten"`
+	Init              bool     `desc:"whether to explicitly (re)initialize the cluster; any known state from previous runs will be forgotten"`
-	BindAddr      string   `id:"bind-addr" desc:"IP address to bind to for cluster membership traffic (cannot be used with --bind-iface)"`
+	BindAddr          string   `id:"bind-addr" desc:"IP address to bind to for cluster membership traffic (cannot be used with --bind-iface)"`
-	BindIface     string   `id:"bind-iface" desc:"Interface to bind to for cluster membership traffic (cannot be used with --bind-addr)"`
+	BindIface         string   `id:"bind-iface" desc:"Interface to bind to for cluster membership traffic (cannot be used with --bind-addr)"`
-	ClusterPort   int      `id:"cluster-port" desc:"port used for membership gossip traffic (both TCP and UDP); must be the same across cluster" default:"7946"`
+	ClusterPort       int      `id:"cluster-port" desc:"port used for membership gossip traffic (both TCP and UDP); must be the same across cluster" default:"7946"`
-	WireguardPort int      `id:"wireguard-port" desc:"port used for wireguard traffic (UDP); must be the same across cluster" default:"51820"`
+	WireguardPort     int      `id:"wireguard-port" desc:"port used for wireguard traffic (UDP); must be the same across cluster" default:"51820"`
-	BaseMtu       int      `id:"mtu" desc:"MTU of the underlying network, taking intermediary hops into account" default:"1500"`
+	BaseMtu           int      `id:"mtu" desc:"MTU of the underlying network, taking intermediary hops into account" default:"1500"`
-	OverlayNet    *network `id:"overlay-net" desc:"the network in which to allocate addresses for the overlay mesh network (CIDR format); smaller networks increase the chance of IP collision" default:"10.0.0.0/8"`
+	OverlayNet        *network `id:"overlay-net" desc:"the network in which to allocate addresses for the overlay mesh network (CIDR format); smaller networks increase the chance of IP collision" default:"10.0.0.0/8"`
-	RoutedNet     *network `id:"routed-net" desc:"network used to filter routes that nodes are allowed to announce (CIDR format)" default:"0.0.0.0/32"`
+	RoutedNet         *network `id:"routed-net" desc:"network used to filter routes that nodes are allowed to announce (CIDR format)" default:"0.0.0.0/32"`
-	Interface     string   `desc:"name of the wireguard interface to create and manage" default:"wgoverlay"`
+	Interface         string   `desc:"name of the wireguard interface to create and manage" default:"wgoverlay"`
-	NoEtcHosts    bool     `id:"no-etc-hosts" desc:"disable writing of entries to /etc/hosts"`
+	NoEtcHosts        bool     `id:"no-etc-hosts" desc:"disable writing of entries to /etc/hosts"`
-	LogLevel      string   `id:"log-level" desc:"set the verbosity (debug/info/warn/error)" default:"warn"`
+	LogLevel          string   `id:"log-level" desc:"set the verbosity (debug/info/warn/error)" default:"warn"`
-	Version       bool     `desc:"display current version and exit"`
+	Version           bool     `desc:"display current version and exit"`
+	KeepaliveInterval string   `id:"keepalive-interval" desc:"interval for which to send keepalive packets" default:"30s"`
 
 	// for easier local testing; will break etchosts entry
 	UseIPAsName bool `id:"ip-as-name" default:"false" opts:"hidden"`

main.go

@@ -39,7 +39,13 @@ func main() {
 	if err != nil {
 		logrus.WithError(err).Fatal("could not create cluster")
 	}
-	wgstate, localNode, err := wg.New(config.Interface, config.WireguardPort, config.BaseMtu, (*net.IPNet)(config.OverlayNet), cluster.LocalName)
+
+	keepaliveDuration, err := time.ParseDuration(config.KeepaliveInterval)
+	if err != nil {
+		logrus.WithError(err).Fatal("could not parse time duration for keepalive")
+	}
+
+	wgstate, localNode, err := wg.New(config.Interface, config.WireguardPort, config.BaseMtu, (*net.IPNet)(config.OverlayNet), cluster.LocalName, &keepaliveDuration)
 	if err != nil {
 		logrus.WithError(err).Fatal("could not instantiate wireguard controller")
 	}

wg/wireguard.go

@@ -4,6 +4,7 @@ import (
 	"hash/fnv"
 	"net"
 	"os"
+	"time"
 
 	"github.com/costela/wesher/common"
 	"github.com/pkg/errors"
@@ -14,19 +15,20 @@ import (
 
 // State holds the configured state of a Wesher Wireguard interface
 type State struct {
-	iface       string
+	iface             string
-	client      *wgctrl.Client
+	client            *wgctrl.Client
-	OverlayAddr net.IPNet
+	OverlayAddr       net.IPNet
-	Port        int
+	Port              int
-	Mtu         int
+	Mtu               int
-	PrivKey     wgtypes.Key
+	PrivKey           wgtypes.Key
-	PubKey      wgtypes.Key
+	PubKey            wgtypes.Key
+	KeepaliveInterval *time.Duration
 }
 
 // New creates a new Wesher Wireguard state
 // The Wireguard keys are generated for every new interface
 // The interface must later be setup using SetUpInterface
-func New(iface string, port int, mtu int, ipnet *net.IPNet, name string) (*State, *common.Node, error) {
+func New(iface string, port int, mtu int, ipnet *net.IPNet, name string, keepaliveInterval *time.Duration) (*State, *common.Node, error) {
 	client, err := wgctrl.New()
 	if err != nil {
 		return nil, nil, errors.Wrap(err, "could not instantiate wireguard client")
@@ -39,12 +41,13 @@ func New(iface string, port int, mtu int, ipnet *net.IPNet, name string) (*State
 	pubKey := privKey.PublicKey()
 
 	state := State{
-		iface:   iface,
+		iface:             iface,
-		client:  client,
+		client:            client,
-		Port:    port,
+		Port:              port,
-		Mtu:     mtu,
+		Mtu:               mtu,
-		PrivKey: privKey,
+		PrivKey:           privKey,
-		PubKey:  pubKey,
+		PubKey:            pubKey,
+		KeepaliveInterval: keepaliveInterval,
 	}
 	state.assignOverlayAddr(ipnet, name)
 
@@ -192,7 +195,8 @@ func (s *State) nodesToPeerConfigs(nodes []common.Node) ([]wgtypes.PeerConfig, e
 				IP:   node.Addr,
 				Port: s.Port,
 			},
-			AllowedIPs: append([]net.IPNet{node.OverlayAddr}, node.Routes...),
+			AllowedIPs:                  append([]net.IPNet{node.OverlayAddr}, node.Routes...),
+			PersistentKeepaliveInterval: s.KeepaliveInterval,
 		}
 	}
 	return peerCfgs, nil