Create weave-no-masq-local ipset

The ipset is used to track locally allocated CIDRs. If requested,
external traffic sent to pods within these CIDR ranges will
avoid SNAT'ing (NYI).

net/bridge.go

@@ -11,7 +11,9 @@ import (
 	"github.com/sirupsen/logrus"
 	"github.com/vishvananda/netlink"
 
+	"github.com/weaveworks/weave/common"
 	"github.com/weaveworks/weave/common/odp"
+	"github.com/weaveworks/weave/net/ipset"
 )
 
 /* This code implements three possible configurations to connect
@@ -47,11 +49,12 @@ datapath of old kernel versions (https://github.com/weaveworks/weave/issues/1577
 */
 
 const (
-	WeaveBridgeName = "weave"
-	DatapathName    = "datapath"
-	DatapathIfName  = "vethwe-datapath"
-	BridgeIfName    = "vethwe-bridge"
-	PcapIfName      = "vethwe-pcap"
+	WeaveBridgeName  = "weave"
+	DatapathName     = "datapath"
+	DatapathIfName   = "vethwe-datapath"
+	BridgeIfName     = "vethwe-bridge"
+	PcapIfName       = "vethwe-pcap"
+	NoMasqLocalIpset = ipset.Name("weave-no-masq-local")
 )
 
 type Bridge interface {
@@ -213,6 +216,7 @@ type BridgeConfig struct {
 	MTU              int
 	Mac              string
 	Port             int
+	NoMasqLocal      bool
 }
 
 func (config *BridgeConfig) configuredBridgeType() Bridge {
@@ -489,8 +493,25 @@ func configureIPTables(config *BridgeConfig) error {
 	if err := ipt.ClearChain("nat", "WEAVE"); err != nil {
 		return errors.Wrap(err, "failed to clear/create nat/WEAVE chain")
 	}
+	if err := ipt.AppendUnique("nat", "POSTROUTING", "-j", "WEAVE"); err != nil {
+		return err
+	}
 
-	return ipt.AppendUnique("nat", "POSTROUTING", "-j", "WEAVE")
+	// k8s-only: Create the ipset to store CIDRs allocated by IPAM for local pods.
+	// External traffic sent to these CIDRs avoids SNAT'ing so that NodePort
+	// with `"externalTrafficPolicy":"Local"` would have the correct src IP addr.
+	if config.NoMasqLocal {
+		ips := ipset.New(common.LogLogger())
+		_ = ips.Destroy(NoMasqLocalIpset)
+		if err := ips.Create(NoMasqLocalIpset, ipset.HashNet); err != nil {
+			return err
+		}
+		if err := ipt.Insert("nat", "WEAVE", 1, "-m", "set", "--match-set", string(NoMasqLocalIpset), "dst", "-j", "RETURN"); err != nil {
+			return err
+		}
+	}
+
+	return nil
 }
 
 func linkSetUpByName(linkName string) error {

net/ipset/ipset.go

@@ -16,6 +16,7 @@ type Type string
 const (
 	ListSet = Type("list:set")
 	HashIP  = Type("hash:ip")
+	HashNet = Type("hash:net")
 )
 
 type Interface interface {

prog/weave-npc/main.go

@@ -18,8 +18,9 @@ import (
 	"k8s.io/client-go/tools/cache"
 
 	"github.com/weaveworks/weave/common"
-	"github.com/weaveworks/weave/npc"
+	"github.com/weaveworks/weave/net"
 	"github.com/weaveworks/weave/net/ipset"
+	"github.com/weaveworks/weave/npc"
 	"github.com/weaveworks/weave/npc/metrics"
 	"github.com/weaveworks/weave/npc/ulogd"
 )
@@ -56,7 +57,7 @@ func resetIPTables(ipt *iptables.IPTables) error {
 }
 
 func resetIPSets(ips ipset.Interface) error {
-	// Remove ipsets prefixed `weave-` only
+	// Remove ipsets prefixed `weave-` excluding weave-no-masq-local (used by weaver).
 
 	sets, err := ips.List(npc.IpsetNamePrefix)
 	if err != nil {
@@ -68,6 +69,9 @@ func resetIPSets(ips ipset.Interface) error {
 
 	// Must remove references to ipsets by other ipsets before they're destroyed
 	for _, s := range sets {
+		if s == ipset.Name(net.NoMasqLocalIpset) {
+			continue
+		}
 		common.Log.Debugf("Flushing ipset '%s'", string(s))
 		if err := ips.Flush(s); err != nil {
 			common.Log.Errorf("Failed to flush ipset '%s'", string(s))
@@ -76,6 +80,9 @@ func resetIPSets(ips ipset.Interface) error {
 	}
 
 	for _, s := range sets {
+		if s == ipset.Name(net.NoMasqLocalIpset) {
+			continue
+		}
 		common.Log.Debugf("Destroying ipset '%s'", string(s))
 		if err := ips.Destroy(s); err != nil {
 			common.Log.Errorf("Failed to destroy ipset '%s'", string(s))

prog/weaver/main.go

@@ -221,6 +221,7 @@ func main() {
 	mflag.BoolVar(&pluginConfig.EnableV2Multicast, []string{"-plugin-v2-multicast"}, false, "enable multicast for Docker plugin (v2)")
 	mflag.StringVar(&pluginConfig.Socket, []string{"-plugin-socket"}, "/run/docker/plugins/weave.sock", "plugin socket on which to listen")
 	mflag.StringVar(&pluginConfig.MeshSocket, []string{"-plugin-mesh-socket"}, "/run/docker/plugins/weavemesh.sock", "plugin socket on which to listen in mesh mode")
+	mflag.BoolVar(&bridgeConfig.NoMasqLocal, []string{"-no-masq-local"}, false, "do not SNAT external traffic sent to pods running on this node (Kubernetes only)")
 
 	proxyConfig := newProxyConfig()