Files
weave/prog/weave-npc/main.go
T
levaliciousandClaude Opus 5.5 9bc3415111
images / images (push) Successful in 14m46s
lb: LoadBalancer Services from addresses nodes hold, client address kept
A Service of type LoadBalancer with loadBalancerClass
reweave.levk.dev/node-address names its address in spec.loadBalancerIP.
Each node's weave-npc publishes the IPv4 addresses on its host interfaces
and its weave bridge (Node annotations reweave.levk.dev/addresses and
reweave.levk.dev/weave). The one node that holds a Service's address is
its holder: it alone writes status.loadBalancer (ipMode Proxy, so
kube-proxy installs nothing for the address) and DNATs the address's ports
to ready endpoints (EndpointSlices), without SNAT, from nat PREROUTING
ahead of kube-proxy. Every other node marks the connections the holder
forwards to its pods (new, from the holder's bridge MAC, from outside the
pod network) and routes their replies back to the holder's bridge through
a per-holder policy rule and table, so the holder's conntrack undoes the
DNAT and the reply leaves from the address the client connected to.

Fail closed: an address no node holds, or that several nodes hold, is
served by no node, and its status is cleared (by its holder, or by the
first node by name if none holds it). Compute is the policy as a pure
function; Datapath replaces exactly the chains it owns with
iptables-restore --noflush and keeps its rules and tables exact.

weave-npc runs it unless --lb=false; the ClusterRole gains services
(watch), services/status (patch), endpointslices (watch), nodes (patch).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UHXzGrsxfkGk8LmwrBbHNZ
2026-10-13 12:45:54 -04:00

368 lines
12 KiB
Go

package main
import (
"context"
"fmt"
"os"
"os/signal"
goruntime "runtime"
"syscall"
"github.com/coreos/go-iptables/iptables"
"github.com/spf13/cobra"
coreapi "k8s.io/api/core/v1"
networkingv1 "k8s.io/api/networking/v1"
"k8s.io/apimachinery/pkg/fields"
"k8s.io/apimachinery/pkg/runtime"
"k8s.io/apimachinery/pkg/util/wait"
"k8s.io/client-go/kubernetes"
"k8s.io/client-go/rest"
"k8s.io/client-go/tools/cache"
"github.com/rajch/weave/common"
"github.com/rajch/weave/common/chains"
"github.com/rajch/weave/lb"
"github.com/rajch/weave/net"
"github.com/rajch/weave/net/ipset"
"github.com/rajch/weave/npc"
"github.com/rajch/weave/npc/metrics"
"github.com/rajch/weave/npc/ulogd"
)
var (
version = "unreleased"
metricsAddr string
logLevel string
allowMcast bool
nodeName string
maxList int
bridgePortName string
serveLB bool
)
func handleError(err error) {
if err != nil {
metrics.PolicyEnforcementErrors.Inc()
common.Log.Error(err)
}
}
func handleFatal(err error) { common.CheckFatal(err) }
func makeController(getter cache.Getter, resource string,
objType runtime.Object, handlers cache.ResourceEventHandlerFuncs) cache.Controller {
listWatch := cache.NewListWatchFromClient(getter, resource, "", fields.Everything())
_, controller := cache.NewInformer(listWatch, objType, 0, handlers)
return controller
}
func resetIPTables(ipt *iptables.IPTables) error {
// Flush chains first so there are no refs to extant ipsets
if err := ipt.ClearChain(npc.TableFilter, chains.IngressChain); err != nil {
return err
}
if err := ipt.ClearChain(npc.TableFilter, chains.DefaultChain); err != nil {
return err
}
if err := ipt.ClearChain(npc.TableFilter, chains.MainChain); err != nil {
return err
}
if err := ipt.ClearChain(npc.TableFilter, chains.EgressMarkChain); err != nil {
return err
}
if err := ipt.ClearChain(npc.TableFilter, chains.EgressCustomChain); err != nil {
return err
}
if err := ipt.ClearChain(npc.TableFilter, chains.EgressDefaultChain); err != nil {
return err
}
// We do not clear npc.EgressChain here because otherwise, in the case of restarting
// weave-npc process, all egress traffic is allowed for a short period of time.
// The chain is created in createBaseRules.
return nil
}
func resetIPSets(ips ipset.Interface) error {
// Remove ipsets prefixed `weave-` only.
sets, err := ips.List(npc.IpsetNamePrefix)
if err != nil {
common.Log.Errorf("Failed to retrieve list of ipsets")
return err
}
common.Log.Debugf("Got list of ipsets: %v", sets)
// Must remove references to ipsets by other ipsets before they're destroyed
for _, s := range sets {
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))
return err
}
}
for _, s := range sets {
// LocalIPset might be used by WEAVE-NPC-EGRESS chain which we do not
// flush, so we cannot destroy it.
if s == npc.LocalIpset {
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))
return err
}
}
return nil
}
func createBaseRules(ipt *iptables.IPTables, ips ipset.Interface) error {
// Configure main chain static rules
if err := ipt.Append(npc.TableFilter, chains.MainChain,
"-m", "state", "--state", "RELATED,ESTABLISHED", "-j", "ACCEPT"); err != nil {
return err
}
if allowMcast {
if err := ipt.Append(npc.TableFilter, chains.MainChain,
"-d", "224.0.0.0/4", "-j", "ACCEPT"); err != nil {
return err
}
}
// If the destination address is not any of the local pods, let it through
if err := ipt.Append(npc.TableFilter, chains.MainChain,
"-m", "physdev", "--physdev-is-bridged", "--physdev-out="+bridgePortName, "-j", "ACCEPT"); err != nil {
return err
}
if err := ipt.Append(npc.TableFilter, chains.MainChain,
"-m", "state", "--state", "NEW", "-j", chains.DefaultChain); err != nil {
return err
}
if err := ipt.Append(npc.TableFilter, chains.MainChain,
"-m", "state", "--state", "NEW", "-j", chains.IngressChain); err != nil {
return err
}
if err := ipt.Append(npc.TableFilter, chains.EgressMarkChain,
"-j", "MARK", "--set-xmark", npc.EgressMark); err != nil {
return err
}
// Egress rules:
//
// -A WEAVE-NPC-EGRESS -m state --state RELATED,ESTABLISHED -j ACCEPT
// -A WEAVE-NPC-EGRESS -m physdev --physdev-in vethwe-bridge --physdev-is-bridged -j RETURN
// -A WEAVE-NPC-EGRESS -m addrtype --dst-type LOCAL -j RETURN
// -A WEAVE-NPC-EGRESS -m state --state NEW -j WEAVE-NPC-EGRESS-DEFAULT
// -A WEAVE-NPC-EGRESS -m state --state NEW -m mark ! --mark 0x40000/0x40000 -j WEAVE-NPC-EGRESS-CUSTOM
// -A WEAVE-NPC-EGRESS -m state --state NEW -m mark ! --mark 0x40000/0x40000 -j NFLOG --nflog-group 86
// -A WEAVE-NPC-EGRESS -m mark ! --mark 0x40000/0x40000 -j DROP
//
// -A WEAVE-NPC-EGRESS-CUSTOM <rulespec> -j MARK --set-xmark 0x40000/0x40000
// -A WEAVE-NPC-EGRESS-CUSTOM <rulespec> -j RETURN
//
// -A WEAVE-NPC-EGRESS-DEFAULT <rulespec> -j MARK --set-xmark 0x40000/0x40000
// -A WEAVE-NPC-EGRESS-DEFAULT <rulespec> -j RETURN
//
// For each rule we create two (mark and return). We cannot just accept
// a packet if it matches any rule, as a packet might need to traverse
// the ingress npc as well which happens later in the chain (in some cases
// we cannot detect whether packet is ingress or egress, so we need to
// check both chains).
ruleSpecs := [][]string{
{"-m", "state", "--state", "RELATED,ESTABLISHED", "-j", "ACCEPT"},
// skip running through egress network policies for the traffic not coming from local pods
{"-m", "physdev", "--physdev-is-bridged", "--physdev-in=" + bridgePortName, "-j", "RETURN"},
// skip running through egress network policies for the traffic bound for IP address assigned for the bridge
{"-m", "addrtype", "--dst-type", "LOCAL", "-j", "RETURN"},
}
if allowMcast {
ruleSpecs = append(ruleSpecs, []string{"-d", "224.0.0.0/4", "-j", "RETURN"})
}
ruleSpecs = append(ruleSpecs, [][]string{
{"-m", "state", "--state", "NEW", "-j", chains.EgressDefaultChain},
{"-m", "state", "--state", "NEW", "-m", "mark", "!", "--mark", npc.EgressMark, "-j", chains.EgressCustomChain},
}...)
if err := net.AddChainWithRules(ipt, npc.TableFilter, chains.EgressChain, ruleSpecs); err != nil {
return err
}
// delete `weave-local-pods` ipset which is no longer used by weave-npc
weaveLocalPodExist, err := ips.Exists(npc.LocalIpset)
if err != nil {
common.Log.Errorf("Failed to look if ipset '%s' exists", npc.LocalIpset)
} else if weaveLocalPodExist {
common.Log.Debugf("Destroying ipset '%s'", npc.LocalIpset)
if err := ips.Destroy(npc.LocalIpset); err != nil {
common.Log.Errorf("Failed to destroy ipset '%s'", npc.LocalIpset)
}
}
return nil
}
// Indirection in case we come up with a more interesting requirement
func hCtx() context.Context {
return context.Background()
}
func stopOnPanicRecover() {
if r := recover(); r != nil {
buf := make([]byte, 1<<20)
stacklen := goruntime.Stack(buf, false)
handleFatal(fmt.Errorf("panic: %v \n%s", r, buf[:stacklen]))
}
}
func root(cmd *cobra.Command, args []string) {
var npController cache.Controller
common.SetLogLevel(logLevel)
if nodeName == "" {
// HOSTNAME is set by Kubernetes for pods in the host network namespace
nodeName = os.Getenv("HOSTNAME")
}
if nodeName == "" {
common.Log.Fatalf("Must set node name via --node-name or $HOSTNAME")
}
common.Log.Infof("Starting Weaveworks NPC %s; node name %q", version, nodeName)
if err := metrics.Start(metricsAddr); err != nil {
common.Log.Fatalf("Failed to start metrics: %v", err)
}
if err := ulogd.Start(); err != nil {
common.Log.Fatalf("Failed to start ulogd: %v", err)
}
config, err := rest.InClusterConfig()
handleFatal(err)
client, err := kubernetes.NewForConfig(config)
handleFatal(err)
ipt, err := iptables.New()
handleFatal(err)
ips := ipset.New(common.LogLogger(), maxList)
handleFatal(resetIPTables(ipt))
handleFatal(resetIPSets(ips))
handleFatal(createBaseRules(ipt, ips))
npc := npc.New(nodeName, ipt, ips, client)
nsController := makeController(client.CoreV1().RESTClient(), "namespaces", &coreapi.Namespace{},
cache.ResourceEventHandlerFuncs{
AddFunc: func(obj interface{}) {
defer stopOnPanicRecover()
handleError(npc.AddNamespace(hCtx(), obj.(*coreapi.Namespace)))
},
DeleteFunc: func(obj interface{}) {
defer stopOnPanicRecover()
switch obj := obj.(type) {
case *coreapi.Namespace:
handleError(npc.DeleteNamespace(hCtx(), obj))
case cache.DeletedFinalStateUnknown:
// We know this object has gone away, but its final state is no longer
// available from the API server. Instead we use the last copy of it
// that we have, which is good enough for our cleanup.
handleError(npc.DeleteNamespace(hCtx(), obj.Obj.(*coreapi.Namespace)))
}
},
UpdateFunc: func(old, new interface{}) {
defer stopOnPanicRecover()
handleError(npc.UpdateNamespace(hCtx(), old.(*coreapi.Namespace), new.(*coreapi.Namespace)))
}})
podController := makeController(client.CoreV1().RESTClient(), "pods", &coreapi.Pod{},
cache.ResourceEventHandlerFuncs{
AddFunc: func(obj interface{}) {
defer stopOnPanicRecover()
handleError(npc.AddPod(hCtx(), obj.(*coreapi.Pod)))
},
DeleteFunc: func(obj interface{}) {
defer stopOnPanicRecover()
switch obj := obj.(type) {
case *coreapi.Pod:
handleError(npc.DeletePod(hCtx(), obj))
case cache.DeletedFinalStateUnknown:
// We know this object has gone away, but its final state is no longer
// available from the API server. Instead we use the last copy of it
// that we have, which is good enough for our cleanup.
handleError(npc.DeletePod(hCtx(), obj.Obj.(*coreapi.Pod)))
}
},
UpdateFunc: func(old, new interface{}) {
defer stopOnPanicRecover()
handleError(npc.UpdatePod(hCtx(), old.(*coreapi.Pod), new.(*coreapi.Pod)))
}})
npHandlers := cache.ResourceEventHandlerFuncs{
AddFunc: func(obj interface{}) {
defer stopOnPanicRecover()
handleError(npc.AddNetworkPolicy(hCtx(), obj))
},
DeleteFunc: func(obj interface{}) {
defer stopOnPanicRecover()
switch obj := obj.(type) {
case cache.DeletedFinalStateUnknown:
// We know this object has gone away, but its final state is no longer
// available from the API server. Instead we use the last copy of it
// that we have, which is good enough for our cleanup.
handleError(npc.DeleteNetworkPolicy(hCtx(), obj.Obj))
default:
handleError(npc.DeleteNetworkPolicy(hCtx(), obj))
}
},
UpdateFunc: func(old, new interface{}) {
defer stopOnPanicRecover()
handleError(npc.UpdateNetworkPolicy(hCtx(), old, new))
},
}
npController = makeController(client.NetworkingV1().RESTClient(), "networkpolicies", &networkingv1.NetworkPolicy{}, npHandlers)
go nsController.Run(wait.NeverStop)
go podController.Run(wait.NeverStop)
go npController.Run(wait.NeverStop)
if serveLB {
// LoadBalancer Services of class lb.Class: addresses this node holds, client address kept
go lb.NewController(nodeName, client, lb.NewDatapath(ipt)).Run(context.Background())
}
signals := make(chan os.Signal, 1)
signal.Notify(signals, syscall.SIGINT, syscall.SIGTERM)
common.Log.Fatalf("Exiting: %v", <-signals)
}
func main() {
rootCmd := &cobra.Command{
Use: "weave-npc",
Short: "Weaveworks Kubernetes Network Policy Controller",
Run: root}
rootCmd.PersistentFlags().StringVar(&metricsAddr, "metrics-addr", ":6781", "metrics server bind address")
rootCmd.PersistentFlags().StringVar(&logLevel, "log-level", "debug", "logging level (debug, info, warning, error)")
rootCmd.PersistentFlags().BoolVar(&allowMcast, "allow-mcast", true, "allow all multicast traffic")
rootCmd.PersistentFlags().StringVar(&nodeName, "node-name", "", "only generate rules that apply to this node")
rootCmd.PersistentFlags().IntVar(&maxList, "max-list-size", 1024, "maximum size of ipset list (for namespaces)")
rootCmd.PersistentFlags().StringVar(&bridgePortName, "bridge-port-name", "vethwe-bridge", "name of the brige port on which packets are received and sent")
rootCmd.PersistentFlags().BoolVar(&serveLB, "lb", true, "serve LoadBalancer Services of class "+lb.Class+" from addresses nodes hold")
handleFatal(rootCmd.Execute())
}