netbird/client/internal/routemanager/iptables_linux.go
Zoltan Papp 0f0c7ec2ed
Routemgr error handling (#1073)
In case the route management feature is not supported 
then do not create unnecessary firewall and manager instances. 
This can happen if the nftables nor iptables is not available on the host OS.

- Move the error handling to upper layer
- Remove fake, useless implementations of interfaces
- Update go-iptables because In Docker the old version can not 
determine well the path of executable file
- update lib to 0.70
2023-08-12 11:42:36 +02:00

494 lines
14 KiB
Go

//go:build !android
package routemanager
import (
"context"
"fmt"
"net/netip"
"os/exec"
"strings"
"sync"
"github.com/coreos/go-iptables/iptables"
log "github.com/sirupsen/logrus"
)
func isIptablesSupported() bool {
_, err4 := exec.LookPath("iptables")
_, err6 := exec.LookPath("ip6tables")
return err4 == nil && err6 == nil
}
// constants needed to manage and create iptable rules
const (
iptablesFilterTable = "filter"
iptablesNatTable = "nat"
iptablesForwardChain = "FORWARD"
iptablesPostRoutingChain = "POSTROUTING"
iptablesRoutingNatChain = "NETBIRD-RT-NAT"
iptablesRoutingForwardingChain = "NETBIRD-RT-FWD"
routingFinalForwardJump = "ACCEPT"
routingFinalNatJump = "MASQUERADE"
)
// some presets for building nftable rules
var (
iptablesDefaultForwardingRule = []string{"-j", iptablesRoutingForwardingChain, "-m", "comment", "--comment"}
iptablesDefaultNetbirdForwardingRule = []string{"-j", "RETURN"}
iptablesDefaultNatRule = []string{"-j", iptablesRoutingNatChain, "-m", "comment", "--comment"}
iptablesDefaultNetbirdNatRule = []string{"-j", "RETURN"}
)
type iptablesManager struct {
ctx context.Context
stop context.CancelFunc
ipv4Client *iptables.IPTables
ipv6Client *iptables.IPTables
rules map[string]map[string][]string
mux sync.Mutex
}
func newIptablesManager(parentCtx context.Context) (*iptablesManager, error) {
ipv4Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
if err != nil {
return nil, err
} else if !isIptablesClientAvailable(ipv4Client) {
return nil, fmt.Errorf("iptables is missing for ipv4")
}
ipv6Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv6)
if err != nil {
log.Debugf("failed to initialize iptables for ipv6: %s", err)
} else if !isIptablesClientAvailable(ipv6Client) {
log.Infof("iptables is missing for ipv6")
ipv6Client = nil
}
ctx, cancel := context.WithCancel(parentCtx)
return &iptablesManager{
ctx: ctx,
stop: cancel,
ipv4Client: ipv4Client,
ipv6Client: ipv6Client,
rules: make(map[string]map[string][]string),
}, nil
}
// CleanRoutingRules cleans existing iptables resources that we created by the agent
func (i *iptablesManager) CleanRoutingRules() {
i.mux.Lock()
defer i.mux.Unlock()
err := i.cleanJumpRules()
if err != nil {
log.Error(err)
}
log.Debug("flushing tables")
errMSGFormat := "iptables: failed cleaning %s chain %s,error: %v"
if i.ipv4Client != nil {
err = i.ipv4Client.ClearAndDeleteChain(iptablesFilterTable, iptablesRoutingForwardingChain)
if err != nil {
log.Errorf(errMSGFormat, ipv4, iptablesRoutingForwardingChain, err)
}
err = i.ipv4Client.ClearAndDeleteChain(iptablesNatTable, iptablesRoutingNatChain)
if err != nil {
log.Errorf(errMSGFormat, ipv4, iptablesRoutingNatChain, err)
}
}
if i.ipv6Client != nil {
err = i.ipv6Client.ClearAndDeleteChain(iptablesFilterTable, iptablesRoutingForwardingChain)
if err != nil {
log.Errorf(errMSGFormat, ipv6, iptablesRoutingForwardingChain, err)
}
err = i.ipv6Client.ClearAndDeleteChain(iptablesNatTable, iptablesRoutingNatChain)
if err != nil {
log.Errorf(errMSGFormat, ipv6, iptablesRoutingNatChain, err)
}
}
log.Info("done cleaning up iptables rules")
}
// RestoreOrCreateContainers restores existing iptables containers (chains and rules)
// if they don't exist, we create them
func (i *iptablesManager) RestoreOrCreateContainers() error {
i.mux.Lock()
defer i.mux.Unlock()
if i.rules[ipv4][ipv4Forwarding] != nil && i.rules[ipv6][ipv6Forwarding] != nil {
return nil
}
errMSGFormat := "iptables: failed creating %s chain %s,error: %v"
if i.ipv4Client != nil {
err := createChain(i.ipv4Client, iptablesFilterTable, iptablesRoutingForwardingChain)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv4, iptablesRoutingForwardingChain, err)
}
err = createChain(i.ipv4Client, iptablesNatTable, iptablesRoutingNatChain)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv4, iptablesRoutingNatChain, err)
}
err = i.restoreRules(i.ipv4Client)
if err != nil {
return fmt.Errorf("iptables: error while restoring ipv4 rules: %v", err)
}
}
if i.ipv6Client != nil {
err := createChain(i.ipv6Client, iptablesFilterTable, iptablesRoutingForwardingChain)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv6, iptablesRoutingForwardingChain, err)
}
err = createChain(i.ipv6Client, iptablesNatTable, iptablesRoutingNatChain)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv6, iptablesRoutingNatChain, err)
}
err = i.restoreRules(i.ipv6Client)
if err != nil {
return fmt.Errorf("iptables: error while restoring ipv6 rules: %v", err)
}
}
err := i.addJumpRules()
if err != nil {
return fmt.Errorf("iptables: error while creating jump rules: %v", err)
}
return nil
}
// addJumpRules create jump rules to send packets to NetBird chains
func (i *iptablesManager) addJumpRules() error {
err := i.cleanJumpRules()
if err != nil {
return err
}
if i.ipv4Client != nil {
rule := append(iptablesDefaultForwardingRule, ipv4Forwarding)
err = i.ipv4Client.Insert(iptablesFilterTable, iptablesForwardChain, 1, rule...)
if err != nil {
return err
}
i.rules[ipv4][ipv4Forwarding] = rule
rule = append(iptablesDefaultNatRule, ipv4Nat)
err = i.ipv4Client.Insert(iptablesNatTable, iptablesPostRoutingChain, 1, rule...)
if err != nil {
return err
}
i.rules[ipv4][ipv4Nat] = rule
}
if i.ipv6Client != nil {
rule := append(iptablesDefaultForwardingRule, ipv6Forwarding)
err = i.ipv6Client.Insert(iptablesFilterTable, iptablesForwardChain, 1, rule...)
if err != nil {
return err
}
i.rules[ipv6][ipv6Forwarding] = rule
rule = append(iptablesDefaultNatRule, ipv6Nat)
err = i.ipv6Client.Insert(iptablesNatTable, iptablesPostRoutingChain, 1, rule...)
if err != nil {
return err
}
i.rules[ipv6][ipv6Nat] = rule
}
return nil
}
// cleanJumpRules cleans jump rules that was sending packets to NetBird chains
func (i *iptablesManager) cleanJumpRules() error {
var err error
errMSGFormat := "iptables: failed cleaning rule from %s chain %s,err: %v"
rule, found := i.rules[ipv4][ipv4Forwarding]
if i.ipv4Client != nil {
if found {
log.Debugf("iptables: removing %s rule: %s ", ipv4, ipv4Forwarding)
err = i.ipv4Client.DeleteIfExists(iptablesFilterTable, iptablesForwardChain, rule...)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv4, iptablesForwardChain, err)
}
}
rule, found = i.rules[ipv4][ipv4Nat]
if found {
log.Debugf("iptables: removing %s rule: %s ", ipv4, ipv4Nat)
err = i.ipv4Client.DeleteIfExists(iptablesNatTable, iptablesPostRoutingChain, rule...)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv4, iptablesPostRoutingChain, err)
}
}
}
if i.ipv6Client == nil {
rule, found = i.rules[ipv6][ipv6Forwarding]
if found {
log.Debugf("iptables: removing %s rule: %s ", ipv6, ipv6Forwarding)
err = i.ipv6Client.DeleteIfExists(iptablesFilterTable, iptablesForwardChain, rule...)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv6, iptablesForwardChain, err)
}
}
rule, found = i.rules[ipv6][ipv6Nat]
if found {
log.Debugf("iptables: removing %s rule: %s ", ipv6, ipv6Nat)
err = i.ipv6Client.DeleteIfExists(iptablesNatTable, iptablesPostRoutingChain, rule...)
if err != nil {
return fmt.Errorf(errMSGFormat, ipv6, iptablesPostRoutingChain, err)
}
}
}
return nil
}
func iptablesProtoToString(proto iptables.Protocol) string {
if proto == iptables.ProtocolIPv6 {
return ipv6
}
return ipv4
}
// restoreRules restores existing NetBird rules
func (i *iptablesManager) restoreRules(iptablesClient *iptables.IPTables) error {
ipVersion := iptablesProtoToString(iptablesClient.Proto())
if i.rules[ipVersion] == nil {
i.rules[ipVersion] = make(map[string][]string)
}
table := iptablesFilterTable
for _, chain := range []string{iptablesForwardChain, iptablesRoutingForwardingChain} {
rules, err := iptablesClient.List(table, chain)
if err != nil {
return err
}
for _, ruleString := range rules {
rule := strings.Fields(ruleString)
id := getRuleRouteID(rule)
if id != "" {
i.rules[ipVersion][id] = rule[2:]
}
}
}
table = iptablesNatTable
for _, chain := range []string{iptablesPostRoutingChain, iptablesRoutingNatChain} {
rules, err := iptablesClient.List(table, chain)
if err != nil {
return err
}
for _, ruleString := range rules {
rule := strings.Fields(ruleString)
id := getRuleRouteID(rule)
if id != "" {
i.rules[ipVersion][id] = rule[2:]
}
}
}
return nil
}
// createChain create NetBird chains
func createChain(iptables *iptables.IPTables, table, newChain string) error {
chains, err := iptables.ListChains(table)
if err != nil {
return fmt.Errorf("couldn't get %s %s table chains, error: %v", iptablesProtoToString(iptables.Proto()), table, err)
}
shouldCreateChain := true
for _, chain := range chains {
if chain == newChain {
shouldCreateChain = false
}
}
if shouldCreateChain {
err = iptables.NewChain(table, newChain)
if err != nil {
return fmt.Errorf("couldn't create %s chain %s in %s table, error: %v", iptablesProtoToString(iptables.Proto()), newChain, table, err)
}
if table == iptablesNatTable {
err = iptables.Append(table, newChain, iptablesDefaultNetbirdNatRule...)
} else {
err = iptables.Append(table, newChain, iptablesDefaultNetbirdForwardingRule...)
}
if err != nil {
return fmt.Errorf("couldn't create %s chain %s default rule, error: %v", iptablesProtoToString(iptables.Proto()), newChain, err)
}
}
return nil
}
// genRuleSpec generates rule specification with comment identifier
func genRuleSpec(jump, id, source, destination string) []string {
return []string{"-s", source, "-d", destination, "-j", jump, "-m", "comment", "--comment", id}
}
// getRuleRouteID returns the rule ID if matches our prefix
func getRuleRouteID(rule []string) string {
for i, flag := range rule {
if flag == "--comment" {
id := rule[i+1]
if strings.HasPrefix(id, "netbird-") {
return id
}
}
}
return ""
}
// InsertRoutingRules inserts an iptables rule pair to the forwarding chain and if enabled, to the nat chain
func (i *iptablesManager) InsertRoutingRules(pair routerPair) error {
i.mux.Lock()
defer i.mux.Unlock()
err := i.insertRoutingRule(forwardingFormat, iptablesFilterTable, iptablesRoutingForwardingChain, routingFinalForwardJump, pair)
if err != nil {
return err
}
err = i.insertRoutingRule(inForwardingFormat, iptablesFilterTable, iptablesRoutingForwardingChain, routingFinalForwardJump, getInPair(pair))
if err != nil {
return err
}
if !pair.masquerade {
return nil
}
err = i.insertRoutingRule(natFormat, iptablesNatTable, iptablesRoutingNatChain, routingFinalNatJump, pair)
if err != nil {
return err
}
err = i.insertRoutingRule(inNatFormat, iptablesNatTable, iptablesRoutingNatChain, routingFinalNatJump, getInPair(pair))
if err != nil {
return err
}
return nil
}
// insertRoutingRule inserts an iptable rule
func (i *iptablesManager) insertRoutingRule(keyFormat, table, chain, jump string, pair routerPair) error {
var err error
prefix := netip.MustParsePrefix(pair.source)
ipVersion := ipv4
iptablesClient := i.ipv4Client
if prefix.Addr().Unmap().Is6() {
iptablesClient = i.ipv6Client
ipVersion = ipv6
}
if iptablesClient == nil {
return fmt.Errorf("unable to insert iptables routing rules. Iptables client is not initialized")
}
ruleKey := genKey(keyFormat, pair.ID)
rule := genRuleSpec(jump, ruleKey, pair.source, pair.destination)
existingRule, found := i.rules[ipVersion][ruleKey]
if found {
err = iptablesClient.DeleteIfExists(table, chain, existingRule...)
if err != nil {
return fmt.Errorf("iptables: error while removing existing %s rule for %s: %v", getIptablesRuleType(table), pair.destination, err)
}
delete(i.rules[ipVersion], ruleKey)
}
err = iptablesClient.Insert(table, chain, 1, rule...)
if err != nil {
return fmt.Errorf("iptables: error while adding new %s rule for %s: %v", getIptablesRuleType(table), pair.destination, err)
}
i.rules[ipVersion][ruleKey] = rule
return nil
}
// RemoveRoutingRules removes an iptables rule pair from forwarding and nat chains
func (i *iptablesManager) RemoveRoutingRules(pair routerPair) error {
i.mux.Lock()
defer i.mux.Unlock()
err := i.removeRoutingRule(forwardingFormat, iptablesFilterTable, iptablesRoutingForwardingChain, pair)
if err != nil {
return err
}
err = i.removeRoutingRule(inForwardingFormat, iptablesFilterTable, iptablesRoutingForwardingChain, getInPair(pair))
if err != nil {
return err
}
if !pair.masquerade {
return nil
}
err = i.removeRoutingRule(natFormat, iptablesNatTable, iptablesRoutingNatChain, pair)
if err != nil {
return err
}
err = i.removeRoutingRule(inNatFormat, iptablesNatTable, iptablesRoutingNatChain, getInPair(pair))
if err != nil {
return err
}
return nil
}
// removeRoutingRule removes an iptables rule
func (i *iptablesManager) removeRoutingRule(keyFormat, table, chain string, pair routerPair) error {
var err error
prefix := netip.MustParsePrefix(pair.source)
ipVersion := ipv4
iptablesClient := i.ipv4Client
if prefix.Addr().Unmap().Is6() {
iptablesClient = i.ipv6Client
ipVersion = ipv6
}
if iptablesClient == nil {
return fmt.Errorf("unable to remove iptables routing rules. Iptables client is not initialized")
}
ruleKey := genKey(keyFormat, pair.ID)
existingRule, found := i.rules[ipVersion][ruleKey]
if found {
err = iptablesClient.DeleteIfExists(table, chain, existingRule...)
if err != nil {
return fmt.Errorf("iptables: error while removing existing %s rule for %s: %v", getIptablesRuleType(table), pair.destination, err)
}
}
delete(i.rules[ipVersion], ruleKey)
return nil
}
func getIptablesRuleType(table string) string {
ruleType := "forwarding"
if table == iptablesNatTable {
ruleType = "nat"
}
return ruleType
}
func isIptablesClientAvailable(client *iptables.IPTables) bool {
_, err := client.ListChains("filter")
return err == nil
}