Host-agent DNS forwarder improvements (fixes #770)

docs/internal.md

@@ -112,3 +112,4 @@ The volume label is "cidata", as defined by [cloud-init NoCloud](https://cloudin
 - `LIMA_CIDATA_SLIRP_GATEWAY`: set to the IP address of the host on the SLIRP network. `192.168.5.2`.
 - `LIMA_CIDATA_SLIRP_DNS`: set to the IP address of the DNS on the SLIRP network. `192.168.5.3`.
 - `LIMA_CIDATA_UDP_DNS_LOCAL_PORT`: set to the udp port number of the hostagent dns server (or 0 when not enabled).
+- `LIMA_CIDATA_TCP_DNS_LOCAL_PORT`: set to the tcp port number of the hostagent dns server (or 0 when not enabled).

docs/network.md

@@ -20,9 +20,23 @@ The loopback addresses of the host is `192.168.5.2` and is accessible from the g
 
 The DNS.
 
-If `useHostResolver` in `lima.yaml` is true, then the hostagent is going to run a DNS server over udp, using the same socket number as `ssh.localPort`. This server does a local lookup using the native host resolver, so will deal correctly with VPN configurations and split-DNS setups, as well a mDNS (for this the hostagent has to be compiled with `CGO_ENABLED=1`).
+If `useHostResolver` in `lima.yaml` is true, then the hostagent is going to run a DNS server over tcp and udp - each on a separate randomly selected free port. This server does a local lookup using the native host resolver, so it will deal correctly with VPN configurations and split-DNS setups, as well as mDNS, local `/etc/hosts` etc. For this the hostagent has to be compiled with `CGO_ENABLED=1` as default Go resolver is [broken](https://github.com/golang/go/issues/12524).
 
-This udp port is then forwarded via iptables rules to `192.168.5.3:53`, overriding the DNS provided by QEMU via slirp.
+These tcp and udp ports are then forwarded via iptables rules to `192.168.5.3:53`, overriding the DNS provided by QEMU via slirp.
+
+Currently following request types are supported:
+
+- A
+- AAAA
+- CNAME
+- TXT
+- NS
+- MX
+- SRV
+
+For all other queries hostagent will redirect the query to the nameservers specified in `/etc/resolv.conf` (or, if that fails - to `8.8.8.8` and `1.1.1.1`).
+
+DNS over tcp is rarely used. It is usually only used either when user explicitly requires it, or when request+response can't fit into a single UDP packet (most likely in case of DNSSEC), or in the case of certain management operations such as domain transfers. Neither DNSSEC nor management operations are currently supported by a hostagent, but on the off chance that the response may contain an unusually long list of records - hostagent will also listen for the tcp traffic.
 
 During initial cloud-init bootstrap, `iptables` may not yet be installed. In that case the repo server is determined using the slirp DNS. After `iptables` has been installed, the forwarding rule is applied, switching over to the hostagent DNS.
 

pkg/cidata/cidata.TEMPLATE.d/boot/07-host-dns-setup.sh

@@ -6,8 +6,19 @@ if command -v iptables >/dev/null 2>&1; then
 	if [ -n "${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}" ] && [ "${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}" -ne 0 ]; then
 		# Only add the rule once
 		if ! iptables-save | grep "udp.*${LIMA_CIDATA_SLIRP_GATEWAY}:${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}"; then
+			iptables -t nat -A PREROUTING -d "${LIMA_CIDATA_SLIRP_DNS}" -p udp --dport 53 -j DNAT \
+				--to-destination "${LIMA_CIDATA_SLIRP_GATEWAY}:${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}"
 			iptables -t nat -A OUTPUT -d "${LIMA_CIDATA_SLIRP_DNS}" -p udp --dport 53 -j DNAT \
 				--to-destination "${LIMA_CIDATA_SLIRP_GATEWAY}:${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}"
 		fi
 	fi
+	if [ -n "${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}" ] && [ "${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}" -ne 0 ]; then
+		# Only add the rule once
+		if ! iptables-save | grep "tcp.*${LIMA_CIDATA_SLIRP_GATEWAY}:${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}"; then
+			iptables -t nat -A PREROUTING -d "${LIMA_CIDATA_SLIRP_DNS}" -p tcp --dport 53 -j DNAT \
+				--to-destination "${LIMA_CIDATA_SLIRP_GATEWAY}:${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}"
+			iptables -t nat -A OUTPUT -d "${LIMA_CIDATA_SLIRP_DNS}" -p tcp --dport 53 -j DNAT \
+				--to-destination "${LIMA_CIDATA_SLIRP_GATEWAY}:${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}"
+		fi
+	fi
 fi

pkg/cidata/cidata.TEMPLATE.d/boot/30-install-packages.sh

@@ -14,7 +14,7 @@ INSTALL_IPTABLES=0
 if [ "${LIMA_CIDATA_CONTAINERD_SYSTEM}" = 1 ] || [ "${LIMA_CIDATA_CONTAINERD_USER}" = 1 ]; then
 	INSTALL_IPTABLES=1
 fi
-if [ "${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}" -ne 0 ]; then
+if [ "${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}" -ne 0 ] || [ "${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}" -ne 0 ]; then
 	INSTALL_IPTABLES=1
 fi
 
@@ -96,7 +96,14 @@ elif command -v apk >/dev/null 2>&1; then
 	fi
 fi
 
+SETUP_DNS=0
 if [ -n "${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}" ] && [ "${LIMA_CIDATA_UDP_DNS_LOCAL_PORT}" -ne 0 ]; then
+	SETUP_DNS=1
+fi
+if [ -n "${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}" ] && [ "${LIMA_CIDATA_TCP_DNS_LOCAL_PORT}" -ne 0 ]; then
+	SETUP_DNS=1
+fi
+if [ "${SETUP_DNS}" = 1 ]; then
 	# Try to setup iptables rule again, in case we just installed iptables
 	"${LIMA_CIDATA_MNT}/boot/07-host-dns-setup.sh"
 fi

pkg/cidata/cidata.TEMPLATE.d/lima.env

@@ -17,3 +17,4 @@ LIMA_CIDATA_CONTAINERD_SYSTEM=
 LIMA_CIDATA_SLIRP_DNS={{.SlirpDNS}}
 LIMA_CIDATA_SLIRP_GATEWAY={{.SlirpGateway}}
 LIMA_CIDATA_UDP_DNS_LOCAL_PORT={{.UDPDNSLocalPort}}
+LIMA_CIDATA_TCP_DNS_LOCAL_PORT={{.TCPDNSLocalPort}}

pkg/cidata/cidata.go

@@ -63,7 +63,7 @@ func setupEnv(y *limayaml.LimaYAML) (map[string]string, error) {
 	return env, nil
 }
 
-func GenerateISO9660(instDir, name string, y *limayaml.LimaYAML, udpDNSLocalPort int, nerdctlArchive string) error {
+func GenerateISO9660(instDir, name string, y *limayaml.LimaYAML, udpDNSLocalPort, tcpDNSLocalPort int, nerdctlArchive string) error {
 	if err := limayaml.Validate(*y, false); err != nil {
 		return err
 	}
@@ -119,6 +119,7 @@ func GenerateISO9660(instDir, name string, y *limayaml.LimaYAML, udpDNSLocalPort
 	}
 	if *y.UseHostResolver {
 		args.UDPDNSLocalPort = udpDNSLocalPort
+		args.TCPDNSLocalPort = tcpDNSLocalPort
 		args.DNSAddresses = append(args.DNSAddresses, qemu.SlirpDNS)
 	} else if len(y.DNS) > 0 {
 		for _, addr := range y.DNS {

pkg/cidata/template.go

@@ -40,6 +40,7 @@ type TemplateArgs struct {
 	SlirpGateway    string
 	SlirpDNS        string
 	UDPDNSLocalPort int
+	TCPDNSLocalPort int
 	Env             map[string]string
 	DNSAddresses    []string
 }

pkg/hostagent/dns.go

@@ -16,6 +16,20 @@ type Handler struct {
 	clients      []*dns.Client
 }
 
+type Server struct {
+	udp *dns.Server
+	tcp *dns.Server
+}
+
+func (s *Server) Shutdown() {
+	if s.udp != nil {
+		_ = s.udp.Shutdown()
+	}
+	if s.tcp != nil {
+		_ = s.tcp.Shutdown()
+	}
+}
+
 func newStaticClientConfig(ips []net.IP) (*dns.ClientConfig, error) {
 	s := ``
 	for _, ip := range ips {
@@ -52,27 +66,111 @@ func (h *Handler) handleQuery(w dns.ResponseWriter, req *dns.Msg) {
 		handled bool
 	)
 	reply.SetReply(req)
-	for _, q := range reply.Question {
+	for _, q := range req.Question {
+		hdr := dns.RR_Header{
+			Name:   q.Name,
+			Rrtype: q.Qtype,
+			Class:  q.Qclass,
+			Ttl:    5,
+		}
 		switch q.Qtype {
-		case dns.TypeA:
+		case dns.TypeCNAME, dns.TypeA, dns.TypeAAAA:
+			cname, err := net.LookupCNAME(q.Name)
+			if err != nil {
+				break
+			}
+			if cname != "" && cname != q.Name {
+				hdr.Rrtype = dns.TypeCNAME
+				a := &dns.CNAME{
+					Hdr:    hdr,
+					Target: cname,
+				}
+				reply.Answer = append(reply.Answer, a)
+				handled = true
+			}
+			if q.Qtype == dns.TypeCNAME {
+				break
+			}
+			hdr.Name = cname
 			addrs, err := net.LookupIP(q.Name)
 			if err == nil && len(addrs) > 0 {
 				for _, ip := range addrs {
-					if ip.To4() != nil {
-						a := &dns.A{
-							Hdr: dns.RR_Header{
-								Name:   q.Name,
-								Rrtype: dns.TypeA,
-								Class:  dns.ClassINET,
-							},
-							A: ip.To4(),
+					var a dns.RR
+					ipv6 := ip.To4() == nil
+					if q.Qtype == dns.TypeA && !ipv6 {
+						hdr.Rrtype = dns.TypeA
+						a = &dns.A{
+							Hdr: hdr,
+							A:   ip.To4(),
+						}
+					} else if q.Qtype == dns.TypeAAAA && ipv6 {
+						hdr.Rrtype = dns.TypeAAAA
+						a = &dns.AAAA{
+							Hdr:  hdr,
+							AAAA: ip.To16(),
+						}
+					} else {
+						continue
+					}
+					reply.Answer = append(reply.Answer, a)
+					handled = true
+				}
+			}
+		case dns.TypeTXT:
+			txt, err := net.LookupTXT(q.Name)
+			if err == nil && len(txt) > 0 {
+				a := &dns.TXT{
+					Hdr: hdr,
+					Txt: txt,
+				}
+				reply.Answer = append(reply.Answer, a)
+				handled = true
+			}
+		case dns.TypeNS:
+			ns, err := net.LookupNS(q.Name)
+			if err == nil && len(ns) > 0 {
+				for _, s := range ns {
+					if s.Host != "" {
+						a := &dns.NS{
+							Hdr: hdr,
+							Ns:  s.Host,
+						}
+						reply.Answer = append(reply.Answer, a)
+						handled = true
+					}
+				}
+			}
+		case dns.TypeMX:
+			mx, err := net.LookupMX(q.Name)
+			if err == nil && len(mx) > 0 {
+				for _, s := range mx {
+					if s.Host != "" {
+						a := &dns.MX{
+							Hdr:        hdr,
+							Mx:         s.Host,
+							Preference: s.Pref,
 						}
 						reply.Answer = append(reply.Answer, a)
 						handled = true
-						break
 					}
 				}
 			}
+		case dns.TypeSRV:
+			_, addrs, err := net.LookupSRV("", "", q.Name)
+			if err == nil {
+				hdr.Rrtype = dns.TypeSRV
+				for _, addr := range addrs {
+					a := &dns.SRV{
+						Hdr:      hdr,
+						Target:   addr.Target,
+						Port:     addr.Port,
+						Priority: addr.Priority,
+						Weight:   addr.Weight,
+					}
+					reply.Answer = append(reply.Answer, a)
+					handled = true
+				}
+			}
 		}
 	}
 	if handled {
@@ -107,17 +205,31 @@ func (h *Handler) ServeDNS(w dns.ResponseWriter, req *dns.Msg) {
 	}
 }
 
-func (a *HostAgent) StartDNS() (*dns.Server, error) {
+func (a *HostAgent) StartDNS() (*Server, error) {
 	h, err := newHandler()
 	if err != nil {
 		panic(err)
 	}
-	addr := fmt.Sprintf("127.0.0.1:%d", a.udpDNSLocalPort)
-	server := &dns.Server{Net: "udp", Addr: addr, Handler: h}
-	go func() {
-		if e := server.ListenAndServe(); e != nil {
-			panic(e)
-		}
-	}()
+	server := &Server{}
+	if a.udpDNSLocalPort > 0 {
+		addr := fmt.Sprintf("127.0.0.1:%d", a.udpDNSLocalPort)
+		s := &dns.Server{Net: "udp", Addr: addr, Handler: h}
+		server.udp = s
+		go func() {
+			if e := s.ListenAndServe(); e != nil {
+				panic(e)
+			}
+		}()
+	}
+	if a.tcpDNSLocalPort > 0 {
+		addr := fmt.Sprintf("127.0.0.1:%d", a.tcpDNSLocalPort)
+		s := &dns.Server{Net: "tcp", Addr: addr, Handler: h}
+		server.tcp = s
+		go func() {
+			if e := s.ListenAndServe(); e != nil {
+				panic(e)
+			}
+		}()
+	}
 	return server, nil
 }

pkg/hostagent/hostagent.go

@@ -35,6 +35,7 @@ type HostAgent struct {
 	y               *limayaml.LimaYAML
 	sshLocalPort    int
 	udpDNSLocalPort int
+	tcpDNSLocalPort int
 	instDir         string
 	sshConfig       *ssh.SSHConfig
 	portForwarder   *portForwarder
@@ -87,15 +88,19 @@ func New(instName string, stdout io.Writer, sigintCh chan os.Signal, opts ...Opt
 		return nil, err
 	}
 
-	var udpDNSLocalPort int
+	var udpDNSLocalPort, tcpDNSLocalPort int
 	if *y.UseHostResolver {
 		udpDNSLocalPort, err = findFreeUDPLocalPort()
 		if err != nil {
 			return nil, err
 		}
+		tcpDNSLocalPort, err = findFreeTCPLocalPort()
+		if err != nil {
+			return nil, err
+		}
 	}
 
-	if err := cidata.GenerateISO9660(inst.Dir, instName, y, udpDNSLocalPort, o.nerdctlArchive); err != nil {
+	if err := cidata.GenerateISO9660(inst.Dir, instName, y, udpDNSLocalPort, tcpDNSLocalPort, o.nerdctlArchive); err != nil {
 		return nil, err
 	}
 
@@ -135,6 +140,7 @@ func New(instName string, stdout io.Writer, sigintCh chan os.Signal, opts ...Opt
 		y:               y,
 		sshLocalPort:    sshLocalPort,
 		udpDNSLocalPort: udpDNSLocalPort,
+		tcpDNSLocalPort: tcpDNSLocalPort,
 		instDir:         inst.Dir,
 		sshConfig:       sshConfig,
 		portForwarder:   newPortForwarder(sshConfig, sshLocalPort, rules),
@@ -244,7 +250,7 @@ func (a *HostAgent) Run(ctx context.Context) error {
 		if err != nil {
 			return fmt.Errorf("cannot start DNS server: %w", err)
 		}
-		defer func() { _ = dnsServer.Shutdown() }()
+		defer dnsServer.Shutdown()
 	}
 
 	qCmd := exec.CommandContext(ctx, a.qExe, a.qArgs...)