Merge remote-tracking branch 'origin/main' into refactor/permissions-manager

This commit is contained in:
pascal
2026-09-11 13:55:02 +02:00
2126 changed files with 284567 additions and 42026 deletions
+8 -2
View File
@@ -146,7 +146,11 @@ func (c *ClaimsExtractor) ToGroups(token *jwt.Token, claimName string) []string
userJWTGroups := make([]string, 0)
if claim, ok := claims[claimName]; ok {
if claimGroups, ok := claim.([]interface{}); ok {
switch claimGroups := claim.(type) {
case string:
// Some IdPs emit a single group claim as a string instead of an array.
userJWTGroups = append(userJWTGroups, claimGroups)
case []any:
for _, g := range claimGroups {
if group, ok := g.(string); ok {
userJWTGroups = append(userJWTGroups, group)
@@ -154,9 +158,11 @@ func (c *ClaimsExtractor) ToGroups(token *jwt.Token, claimName string) []string
log.Debugf("JWT claim %q contains a non-string group (type: %T): %v", claimName, g, g)
}
}
default:
log.Debugf("JWT claim %q is not a string or string array (type: %T): %v", claimName, claim, claim)
}
} else {
log.Debugf("JWT claim %q is not a string array", claimName)
log.Debugf("JWT claim %q is missing", claimName)
}
return userJWTGroups
+9
View File
@@ -249,6 +249,15 @@ func TestClaimsExtractor_ToGroups(t *testing.T) {
groupClaimName: "groups",
expectedGroups: []string{},
},
{
name: "extracts single group string from claim",
claims: jwt.MapClaims{
"sub": "user-123",
"groups": "admin",
},
groupClaimName: "groups",
expectedGroups: []string{"admin"},
},
{
name: "handles custom claim name",
claims: jwt.MapClaims{
+47 -19
View File
@@ -289,36 +289,64 @@ func getPublicKey(token *jwt.Token, jwks *Jwks) (interface{}, error) {
return nil, errKeyNotFound
}
func getPublicKeyFromECDSA(jwk JSONWebKey) (publicKey *ecdsa.PublicKey, err error) {
func curveFromName(crv string) (elliptic.Curve, error) {
switch crv {
case p256:
return elliptic.P256(), nil
case p384:
return elliptic.P384(), nil
case p521:
return elliptic.P521(), nil
default:
return nil, fmt.Errorf("unsupported elliptic curve %q", crv)
}
}
func getPublicKeyFromECDSA(jwk JSONWebKey) (*ecdsa.PublicKey, error) {
if jwk.X == "" || jwk.Y == "" || jwk.Crv == "" {
return nil, fmt.Errorf("ecdsa key incomplete")
}
var xCoordinate []byte
if xCoordinate, err = base64.RawURLEncoding.DecodeString(jwk.X); err != nil {
curve, err := curveFromName(jwk.Crv)
if err != nil {
return nil, err
}
var yCoordinate []byte
if yCoordinate, err = base64.RawURLEncoding.DecodeString(jwk.Y); err != nil {
return nil, err
xCoordinate, err := base64.RawURLEncoding.DecodeString(jwk.X)
if err != nil {
return nil, fmt.Errorf("decode ecdsa x coordinate: %w", err)
}
publicKey = &ecdsa.PublicKey{}
var curve elliptic.Curve
switch jwk.Crv {
case p256:
curve = elliptic.P256()
case p384:
curve = elliptic.P384()
case p521:
curve = elliptic.P521()
yCoordinate, err := base64.RawURLEncoding.DecodeString(jwk.Y)
if err != nil {
return nil, fmt.Errorf("decode ecdsa y coordinate: %w", err)
}
publicKey.Curve = curve
publicKey.X = big.NewInt(0).SetBytes(xCoordinate)
publicKey.Y = big.NewInt(0).SetBytes(yCoordinate)
var x, y big.Int
x.SetBytes(xCoordinate)
y.SetBytes(yCoordinate)
bits := curve.Params().BitSize
if x.BitLen() > bits {
return nil, fmt.Errorf("ecdsa x coordinate is %d bits, exceeds curve %s field size of %d bits", x.BitLen(), jwk.Crv, bits)
}
if y.BitLen() > bits {
return nil, fmt.Errorf("ecdsa y coordinate is %d bits, exceeds curve %s field size of %d bits", y.BitLen(), jwk.Crv, bits)
}
// Round up: P-521's field is 521 bits, so a coordinate needs 66 bytes, not 65.
size := (bits + 7) / 8
// Assemble the SEC 1 uncompressed point (0x04 || X || Y)
point := make([]byte, 1+2*size)
point[0] = 4
x.FillBytes(point[1 : 1+size])
y.FillBytes(point[1+size:])
publicKey, err := ecdsa.ParseUncompressedPublicKey(curve, point)
if err != nil {
return nil, fmt.Errorf("parse ecdsa public key: %w", err)
}
return publicKey, nil
}
+214
View File
@@ -0,0 +1,214 @@
package jwt
import (
"bytes"
"context"
"crypto/ecdsa"
"crypto/elliptic"
"crypto/rand"
"encoding/base64"
"encoding/json"
"net/http"
"net/http/httptest"
"testing"
"time"
"github.com/golang-jwt/jwt/v5"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// ecdsaJWK builds a JWK for pub using uncompressed-point encoding
func ecdsaJWK(t *testing.T, kid string, pub *ecdsa.PublicKey, crv string, size int) JSONWebKey {
t.Helper()
point, err := pub.Bytes()
require.NoError(t, err)
require.Len(t, point, 1+2*size)
require.Equal(t, byte(4), point[0], "expected uncompressed point")
return JSONWebKey{
Kty: "EC",
Kid: kid,
Use: "sig",
Crv: crv,
X: base64.RawURLEncoding.EncodeToString(point[1 : 1+size]),
Y: base64.RawURLEncoding.EncodeToString(point[1+size:]),
}
}
func TestGetPublicKeyFromECDSA_RoundTrip(t *testing.T) {
tests := []struct {
crv string
curve elliptic.Curve
size int
}{
{p256, elliptic.P256(), 32},
{p384, elliptic.P384(), 48},
{p521, elliptic.P521(), 66},
}
for _, tc := range tests {
t.Run(tc.crv, func(t *testing.T) {
priv, err := ecdsa.GenerateKey(tc.curve, rand.Reader)
require.NoError(t, err)
got, err := getPublicKeyFromECDSA(ecdsaJWK(t, "kid", &priv.PublicKey, tc.crv, tc.size))
require.NoError(t, err)
assert.True(t, priv.PublicKey.Equal(got), "parsed key differs from the original")
})
}
}
// TestGetPublicKeyFromECDSA_ShortCoordinate covers IdPs that strip leading zero
// bytes from a coordinate instead of padding to the curve's field size.
func TestGetPublicKeyFromECDSA_ShortCoordinate(t *testing.T) {
var (
priv *ecdsa.PrivateKey
point []byte
)
for i := 0; i < 10000; i++ {
key, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
require.NoError(t, err)
p, err := key.PublicKey.Bytes()
require.NoError(t, err)
if p[1] == 0 || p[33] == 0 {
priv, point = key, p
break
}
}
require.NotNil(t, priv, "no key with a leading zero coordinate byte was generated")
jwk := JSONWebKey{
Kty: "EC",
Kid: "kid",
Crv: p256,
X: base64.RawURLEncoding.EncodeToString(bytes.TrimLeft(point[1:33], "\x00")),
Y: base64.RawURLEncoding.EncodeToString(bytes.TrimLeft(point[33:], "\x00")),
}
got, err := getPublicKeyFromECDSA(jwk)
require.NoError(t, err)
assert.True(t, priv.PublicKey.Equal(got))
}
func TestGetPublicKeyFromECDSA_Invalid(t *testing.T) {
priv, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
require.NoError(t, err)
valid := ecdsaJWK(t, "kid", &priv.PublicKey, p256, 32)
offCurve := valid
x, err := base64.RawURLEncoding.DecodeString(valid.X)
require.NoError(t, err)
x[31] ^= 0xff
offCurve.X = base64.RawURLEncoding.EncodeToString(x)
// 33 non-zero bytes is 264 bits, past P-256's 256-bit field.
oversized := valid
oversized.X = base64.RawURLEncoding.EncodeToString(bytes.Repeat([]byte{0xff}, 33))
// P-521 coordinates occupy 66 bytes but only 521 bits, so a full 66-byte
// 0xff value (528 bits) is over the field size without being over the byte
// length. Only a bit-length bound catches this.
overP521 := JSONWebKey{
Kty: "EC",
Crv: p521,
X: base64.RawURLEncoding.EncodeToString(bytes.Repeat([]byte{0xff}, 66)),
Y: base64.RawURLEncoding.EncodeToString(bytes.Repeat([]byte{0xff}, 66)),
}
zeroPoint := valid
zeroPoint.X = base64.RawURLEncoding.EncodeToString(make([]byte, 32))
zeroPoint.Y = base64.RawURLEncoding.EncodeToString(make([]byte, 32))
tests := []struct {
name string
jwk JSONWebKey
errContains string
}{
{name: "missing crv", jwk: JSONWebKey{Kty: "EC", X: valid.X, Y: valid.Y}},
{name: "missing x", jwk: JSONWebKey{Kty: "EC", Crv: p256, Y: valid.Y}},
{name: "unsupported curve", jwk: JSONWebKey{Kty: "EC", Crv: "P-224", X: valid.X, Y: valid.Y}, errContains: "unsupported elliptic curve"},
{name: "undecodable x", jwk: JSONWebKey{Kty: "EC", Crv: p256, X: "!!not base64!!!", Y: valid.Y}, errContains: "decode ecdsa x coordinate"},
{name: "coordinate over field size", jwk: oversized, errContains: "exceeds curve P-256 field size of 256 bits"},
{name: "p521 coordinate over field size", jwk: overP521, errContains: "exceeds curve P-521 field size of 521 bits"},
{name: "off-curve point", jwk: offCurve, errContains: "parse ecdsa public key"},
{name: "point at infinity", jwk: zeroPoint, errContains: "parse ecdsa public key"},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
key, err := getPublicKeyFromECDSA(tc.jwk)
require.Error(t, err)
assert.Nil(t, key)
if tc.errContains != "" {
assert.ErrorContains(t, err, tc.errContains)
}
})
}
}
// TestValidateAndParse_ECDSA verifies an ES256-signed token end to end, proving
// the parsed key actually validates signatures.
func TestValidateAndParse_ECDSA(t *testing.T) {
const (
kid = "es256-kid"
issuer = "https://issuer.example.com/"
audience = "netbird"
)
priv, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
require.NoError(t, err)
jwks, err := json.Marshal(Jwks{Keys: []JSONWebKey{ecdsaJWK(t, kid, &priv.PublicKey, p256, 32)}})
require.NoError(t, err)
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
w.Header().Set("Content-Type", "application/json")
_, _ = w.Write(jwks)
}))
defer srv.Close()
token := jwt.NewWithClaims(jwt.SigningMethodES256, jwt.MapClaims{
"iss": issuer,
"aud": audience,
"sub": "user-1",
"iat": time.Now().Add(-time.Minute).Unix(),
"exp": time.Now().Add(time.Hour).Unix(),
})
token.Header["kid"] = kid
signed, err := token.SignedString(priv)
require.NoError(t, err)
v := NewValidator(issuer, []string{audience}, srv.URL, false)
parsed, err := v.ValidateAndParse(context.Background(), signed)
require.NoError(t, err)
require.True(t, parsed.Valid)
claims, ok := parsed.Claims.(jwt.MapClaims)
require.True(t, ok)
assert.Equal(t, "user-1", claims["sub"])
// A token signed by a different key of the same curve must be rejected.
other, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
require.NoError(t, err)
forged := jwt.NewWithClaims(jwt.SigningMethodES256, jwt.MapClaims{
"iss": issuer,
"aud": audience,
"sub": "user-1",
"iat": time.Now().Add(-time.Minute).Unix(),
"exp": time.Now().Add(time.Hour).Unix(),
})
forged.Header["kid"] = kid
forgedSigned, err := forged.SignedString(other)
require.NoError(t, err)
_, err = v.ValidateAndParse(context.Background(), forgedSigned)
require.Error(t, err)
}
+2
View File
@@ -3,6 +3,8 @@ package context
const (
RequestIDKey = "requestID"
AccountIDKey = "accountID"
RoleKey = "role"
UserIDKey = "userID"
PeerIDKey = "peerID"
UserAgentKey = "userAgent"
)
+153
View File
@@ -0,0 +1,153 @@
// Package llm holds LLM model-identifier helpers shared by the proxy and
// the management server. The proxy normalizes model ids parsed off inbound
// requests; management normalizes the operator's registered model ids at
// synthesis time so both sides of the pricing / routing contract compare
// equal.
package llm
import (
"regexp"
"strings"
)
// bedrockVendorNamespaces are the vendor segments a Bedrock model id is
// published under. They identify the geography in front of a cross-region
// inference profile without knowing the geography: in
// "eu.anthropic.claude-...", what makes "eu" a geography is that "anthropic"
// follows it.
//
// A vendor missing from here is not fatal — bedrockGeographies covers the
// same id from the other side — but it is one of the two ways an id can go
// unrecognised, and the list needs a new entry whenever AWS onboards a
// vendor. A live listing found "global.xai.grok-4.6" days after this was
// first written.
var bedrockVendorNamespaces = map[string]struct{}{
"ai21": {},
"amazon": {},
"anthropic": {},
"cohere": {},
"deepseek": {},
"luma": {},
"meta": {},
"mistral": {},
"openai": {},
"qwen": {},
"stability": {},
"twelvelabs": {},
"writer": {},
"xai": {},
}
// bedrockGeographies are the geography segments AWS issues cross-region
// inference profiles under. They recognise a profile whose vendor we have
// never seen, which is the case bedrockVendorNamespaces alone gets wrong:
// "global.xai.grok-4.6" is a geography and a model whether or not "xai" is
// a name we know.
//
// Neither list is sufficient alone. A geography list on its own is what this
// file started with, and it aged badly — it held us, eu, apac and global, so
// every profile issued under jp, au, ca, sa or us-gov carried its prefix into
// the pricing key, matched no catalog entry, and reported the model unpriced.
// A vendor list on its own misses a new vendor under a known geography.
// Together, an id has to be new on both axes at once to go unrecognised.
var bedrockGeographies = map[string]struct{}{
"apac": {},
"au": {},
"ca": {},
"eu": {},
"global": {},
"jp": {},
"sa": {},
"us": {},
"us-gov": {},
}
// stripBedrockGeography removes the cross-region inference-profile geography
// from a Bedrock model id, leaving the "<vendor>.<model>" form the catalog and
// the pricing table key on.
//
// A leading segment counts as a geography when it is one we know, or when a
// known vendor follows it. Either alone is enough: the id has to be new on
// both axes before its geography survives.
//
// The segment has to be followed by two more, so "amazon.nova-pro" stays a
// vendor and a model rather than becoming a geography and a model — cutting
// its first segment would strip the vendor away. Over-stripping is the
// dangerous direction, because the result also decides which route may claim
// a model.
func stripBedrockGeography(modelID string) string {
geo, rest, found := strings.Cut(modelID, ".")
if !found || geo == "" {
return modelID
}
vendor, _, found := strings.Cut(rest, ".")
if !found {
return modelID
}
if _, ok := bedrockGeographies[geo]; ok {
return rest
}
if _, ok := bedrockVendorNamespaces[vendor]; ok {
return rest
}
return modelID
}
// bedrockVersionSuffix matches the trailing "-vN[:N]" or "-YYYYMMDD-vN[:N]"
// version/throughput suffix of a Bedrock model id.
var bedrockVersionSuffix = regexp.MustCompile(`-(\d{8}-)?v\d+(:\d+)?$`)
// NormalizeBedrockModel strips an ARN wrapper, a cross-region inference-profile
// prefix, and the version/throughput suffix from a Bedrock model id so it
// matches the catalog/pricing key, e.g.
// "eu.anthropic.claude-sonnet-4-5-20250929-v1:0" -> "anthropic.claude-sonnet-4-5"
// and the inference-profile ARN's last segment likewise. It is the single
// source of truth shared by the proxy's request parser (which normalizes the
// request model from the URL path), the proxy's router (which normalizes the
// operator's registered Bedrock model ids so both sides compare equal), and
// the management synthesizer (which keys per-provider pricing entries by the
// normalized id the parser will emit at billing time).
func NormalizeBedrockModel(modelID string) string {
m := modelID
// A full ARN (inference-profile / provisioned-throughput / foundation-model)
// carries the model id in its last path segment.
if strings.HasPrefix(m, "arn:") {
if i := strings.LastIndex(m, "/"); i >= 0 {
m = m[i+1:]
}
}
m = stripBedrockGeography(m)
return bedrockVersionSuffix.ReplaceAllString(m, "")
}
// anthropicDatedModel matches a Claude model id carrying the trailing
// "-YYYYMMDD" release-date suffix Anthropic appends to a pinned release,
// capturing the id without it. The "claude" anchor is load-bearing: pricing
// looks every model up through this helper regardless of surface, and an
// operator may register a custom id with any shape at all, so an unanchored
// "-\d{8}$" would let "internal-llm-20250101" silently inherit the rate
// registered for "internal-llm". The anchor also covers the vendor-prefixed
// forms ("anthropic.claude-...", "us.anthropic.claude-...").
var anthropicDatedModel = regexp.MustCompile(`(?i)^(.*claude.*)-\d{8}$`)
// NormalizeAnthropicModel strips the trailing release-date suffix from a
// Claude model id, e.g. "claude-sonnet-4-5-20250929" -> "claude-sonnet-4-5",
// so a dated id a client pins matches the undated one the operator
// registered. Ids that are not Claude-family are returned untouched.
// Callers try the verbatim id first and fall back to this, so two dated
// releases of the same family stay distinct wherever both are registered
// explicitly.
func NormalizeAnthropicModel(modelID string) string {
return anthropicDatedModel.ReplaceAllString(modelID, "$1")
}
// NormalizeVertexModel strips the "@version" suffix from a Vertex AI model id
// (e.g. "claude-sonnet-4-5@20250929" -> "claude-sonnet-4-5") so it matches
// the catalog/pricing key. Vertex publisher models are priced under their
// vendor surface with the bare, unversioned id.
func NormalizeVertexModel(modelID string) string {
if at := strings.Index(modelID, "@"); at >= 0 {
return modelID[:at]
}
return modelID
}
+120
View File
@@ -0,0 +1,120 @@
package llm
import (
"testing"
"github.com/stretchr/testify/require"
)
func TestNormalizeBedrockModel(t *testing.T) {
cases := map[string]string{
"eu.anthropic.claude-sonnet-4-5-20250929-v1:0": "anthropic.claude-sonnet-4-5",
"us.anthropic.claude-haiku-4-5": "anthropic.claude-haiku-4-5",
"us.anthropic.claude-opus-4-8-20250101-v1:0": "anthropic.claude-opus-4-8",
"apac.anthropic.claude-haiku-4-5-v1:0": "anthropic.claude-haiku-4-5",
"amazon.nova-2-lite-v1:0": "amazon.nova-2-lite",
"anthropic.claude-sonnet-4-5-20250929-v1:0": "anthropic.claude-sonnet-4-5",
"meta.llama3-3-70b-instruct-v1:0": "meta.llama3-3-70b-instruct",
"amazon.nova-pro-v1:0": "amazon.nova-pro",
// Inference-profile ARN — model id lives in the last path segment.
"arn:aws:bedrock:eu-central-1:123456789012:inference-profile/eu.anthropic.claude-sonnet-4-5-20250929-v1:0": "anthropic.claude-sonnet-4-5",
}
for in, want := range cases {
require.Equal(t, want, NormalizeBedrockModel(in), "normalize %q", in)
}
}
func TestNormalizeVertexModel(t *testing.T) {
cases := map[string]string{
"claude-sonnet-4-5@20250929": "claude-sonnet-4-5",
"claude-haiku-4-5": "claude-haiku-4-5",
"gpt-4o@2024-08-06": "gpt-4o",
}
for in, want := range cases {
require.Equal(t, want, NormalizeVertexModel(in), "normalize %q", in)
}
}
func TestNormalizeAnthropicModel(t *testing.T) {
cases := map[string]string{
"claude-sonnet-4-5-20250929": "claude-sonnet-4-5",
"claude-3-5-haiku-20241022": "claude-3-5-haiku",
"claude-sonnet-5": "claude-sonnet-5",
"claude-opus-4-8": "claude-opus-4-8",
"anthropic.claude-haiku-4-5": "anthropic.claude-haiku-4-5",
"anthropic.claude-sonnet-4-5-20250929": "anthropic.claude-sonnet-4-5",
"us.anthropic.claude-opus-4-8-20250101": "us.anthropic.claude-opus-4-8",
// Non-Claude ids must survive untouched even when they end in eight
// consecutive digits: an operator can register a custom model under
// any id, and pricing looks every one of them up through this helper.
"gpt-4o": "gpt-4o",
"gpt-4o-2024-08-06": "gpt-4o-2024-08-06",
"gpt-4o-20240806": "gpt-4o-20240806",
"internal-llm-20250101": "internal-llm-20250101",
"deepseek-r1-20250120": "deepseek-r1-20250120",
"Qwen/Qwen2.5-20250101": "Qwen/Qwen2.5-20250101",
"gemini-2-5-pro-20250101": "gemini-2-5-pro-20250101",
"": "",
}
for in, want := range cases {
require.Equal(t, want, NormalizeAnthropicModel(in), "normalize %q", in)
}
}
// TestNormalizeBedrockModel_GeographiesBeyondTheOriginalFour covers the bug
// that made this vendor-anchored: the geography used to be matched against a
// list of four, so a profile issued anywhere else kept its prefix, missed the
// catalog key it was supposed to match, and reported the model unpriced.
func TestNormalizeBedrockModel_GeographiesBeyondTheOriginalFour(t *testing.T) {
for _, geo := range []string{"us", "eu", "apac", "global", "jp", "au", "ca", "sa", "us-gov", "il", "mx"} {
t.Run(geo, func(t *testing.T) {
got := NormalizeBedrockModel(geo + ".anthropic.claude-sonnet-5-20260514-v1:0")
require.Equal(t, "anthropic.claude-sonnet-5", got,
"a cross-region profile must reduce to the catalog key whatever geography issued it")
})
}
}
// TestNormalizeBedrockModel_KeepsAVendorItCannotMistakeForAGeography pins the
// direction that must never break: a plain "<vendor>.<model>" id has no
// geography, and cutting its first segment would strip the vendor away and
// hand the id to whichever route claims the bare model name.
func TestNormalizeBedrockModel_KeepsAVendorItCannotMistakeForAGeography(t *testing.T) {
cases := map[string]string{
"amazon.nova-pro-v1:0": "amazon.nova-pro",
"anthropic.claude-sonnet-5-v1:0": "anthropic.claude-sonnet-5",
"meta.llama3-3-70b-instruct-v1:0": "meta.llama3-3-70b-instruct",
"cohere.command-r-plus-v1:0": "cohere.command-r-plus",
// Unknown on both axes: neither the leading segment nor the one
// after it is a name we hold, so the id is left exactly as it came.
"xx.unknownvendor.some-model-v1:0": "xx.unknownvendor.some-model",
"Qwen/Qwen2.5-0.5B-Instruct": "Qwen/Qwen2.5-0.5B-Instruct",
}
for in, want := range cases {
t.Run(in, func(t *testing.T) {
require.Equal(t, want, NormalizeBedrockModel(in))
})
}
}
// TestNormalizeBedrockModel_RecognisesAnIdNewOnOneAxis covers what a live
// eu-central-1 listing returned days after the vendor list was written:
// "global.xai.grok-4.6", a vendor the list did not hold. Anchoring only on the
// vendor left the geography in the key, so the id matched no catalog entry and
// the model metered at zero. Each id below is unfamiliar on one axis and
// recognised through the other.
func TestNormalizeBedrockModel_RecognisesAnIdNewOnOneAxis(t *testing.T) {
cases := map[string]string{
// Known geography, vendor we had never seen (the live case).
"global.xai.grok-4.6": "xai.grok-4.6",
"eu.xai.grok-4.6": "xai.grok-4.6",
// Known vendor, geography outside the list.
"il.anthropic.claude-sonnet-5-20260514-v1:0": "anthropic.claude-sonnet-5",
"mx.amazon.nova-2-lite-v1:0": "amazon.nova-2-lite",
}
for in, want := range cases {
t.Run(in, func(t *testing.T) {
require.Equal(t, want, NormalizeBedrockModel(in))
})
}
}
+5 -2
View File
@@ -12,13 +12,16 @@ import (
// Client is the interface for the management service client.
type Client interface {
io.Closer
Sync(ctx context.Context, sysInfo *system.Info, msgHandler func(msg *proto.SyncResponse) error) error
Sync(ctx context.Context, getInfo func(ctx context.Context) *system.Info, msgHandler func(msg *proto.SyncResponse) error) error
Job(ctx context.Context, msgHandler func(msg *proto.JobRequest) *proto.JobResponse) error
Register(setupKey string, jwtToken string, sysInfo *system.Info, sshKey []byte, dnsLabels domain.List) (*proto.LoginResponse, error)
Login(sysInfo *system.Info, sshKey []byte, dnsLabels domain.List) (*proto.LoginResponse, error)
// ExtendAuthSession refreshes the peer's SSO session deadline using a fresh JWT.
// Returns the new absolute deadline; zero time when the server reports the peer
// is not eligible for session extension.
ExtendAuthSession(sysInfo *system.Info, jwtToken string) (*proto.ExtendAuthSessionResponse, error)
GetDeviceAuthorizationFlow() (*proto.DeviceAuthorizationFlow, error)
GetPKCEAuthorizationFlow() (*proto.PKCEAuthorizationFlow, error)
GetNetworkMap(sysInfo *system.Info) (*proto.NetworkMap, error)
GetServerURL() string
// IsHealthy returns the current connection status without blocking.
// Used by the engine to monitor connectivity in the background.
+146 -15
View File
@@ -2,24 +2,26 @@ package client
import (
"context"
"fmt"
"net"
"os"
"sync"
"sync/atomic"
"testing"
"time"
"github.com/golang/mock/gomock"
log "github.com/sirupsen/logrus"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"go.uber.org/mock/gomock"
"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
"google.golang.org/grpc"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/status"
"github.com/netbirdio/management-integrations/integrations"
ephemeral_manager "github.com/netbirdio/netbird/management/internals/modules/peers/ephemeral/manager"
"github.com/netbirdio/netbird/management/server/integrations/integrated_validator/validator"
"github.com/netbirdio/netbird/management/internals/modules/permissions"
"github.com/netbirdio/netbird/management/internals/controllers/network_map/controller"
@@ -90,7 +92,7 @@ func startManagement(t *testing.T) (*grpc.Server, net.Listener) {
gomock.Any(),
gomock.Any(),
).
Return(true, nil).
Return(true, context.Background(), nil).
AnyTimes()
peersManger := peers.NewManager(store)
@@ -104,7 +106,7 @@ func startManagement(t *testing.T) (*grpc.Server, net.Listener) {
t.Fatal(err)
}
ia, _ := integrations.NewIntegratedValidator(ctx, peersManger, settingsManagerMock, eventStore, cacheStore)
ia, _ := validator.NewIntegratedValidator(ctx, peersManger, settingsManagerMock, eventStore, cacheStore)
metrics, err := telemetry.NewDefaultAppMetrics(ctx)
require.NoError(t, err)
@@ -127,7 +129,7 @@ func startManagement(t *testing.T) (*grpc.Server, net.Listener) {
updateManager := update_channel.NewPeersUpdateManager(metrics)
requestBuffer := mgmt.NewAccountRequestBuffer(ctx, store)
networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), ephemeral_manager.NewEphemeralManager(store, peersManger), config)
networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), ephemeral_manager.NewEphemeralManager(store, peersManger), config, nil)
accountManager, err := mgmt.BuildManager(context.Background(), config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, ia, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManagerMock, false, cacheStore)
if err != nil {
t.Fatal(err)
@@ -139,7 +141,7 @@ func startManagement(t *testing.T) (*grpc.Server, net.Listener) {
if err != nil {
t.Fatal(err)
}
mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, mgmt.MockIntegratedValidator{}, networkMapController, nil)
mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, mgmt.MockIntegratedValidator{}, networkMapController, nil, nil)
if err != nil {
t.Fatal(err)
}
@@ -306,7 +308,7 @@ func TestClient_Sync(t *testing.T) {
defer cancel()
go func() {
err = client.Sync(ctx, info, func(msg *mgmtProto.SyncResponse) error {
err = client.Sync(ctx, func(context.Context) *system.Info { return info }, func(msg *mgmtProto.SyncResponse) error {
ch <- msg
return nil
})
@@ -317,27 +319,156 @@ func TestClient_Sync(t *testing.T) {
select {
case resp := <-ch:
if resp.GetPeerConfig() == nil {
if resp.GetPeerConfig() == nil && resp.GetNetworkMap().GetPeerConfig() == nil {
t.Error("expecting non nil PeerConfig got nil")
}
if resp.GetNetbirdConfig() == nil {
t.Error("expecting non nil NetbirdConfig got nil")
}
if len(resp.GetRemotePeers()) != 1 {
t.Errorf("expecting RemotePeers size %d got %d", 1, len(resp.GetRemotePeers()))
// Top-level RemotePeers is deprecated and must stay empty for
// v0.29.3+ (and dev) clients — the field rides inside NetworkMap
// (legacy) or the NetworkMapEnvelope (components) instead.
if len(resp.GetRemotePeers()) != 0 {
t.Error("expecting top-level RemotePeers to be empty for v0.29.3+ clients")
}
// Component-capable clients receive a NetworkMapEnvelope; the
// remote-peers list is encoded inside it. Decode it and check the
// envelope's peers slice. Legacy peers populate NetworkMap.RemotePeers;
// both shapes must surface exactly one remote peer.
remotePeerKeys := remotePeerKeysFromSync(resp, testKey.PublicKey().String())
if len(remotePeerKeys) != 1 {
t.Errorf("expecting RemotePeers size %d got %d", 1, len(remotePeerKeys))
return
}
if resp.GetRemotePeersIsEmpty() == true {
if resp.GetNetworkMap() != nil && resp.GetNetworkMap().GetRemotePeersIsEmpty() {
t.Error("expecting RemotePeers property to be false, got true")
}
if resp.GetRemotePeers()[0].GetWgPubKey() != remoteKey.PublicKey().String() {
t.Errorf("expecting RemotePeer public key %s got %s", remoteKey.PublicKey().String(), resp.GetRemotePeers()[0].GetWgPubKey())
if remotePeerKeys[0] != remoteKey.PublicKey().String() {
t.Errorf("expecting RemotePeer public key %s got %s", remoteKey.PublicKey().String(), remotePeerKeys[0])
}
case <-time.After(3 * time.Second):
t.Error("timeout waiting for test to finish")
}
}
// remotePeerKeysFromSync extracts the remote-peer WG keys from either the
// legacy NetworkMap.RemotePeers list or the components NetworkMapEnvelope's
// inner peers slice (filtering out the local receiving peer identified by
// localKey, since the envelope's peers list is index-addressed and includes
// the local peer alongside remotes).
func remotePeerKeysFromSync(resp *mgmtProto.SyncResponse, localKey string) []string {
if rp := resp.GetRemotePeers(); len(rp) > 0 {
out := make([]string, 0, len(rp))
for _, p := range rp {
out = append(out, p.GetWgPubKey())
}
return out
}
if rp := resp.GetNetworkMap().GetRemotePeers(); len(rp) > 0 {
out := make([]string, 0, len(rp))
for _, p := range rp {
out = append(out, p.GetWgPubKey())
}
return out
}
env := resp.GetNetworkMapEnvelope().GetFull()
if env == nil {
return nil
}
out := make([]string, 0, len(env.GetPeers()))
for _, p := range env.GetPeers() {
key := wgKeyFromBytes(p.GetWgPubKey())
if key == "" || key == localKey {
continue
}
out = append(out, key)
}
return out
}
// wgKeyFromBytes mirrors the client-side decoder: the envelope ships raw 32
// bytes; reconstruct the standard base64 key the test compares against.
func wgKeyFromBytes(raw []byte) string {
if len(raw) == 0 {
return ""
}
var k wgtypes.Key
if len(raw) != len(k) {
return ""
}
copy(k[:], raw)
return k.String()
}
func TestClient_SyncGathersInfoOnEveryConnect(t *testing.T) {
s, lis, mgmtMockServer, serverKey := startMockManagement(t)
defer s.GracefulStop()
testKey, err := wgtypes.GenerateKey()
require.NoError(t, err)
hostnames := make(chan string, 2)
mgmtMockServer.SyncFunc = func(msg *mgmtProto.EncryptedMessage, _ mgmtProto.ManagementService_SyncServer) error {
peerKey, err := wgtypes.ParseKey(msg.GetWgPubKey())
if err != nil {
t.Errorf("invalid peer key: %v", err)
return status.Error(codes.InvalidArgument, err.Error())
}
syncReq := &mgmtProto.SyncRequest{}
if err := encryption.DecryptMessage(peerKey, serverKey, msg.Body, syncReq); err != nil {
t.Errorf("decrypt sync request: %v", err)
return status.Error(codes.InvalidArgument, err.Error())
}
select {
case hostnames <- syncReq.GetMeta().GetHostname():
default:
}
// Returning closes the stream, so the client reconnects and gathers again.
return nil
}
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
client, err := NewClient(ctx, lis.Addr().String(), testKey, false)
require.NoError(t, err)
var gathers atomic.Int32
done := make(chan struct{})
go func() {
defer close(done)
_ = client.Sync(ctx, func(ctx context.Context) *system.Info {
info := system.GetInfo(ctx)
info.Hostname = fmt.Sprintf("host-%d", gathers.Add(1))
return info
}, func(*mgmtProto.SyncResponse) error { return nil })
}()
// A connect attempt can fail before it reaches the server, so the sequence
// numbers seen here may skip. What matters is that the reconnect carries a
// newly gathered info instead of the one sent on the previous stream.
var seen []int
for len(seen) < 2 {
select {
case got := <-hostnames:
var n int
_, err := fmt.Sscanf(got, "host-%d", &n)
require.NoError(t, err, "hostname should carry the gather sequence number")
seen = append(seen, n)
case <-time.After(10 * time.Second):
t.Fatalf("timeout waiting for the second sync request, got %v", seen)
}
}
assert.Greater(t, seen[1], seen[0], "the reconnect should carry a newly gathered info")
cancel()
select {
case <-done:
case <-time.After(10 * time.Second):
t.Fatal("timeout waiting for Sync to return after cancel")
}
}
func Test_SystemMetaDataFromClient(t *testing.T) {
s, lis, mgmtMockServer, serverKey := startMockManagement(t)
defer s.GracefulStop()
@@ -532,7 +663,7 @@ func Test_GetPKCEAuthorizationFlow(t *testing.T) {
expectedFlowInfo := &mgmtProto.PKCEAuthorizationFlow{
ProviderConfig: &mgmtProto.ProviderConfig{
ClientID: "client",
ClientSecret: "secret",
ClientSecret: "secret", //nolint:staticcheck
},
}
+223 -105
View File
@@ -21,9 +21,11 @@ import (
"google.golang.org/grpc/connectivity"
nbgrpc "github.com/netbirdio/netbird/client/grpc"
"github.com/netbirdio/netbird/client/netevents"
"github.com/netbirdio/netbird/client/system"
"github.com/netbirdio/netbird/encryption"
"github.com/netbirdio/netbird/shared/management/domain"
nbmgmtgrpc "github.com/netbirdio/netbird/shared/management/grpc"
"github.com/netbirdio/netbird/shared/management/proto"
"github.com/netbirdio/netbird/util/wsproxy"
)
@@ -33,10 +35,15 @@ const ConnectTimeout = 10 * time.Second
const healthCheckTimeout = 5 * time.Second
const (
// EnvMaxRecvMsgSize overrides the default gRPC max receive message size (4 MB)
// EnvMaxRecvMsgSize overrides the default gRPC max receive message size
// for the management client connection. Value is in bytes.
EnvMaxRecvMsgSize = "NB_MANAGEMENT_GRPC_MAX_MSG_SIZE"
// defaultMaxRecvMsgSize is the max gRPC receive message size used for the
// management client connection when EnvMaxRecvMsgSize is unset or invalid.
// It overrides the gRPC library default of 4 MB.
defaultMaxRecvMsgSize = 1024 * 1024 * 16
errMsgMgmtPublicKey = "failed getting Management Service public key: %s"
errMsgNoMgmtConnection = "no connection to management"
)
@@ -55,6 +62,18 @@ type GrpcClient struct {
connStateCallback ConnStateNotifier
connStateCallbackLock sync.RWMutex
serverURL string
// netMgr gates the stream retry loop on OS-reported network
// availability and sweeps the transport on network change.
netMgr *netevents.Manager
// syncStreamErr holds the last Sync stream error, or nil while the stream
// is established and healthy. GetServerKey succeeds even when the peer
// cannot sync (e.g. the server returns "settings not found"), so the
// health probe must consult this to avoid reporting a healthy management
// connection while the Sync stream keeps failing.
syncStreamMu sync.RWMutex
syncStreamErr error
}
type ExposeRequest struct {
@@ -76,37 +95,58 @@ type ExposeResponse struct {
}
// MaxRecvMsgSize returns the configured max gRPC receive message size from
// the environment, or 0 if unset (which uses the gRPC default of 4 MB).
// the environment, or defaultMaxRecvMsgSize (16 MB) if unset or invalid.
func MaxRecvMsgSize() int {
val := os.Getenv(EnvMaxRecvMsgSize)
if val == "" {
return 0
return defaultMaxRecvMsgSize
}
size, err := strconv.Atoi(val)
if err != nil {
log.Warnf("invalid %s value %q, using default: %v", EnvMaxRecvMsgSize, val, err)
return 0
return defaultMaxRecvMsgSize
}
if size <= 0 {
log.Warnf("invalid %s value %d, must be positive, using default", EnvMaxRecvMsgSize, size)
return 0
return defaultMaxRecvMsgSize
}
return size
}
// Option configures optional GrpcClient behavior.
type Option func(*GrpcClient)
// WithNetEvents injects the OS network event handling.
func WithNetEvents(events *netevents.Manager) Option {
return func(c *GrpcClient) { c.netMgr = events }
}
// NewClient creates a new client to Management service
func NewClient(ctx context.Context, addr string, ourPrivateKey wgtypes.Key, tlsEnabled bool) (*GrpcClient, error) {
var conn *grpc.ClientConn
func NewClient(ctx context.Context, addr string, ourPrivateKey wgtypes.Key, tlsEnabled bool, opts ...Option) (*GrpcClient, error) {
// Options apply before dialing: the sweeper must wrap the first connection too.
c := &GrpcClient{
key: ourPrivateKey,
ctx: ctx,
connStateCallbackLock: sync.RWMutex{},
serverURL: addr,
}
for _, opt := range opts {
opt(c)
}
var extraOpts []grpc.DialOption
if maxSize := MaxRecvMsgSize(); maxSize > 0 {
extraOpts = append(extraOpts, grpc.WithDefaultCallOptions(grpc.MaxCallRecvMsgSize(maxSize)))
log.Infof("management gRPC max receive message size set to %d bytes", maxSize)
}
if c.netMgr != nil {
extraOpts = append(extraOpts, nbgrpc.WithSweeper(c.netMgr))
}
var conn *grpc.ClientConn
operation := func() error {
var err error
conn, err = nbgrpc.CreateConnection(ctx, addr, tlsEnabled, wsproxy.ManagementComponent, extraOpts...)
@@ -122,16 +162,9 @@ func NewClient(ctx context.Context, addr string, ourPrivateKey wgtypes.Key, tlsE
return nil, err
}
realClient := proto.NewManagementServiceClient(conn)
return &GrpcClient{
key: ourPrivateKey,
realClient: realClient,
ctx: ctx,
conn: conn,
connStateCallbackLock: sync.RWMutex{},
serverURL: addr,
}, nil
c.conn = conn
c.realClient = proto.NewManagementServiceClient(conn)
return c, nil
}
// GetServerURL returns the management server URL
@@ -172,17 +205,17 @@ func (c *GrpcClient) ready() bool {
// Sync wraps the real client's Sync endpoint call and takes care of retries and encryption/decryption of messages
// Blocking request. The result will be sent via msgHandler callback function
func (c *GrpcClient) Sync(ctx context.Context, sysInfo *system.Info, msgHandler func(msg *proto.SyncResponse) error) error {
return c.withMgmtStream(ctx, func(ctx context.Context, serverPubKey wgtypes.Key) error {
return c.handleSyncStream(ctx, serverPubKey, sysInfo, msgHandler)
func (c *GrpcClient) Sync(ctx context.Context, getInfo func(ctx context.Context) *system.Info, msgHandler func(msg *proto.SyncResponse) error) error {
return c.withMgmtStream(ctx, func(ctx context.Context, serverPubKey wgtypes.Key, backOff backoff.BackOff) error {
return c.handleSyncStream(ctx, serverPubKey, getInfo, msgHandler, backOff)
})
}
// Job wraps the real client's Job endpoint call and takes care of retries and encryption/decryption of messages
// Blocking request. The result will be sent via msgHandler callback function
func (c *GrpcClient) Job(ctx context.Context, msgHandler func(msg *proto.JobRequest) *proto.JobResponse) error {
return c.withMgmtStream(ctx, func(ctx context.Context, serverPubKey wgtypes.Key) error {
return c.handleJobStream(ctx, serverPubKey, msgHandler)
return c.withMgmtStream(ctx, func(ctx context.Context, serverPubKey wgtypes.Key, backOff backoff.BackOff) error {
return c.handleJobStream(ctx, serverPubKey, msgHandler, backOff)
})
}
@@ -190,18 +223,38 @@ func (c *GrpcClient) Job(ctx context.Context, msgHandler func(msg *proto.JobRequ
// It takes care of retries, connection readiness, and fetching server public key.
func (c *GrpcClient) withMgmtStream(
ctx context.Context,
handler func(ctx context.Context, serverPubKey wgtypes.Key) error,
handler func(ctx context.Context, serverPubKey wgtypes.Key, backOff backoff.BackOff) error,
) error {
backOff := defaultBackoff(ctx)
backOff := c.netMgr.QuickRetryBackoff(ctx, defaultBackoff(ctx))
operation := func() error {
log.Debugf("management connection state %v", c.conn.GetState())
connState := c.conn.GetState()
// suspend reconnect attempts while the OS reports no usable network.
// Wait only errors on a cancelled context, which means shutdown, so
// stop the loop without reporting a failure.
if waited, err := c.netMgr.Wait(ctx); err != nil {
log.Debugf("management connection context has been canceled while offline, this usually indicates shutdown")
return nil //nolint:nilerr // a cancelled context means shutdown, not a retryable failure
} else if waited {
backOff.Reset()
// dials attempted while offline grew the channel's internal backoff;
// reset it too, or the reconnect waits out that timer first
c.conn.ResetConnectBackoff()
}
connState := c.conn.GetState()
log.Debugf("management connection state %v", connState)
if connState == connectivity.Shutdown {
return backoff.Permanent(fmt.Errorf("connection to management has been shut down"))
} else if !(connState == connectivity.Ready || connState == connectivity.Idle) {
}
if !(connState == connectivity.Ready || connState == connectivity.Idle) {
// A dial may already be in flight (e.g. the other stream triggered
// it after a network change); wait for it to settle and proceed if
// the channel became usable, instead of burning a backoff round on
// a successful dial. A failed dial errors out as before.
c.conn.WaitForStateChange(ctx, connState)
return fmt.Errorf("connection to management is not ready and in %s state", connState)
connState = c.conn.GetState()
if !(connState == connectivity.Ready || connState == connectivity.Idle) {
return fmt.Errorf("connection to management is not ready and in %s state", connState)
}
}
serverPubKey, err := c.getServerPublicKey()
@@ -210,10 +263,10 @@ func (c *GrpcClient) withMgmtStream(
return err
}
return handler(ctx, *serverPubKey)
return handler(ctx, *serverPubKey, backOff)
}
err := backoff.Retry(operation, backOff)
err := nbgrpc.Retry(ctx, operation, backOff, c.netMgr)
if err != nil {
log.Warnf("exiting the Management service connection retry loop due to the unrecoverable error: %s", err)
}
@@ -225,6 +278,7 @@ func (c *GrpcClient) handleJobStream(
ctx context.Context,
serverPubKey wgtypes.Key,
msgHandler func(msg *proto.JobRequest) *proto.JobResponse,
backOff backoff.BackOff,
) error {
ctx, cancelStream := context.WithCancel(ctx)
defer cancelStream()
@@ -242,33 +296,40 @@ func (c *GrpcClient) handleJobStream(
log.Debug("job stream handshake sent successfully")
// The stream is up, so reset the backoff. This matters for two reasons,
// both caused by the backoff lib not resetting its state on a successful
// connection:
// 1. Without a reset, after a connect followed by an error the next retry
// starts from the accumulated (large) interval instead of retrying
// promptly, delaying reconnection.
// 2. Worse, once the accumulated elapsed time exceeds MaxElapsedTime, the
// next stream error makes NextBackOff() return Stop, so the retry loop
// exits immediately. That error is then mislabeled unrecoverable and
// bubbles up to trigger a full engine restart / data-plane teardown
// instead of a silent reconnection.
backOff.Reset()
// Main loop: receive, process, respond
for {
jobReq, err := c.receiveJobRequest(ctx, stream, serverPubKey)
if err != nil {
if ctx.Err() != nil {
log.Debugf("job stream context has been canceled, this usually indicates shutdown")
return nil
}
if s, ok := gstatus.FromError(err); ok {
switch s.Code() {
case codes.PermissionDenied:
c.notifyDisconnected(err)
return backoff.Permanent(err) // unrecoverable error, propagate to the upper layer
case codes.Canceled:
log.Debugf("management connection context has been canceled, this usually indicates shutdown")
return err
case codes.Unimplemented:
log.Warn("Job feature is not supported by the current management server version. " +
"Please update the management service to use this feature.")
return nil
default:
c.notifyDisconnected(err)
log.Warnf("disconnected from the Management service but will retry silently. Reason: %v", err)
return err
}
} else {
// non-gRPC error
c.notifyDisconnected(err)
log.Warnf("disconnected from the Management service but will retry silently. Reason: %v", err)
return err
}
log.Warnf("job stream disconnected, will retry silently. Reason: %v", err)
return err
}
if jobReq == nil || len(jobReq.ID) == 0 {
@@ -363,13 +424,15 @@ func (c *GrpcClient) sendJobResponse(
return nil
}
func (c *GrpcClient) handleSyncStream(ctx context.Context, serverPubKey wgtypes.Key, sysInfo *system.Info, msgHandler func(msg *proto.SyncResponse) error) error {
func (c *GrpcClient) handleSyncStream(ctx context.Context, serverPubKey wgtypes.Key, getInfo func(ctx context.Context) *system.Info, msgHandler func(msg *proto.SyncResponse) error, backOff backoff.BackOff) error {
ctx, cancelStream := context.WithCancel(ctx)
defer cancelStream()
stream, err := c.connectToSyncStream(ctx, serverPubKey, sysInfo)
stream, err := c.connectToSyncStream(ctx, serverPubKey, getInfo(ctx))
if err != nil {
log.Debugf("failed to open Management Service stream: %s", err)
c.notifyDisconnected(err)
c.setSyncStreamDisconnected(err)
if s, ok := gstatus.FromError(err); ok && s.Code() == codes.PermissionDenied {
return backoff.Permanent(err) // unrecoverable error, propagate to the upper layer
}
@@ -378,75 +441,40 @@ func (c *GrpcClient) handleSyncStream(ctx context.Context, serverPubKey wgtypes.
log.Infof("connected to the Management Service stream")
c.notifyConnected()
c.setSyncStreamConnected()
// The stream is up, so reset the backoff. This matters for two reasons,
// both caused by the backoff lib not resetting its state on a successful
// connection:
// 1. Without a reset, after a connect followed by an error the next retry
// starts from the accumulated (large) interval instead of retrying
// promptly, delaying reconnection.
// 2. Worse, once the accumulated elapsed time exceeds MaxElapsedTime, the
// next stream error makes NextBackOff() return Stop, so the retry loop
// exits immediately. That error is then mislabeled unrecoverable and
// bubbles up to trigger a full engine restart / data-plane teardown
// instead of a silent reconnection.
backOff.Reset()
// blocking until error
err = c.receiveUpdatesEvents(stream, serverPubKey, msgHandler)
if err != nil {
c.notifyDisconnected(err)
if s, ok := gstatus.FromError(err); ok {
switch s.Code() {
case codes.PermissionDenied:
return backoff.Permanent(err) // unrecoverable error, propagate to the upper layer
case codes.Canceled:
log.Debugf("management connection context has been canceled, this usually indicates shutdown")
return nil
default:
log.Warnf("disconnected from the Management service but will retry silently. Reason: %v", err)
return err
}
} else {
// non-gRPC error
log.Warnf("disconnected from the Management service but will retry silently. Reason: %v", err)
return err
c.setSyncStreamDisconnected(err)
if ctx.Err() != nil {
log.Debugf("management connection context has been canceled, this usually indicates shutdown")
return nil
}
if s, ok := gstatus.FromError(err); ok && s.Code() == codes.PermissionDenied {
return backoff.Permanent(err) // unrecoverable error, propagate to the upper layer
}
log.Warnf("disconnected from the Management service but will retry silently. Reason: %v", err)
return err
}
return nil
}
// GetNetworkMap return with the network map
func (c *GrpcClient) GetNetworkMap(sysInfo *system.Info) (*proto.NetworkMap, error) {
serverPubKey, err := c.getServerPublicKey()
if err != nil {
log.Debugf("failed getting Management Service public key: %s", err)
return nil, err
}
ctx, cancelStream := context.WithCancel(c.ctx)
defer cancelStream()
stream, err := c.connectToSyncStream(ctx, *serverPubKey, sysInfo)
if err != nil {
log.Debugf("failed to open Management Service stream: %s", err)
return nil, err
}
defer func() {
_ = stream.CloseSend()
}()
update, err := stream.Recv()
if err == io.EOF {
log.Debugf("Management stream has been closed by server: %s", err)
return nil, err
}
if err != nil {
log.Debugf("disconnected from Management Service sync stream: %v", err)
return nil, err
}
decryptedResp := &proto.SyncResponse{}
err = encryption.DecryptMessage(*serverPubKey, c.key, update.Body, decryptedResp)
if err != nil {
log.Errorf("failed decrypting update message from Management Service: %s", err)
return nil, err
}
if decryptedResp.GetNetworkMap() == nil {
return nil, fmt.Errorf("invalid msg, required network map")
}
return decryptedResp.GetNetworkMap(), nil
}
func (c *GrpcClient) connectToSyncStream(ctx context.Context, serverPubKey wgtypes.Key, sysInfo *system.Info) (proto.ManagementService_SyncClient, error) {
req := &proto.SyncRequest{Meta: infoToMetaData(sysInfo)}
@@ -537,12 +565,19 @@ func (c *GrpcClient) IsHealthy() bool {
ctx, cancel := context.WithTimeout(c.ctx, healthCheckTimeout)
defer cancel()
_, err := c.realClient.GetServerKey(ctx, &proto.Empty{})
_, err := c.realClient.IsHealthy(ctx, &proto.Empty{})
if err != nil {
c.notifyDisconnected(err)
log.Warnf("health check returned: %s", err)
return false
}
if syncErr := c.syncStreamError(); syncErr != nil {
c.notifyDisconnected(syncErr)
log.Warnf("management transport is up but the Sync stream is unhealthy: %s", syncErr)
return false
}
c.notifyConnected()
return true
}
@@ -620,6 +655,49 @@ func (c *GrpcClient) Login(sysInfo *system.Info, pubSSHKey []byte, dnsLabels dom
return c.login(&proto.LoginRequest{Meta: infoToMetaData(sysInfo), PeerKeys: keys, DnsLabels: dnsLabels.ToPunycodeList()})
}
// ExtendAuthSession refreshes the peer's SSO session deadline on the management
// server using a freshly issued JWT. The tunnel is untouched: no network map
// sync, no peer reconnect. Returns the new absolute UTC deadline (zero time
// when the server reports the field empty).
func (c *GrpcClient) ExtendAuthSession(sysInfo *system.Info, jwtToken string) (*proto.ExtendAuthSessionResponse, error) {
if !c.ready() {
return nil, errors.New(errMsgNoMgmtConnection)
}
serverKey, err := c.getServerPublicKey()
if err != nil {
return nil, err
}
reqBody, err := encryption.EncryptMessage(*serverKey, c.key, &proto.ExtendAuthSessionRequest{
JwtToken: jwtToken,
Meta: infoToMetaData(sysInfo),
})
if err != nil {
log.Errorf("failed to encrypt extend auth session message: %s", err)
return nil, err
}
mgmCtx, cancel := context.WithTimeout(c.ctx, ConnectTimeout)
defer cancel()
resp, err := c.realClient.ExtendAuthSession(mgmCtx, &proto.EncryptedMessage{
WgPubKey: c.key.PublicKey().String(),
Body: reqBody,
})
if err != nil {
log.Errorf("failed to extend auth session on Management Service: %v", err)
return nil, err
}
out := &proto.ExtendAuthSessionResponse{}
if err := encryption.DecryptMessage(*serverKey, c.key, resp.Body, out); err != nil {
log.Errorf("failed to decrypt extend auth session response: %s", err)
return nil, err
}
return out, nil
}
// GetDeviceAuthorizationFlow returns a device authorization flow information.
// It also takes care of encrypting and decrypting messages.
func (c *GrpcClient) GetDeviceAuthorizationFlow() (*proto.DeviceAuthorizationFlow, error) {
@@ -729,6 +807,24 @@ func (c *GrpcClient) SyncMeta(sysInfo *system.Info) error {
return err
}
func (c *GrpcClient) setSyncStreamConnected() {
c.syncStreamMu.Lock()
defer c.syncStreamMu.Unlock()
c.syncStreamErr = nil
}
func (c *GrpcClient) setSyncStreamDisconnected(err error) {
c.syncStreamMu.Lock()
defer c.syncStreamMu.Unlock()
c.syncStreamErr = err
}
func (c *GrpcClient) syncStreamError() error {
c.syncStreamMu.RLock()
defer c.syncStreamMu.RUnlock()
return c.syncStreamErr
}
func (c *GrpcClient) notifyDisconnected(err error) {
c.connStateCallbackLock.RLock()
defer c.connStateCallbackLock.RUnlock()
@@ -943,6 +1039,7 @@ func infoToMetaData(info *system.Info) *proto.PeerSystemMeta {
RosenpassEnabled: info.RosenpassEnabled,
RosenpassPermissive: info.RosenpassPermissive,
ServerSSHAllowed: info.ServerSSHAllowed,
RemoteJobsAllowed: info.RemoteJobsAllowed,
DisableClientRoutes: info.DisableClientRoutes,
DisableServerRoutes: info.DisableServerRoutes,
@@ -950,8 +1047,29 @@ func infoToMetaData(info *system.Info) *proto.PeerSystemMeta {
DisableFirewall: info.DisableFirewall,
BlockLANAccess: info.BlockLANAccess,
BlockInbound: info.BlockInbound,
LazyConnectionEnabled: info.LazyConnectionEnabled,
DisableIPv6: info.DisableIPv6,
},
Capabilities: peerCapabilities(*info),
SyncMessageVersion: syncMessageVersion(*info),
}
}
// peerCapabilities returns the capabilities this client supports.
func peerCapabilities(info system.Info) []proto.PeerCapability {
caps := []proto.PeerCapability{
proto.PeerCapability_PeerCapabilitySourcePrefixes,
}
if !info.DisableIPv6 {
caps = append(caps, proto.PeerCapability_PeerCapabilityIPv6Overlay)
}
return caps
}
func syncMessageVersion(info system.Info) int32 {
if info.SyncMessageVersion != nil {
return int32(*info.SyncMessageVersion)
}
return int32(nbmgmtgrpc.HighestSyncMessageVersion)
}
+4 -4
View File
@@ -21,11 +21,11 @@ func TestMaxRecvMsgSize(t *testing.T) {
envValue string
expected int
}{
{name: "unset returns 0", envValue: "", expected: 0},
{name: "unset returns default", envValue: "", expected: defaultMaxRecvMsgSize},
{name: "valid value", envValue: "10485760", expected: 10485760},
{name: "non-numeric returns 0", envValue: "abc", expected: 0},
{name: "negative returns 0", envValue: "-1", expected: 0},
{name: "zero returns 0", envValue: "0", expected: 0},
{name: "non-numeric returns default", envValue: "abc", expected: defaultMaxRecvMsgSize},
{name: "negative returns default", envValue: "-1", expected: defaultMaxRecvMsgSize},
{name: "zero returns default", envValue: "0", expected: defaultMaxRecvMsgSize},
}
for _, tt := range tests {
+11 -8
View File
@@ -11,9 +11,10 @@ import (
// MockClient is a mock implementation of the Client interface for testing.
type MockClient struct {
CloseFunc func() error
SyncFunc func(ctx context.Context, sysInfo *system.Info, msgHandler func(msg *proto.SyncResponse) error) error
SyncFunc func(ctx context.Context, getInfo func(ctx context.Context) *system.Info, msgHandler func(msg *proto.SyncResponse) error) error
RegisterFunc func(setupKey string, jwtToken string, info *system.Info, sshKey []byte, dnsLabels domain.List) (*proto.LoginResponse, error)
LoginFunc func(info *system.Info, sshKey []byte, dnsLabels domain.List) (*proto.LoginResponse, error)
ExtendAuthSessionFunc func(info *system.Info, jwtToken string) (*proto.ExtendAuthSessionResponse, error)
GetDeviceAuthorizationFlowFunc func() (*proto.DeviceAuthorizationFlow, error)
GetPKCEAuthorizationFlowFunc func() (*proto.PKCEAuthorizationFlow, error)
GetServerURLFunc func() string
@@ -37,11 +38,11 @@ func (m *MockClient) Close() error {
return m.CloseFunc()
}
func (m *MockClient) Sync(ctx context.Context, sysInfo *system.Info, msgHandler func(msg *proto.SyncResponse) error) error {
func (m *MockClient) Sync(ctx context.Context, getInfo func(ctx context.Context) *system.Info, msgHandler func(msg *proto.SyncResponse) error) error {
if m.SyncFunc == nil {
return nil
}
return m.SyncFunc(ctx, sysInfo, msgHandler)
return m.SyncFunc(ctx, getInfo, msgHandler)
}
func (m *MockClient) Job(ctx context.Context, msgHandler func(msg *proto.JobRequest) *proto.JobResponse) error {
@@ -65,6 +66,13 @@ func (m *MockClient) Login(info *system.Info, sshKey []byte, dnsLabels domain.Li
return m.LoginFunc(info, sshKey, dnsLabels)
}
func (m *MockClient) ExtendAuthSession(info *system.Info, jwtToken string) (*proto.ExtendAuthSessionResponse, error) {
if m.ExtendAuthSessionFunc == nil {
return nil, nil
}
return m.ExtendAuthSessionFunc(info, jwtToken)
}
func (m *MockClient) GetDeviceAuthorizationFlow() (*proto.DeviceAuthorizationFlow, error) {
if m.GetDeviceAuthorizationFlowFunc == nil {
return nil, nil
@@ -86,11 +94,6 @@ func (m *MockClient) HealthCheck() error {
return m.HealthCheckFunc()
}
// GetNetworkMap mock implementation of GetNetworkMap from Client interface.
func (m *MockClient) GetNetworkMap(_ *system.Info) (*proto.NetworkMap, error) {
return nil, nil
}
// GetServerURL mock implementation of GetServerURL from mgm.Client interface
func (m *MockClient) GetServerURL() string {
if m.GetServerURLFunc == nil {
@@ -0,0 +1,417 @@
package rest
import (
"bytes"
"context"
"encoding/json"
"io"
"net/http"
"github.com/netbirdio/netbird/shared/management/http/api"
)
// AgentNetworkAPI APIs for the Agent Network (AI/LLM gateway), do not use directly
// see more: https://docs.netbird.io/api/resources/agent-network
type AgentNetworkAPI struct {
c *Client
}
// ListCatalogProviders lists the catalog of supported upstream AI providers
// (openai_api, anthropic_api, bedrock_api, ...) with their default models and
// pricing, used to prefill provider create forms.
func (a *AgentNetworkAPI) ListCatalogProviders(ctx context.Context) ([]api.AgentNetworkCatalogProvider, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/catalog/providers", nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[[]api.AgentNetworkCatalogProvider](resp)
return ret, err
}
// ListProviders lists all Agent Network providers
func (a *AgentNetworkAPI) ListProviders(ctx context.Context) ([]api.AgentNetworkProvider, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/providers", nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[[]api.AgentNetworkProvider](resp)
return ret, err
}
// GetProvider gets Agent Network provider info
func (a *AgentNetworkAPI) GetProvider(ctx context.Context, providerID string) (*api.AgentNetworkProvider, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/providers/"+providerID, nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkProvider](resp)
return &ret, err
}
// CreateProvider creates a new Agent Network provider. Providers have no
// settings side effects — bootstrap the account's gateway endpoint separately
// via CreateSettings.
func (a *AgentNetworkAPI) CreateProvider(ctx context.Context, request api.PostApiAgentNetworkProvidersJSONRequestBody) (*api.AgentNetworkProvider, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "POST", "/api/agent-network/providers", bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkProvider](resp)
return &ret, err
}
// UpdateProvider updates an Agent Network provider. The request replaces the
// provider's mutable state; only an omitted api_key keeps the stored key
// (secrets are never required to round-trip).
func (a *AgentNetworkAPI) UpdateProvider(ctx context.Context, providerID string, request api.PutApiAgentNetworkProvidersProviderIdJSONRequestBody) (*api.AgentNetworkProvider, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "PUT", "/api/agent-network/providers/"+providerID, bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkProvider](resp)
return &ret, err
}
// DeleteProvider deletes an Agent Network provider. Fails while any policy
// still references the provider — detach it first.
func (a *AgentNetworkAPI) DeleteProvider(ctx context.Context, providerID string) error {
resp, err := a.c.NewRequest(ctx, "DELETE", "/api/agent-network/providers/"+providerID, nil, nil)
if err != nil {
return err
}
if resp.Body != nil {
defer resp.Body.Close()
}
return nil
}
// ListPolicies lists all Agent Network policies
func (a *AgentNetworkAPI) ListPolicies(ctx context.Context) ([]api.AgentNetworkPolicy, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/policies", nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[[]api.AgentNetworkPolicy](resp)
return ret, err
}
// GetPolicy gets Agent Network policy info
func (a *AgentNetworkAPI) GetPolicy(ctx context.Context, policyID string) (*api.AgentNetworkPolicy, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/policies/"+policyID, nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkPolicy](resp)
return &ret, err
}
// CreatePolicy creates a new Agent Network policy
func (a *AgentNetworkAPI) CreatePolicy(ctx context.Context, request api.PostApiAgentNetworkPoliciesJSONRequestBody) (*api.AgentNetworkPolicy, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "POST", "/api/agent-network/policies", bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkPolicy](resp)
return &ret, err
}
// UpdatePolicy updates an Agent Network policy
func (a *AgentNetworkAPI) UpdatePolicy(ctx context.Context, policyID string, request api.PutApiAgentNetworkPoliciesPolicyIdJSONRequestBody) (*api.AgentNetworkPolicy, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "PUT", "/api/agent-network/policies/"+policyID, bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkPolicy](resp)
return &ret, err
}
// DeletePolicy deletes an Agent Network policy
func (a *AgentNetworkAPI) DeletePolicy(ctx context.Context, policyID string) error {
resp, err := a.c.NewRequest(ctx, "DELETE", "/api/agent-network/policies/"+policyID, nil, nil)
if err != nil {
return err
}
if resp.Body != nil {
defer resp.Body.Close()
}
return nil
}
// ListGuardrails lists all Agent Network guardrails
func (a *AgentNetworkAPI) ListGuardrails(ctx context.Context) ([]api.AgentNetworkGuardrail, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/guardrails", nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[[]api.AgentNetworkGuardrail](resp)
return ret, err
}
// GetGuardrail gets Agent Network guardrail info
func (a *AgentNetworkAPI) GetGuardrail(ctx context.Context, guardrailID string) (*api.AgentNetworkGuardrail, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/guardrails/"+guardrailID, nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkGuardrail](resp)
return &ret, err
}
// CreateGuardrail creates a new Agent Network guardrail
func (a *AgentNetworkAPI) CreateGuardrail(ctx context.Context, request api.PostApiAgentNetworkGuardrailsJSONRequestBody) (*api.AgentNetworkGuardrail, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "POST", "/api/agent-network/guardrails", bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkGuardrail](resp)
return &ret, err
}
// UpdateGuardrail updates an Agent Network guardrail
func (a *AgentNetworkAPI) UpdateGuardrail(ctx context.Context, guardrailID string, request api.PutApiAgentNetworkGuardrailsGuardrailIdJSONRequestBody) (*api.AgentNetworkGuardrail, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "PUT", "/api/agent-network/guardrails/"+guardrailID, bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkGuardrail](resp)
return &ret, err
}
// DeleteGuardrail deletes an Agent Network guardrail
func (a *AgentNetworkAPI) DeleteGuardrail(ctx context.Context, guardrailID string) error {
resp, err := a.c.NewRequest(ctx, "DELETE", "/api/agent-network/guardrails/"+guardrailID, nil, nil)
if err != nil {
return err
}
if resp.Body != nil {
defer resp.Body.Close()
}
return nil
}
// ListBudgetRules lists all account-level Agent Network budget rules
func (a *AgentNetworkAPI) ListBudgetRules(ctx context.Context) ([]api.AgentNetworkBudgetRule, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/budget-rules", nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[[]api.AgentNetworkBudgetRule](resp)
return ret, err
}
// GetBudgetRule gets Agent Network budget rule info
func (a *AgentNetworkAPI) GetBudgetRule(ctx context.Context, ruleID string) (*api.AgentNetworkBudgetRule, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/budget-rules/"+ruleID, nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkBudgetRule](resp)
return &ret, err
}
// CreateBudgetRule creates a new Agent Network budget rule
func (a *AgentNetworkAPI) CreateBudgetRule(ctx context.Context, request api.PostApiAgentNetworkBudgetRulesJSONRequestBody) (*api.AgentNetworkBudgetRule, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "POST", "/api/agent-network/budget-rules", bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkBudgetRule](resp)
return &ret, err
}
// UpdateBudgetRule updates an Agent Network budget rule
func (a *AgentNetworkAPI) UpdateBudgetRule(ctx context.Context, ruleID string, request api.PutApiAgentNetworkBudgetRulesRuleIdJSONRequestBody) (*api.AgentNetworkBudgetRule, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "PUT", "/api/agent-network/budget-rules/"+ruleID, bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkBudgetRule](resp)
return &ret, err
}
// DeleteBudgetRule deletes an Agent Network budget rule
func (a *AgentNetworkAPI) DeleteBudgetRule(ctx context.Context, ruleID string) error {
resp, err := a.c.NewRequest(ctx, "DELETE", "/api/agent-network/budget-rules/"+ruleID, nil, nil)
if err != nil {
return err
}
if resp.Body != nil {
defer resp.Body.Close()
}
return nil
}
// GetSettings gets the account's Agent Network gateway settings (endpoint,
// proxy address, collection toggles). An account that has not been
// bootstrapped yet — via CreateSettings — reads as the defaults with an empty
// Endpoint and ProxyAddress. Management servers prior to that contract
// answered 200 with a JSON null body instead; that legacy shape is translated
// to an APIError matchable via IsNotFound rather than fabricating defaults
// the server never stated.
func (a *AgentNetworkAPI) GetSettings(ctx context.Context) (*api.AgentNetworkSettings, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/agent-network/settings", nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
body, err := io.ReadAll(resp.Body)
if err != nil {
return nil, err
}
if trimmed := bytes.TrimSpace(body); len(trimmed) == 0 || bytes.Equal(trimmed, []byte("null")) {
return nil, &APIError{StatusCode: http.StatusNotFound, Message: "agent network settings not found"}
}
var ret api.AgentNetworkSettings
if err := json.Unmarshal(body, &ret); err != nil {
return nil, err
}
return &ret, nil
}
// CreateSettings bootstraps the account's Agent Network settings row,
// assigning the immutable endpoint. Exactly one of request.ProxyAddress
// (labeled endpoint beneath that cluster; the server allocates the label) and
// request.Endpoint (self-addressed dedicated endpoint, claimed verbatim) must
// be set. Returns a conflict when the account already has a settings row.
func (a *AgentNetworkAPI) CreateSettings(ctx context.Context, request api.PostApiAgentNetworkSettingsJSONRequestBody) (*api.AgentNetworkSettings, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "POST", "/api/agent-network/settings", bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkSettings](resp)
return &ret, err
}
// UpdateSettings updates the account's Agent Network settings; the request
// carries every field, replacing the mutable ones (collection toggles and
// retention). The endpoint and proxy address are assigned at bootstrap
// (CreateSettings) and immutable — the request must echo them unchanged, and
// a request carrying different values is rejected. Returns not-found until
// the account is bootstrapped.
func (a *AgentNetworkAPI) UpdateSettings(ctx context.Context, request api.PutApiAgentNetworkSettingsJSONRequestBody) (*api.AgentNetworkSettings, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "PUT", "/api/agent-network/settings", bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.AgentNetworkSettings](resp)
return &ret, err
}
// DeleteSettings deletes the account's Agent Network settings row, releasing
// the endpoint. The server refuses (precondition failed) while any provider
// exists for the account or while a proxy is actively serving the endpoint.
// Bootstrapping again afterwards allocates a new endpoint.
func (a *AgentNetworkAPI) DeleteSettings(ctx context.Context) error {
resp, err := a.c.NewRequest(ctx, "DELETE", "/api/agent-network/settings", nil, nil)
if err != nil {
return err
}
if resp.Body != nil {
defer resp.Body.Close()
}
return nil
}
@@ -0,0 +1,561 @@
//go:build integration
package rest_test
import (
"context"
"encoding/json"
"io"
"net/http"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/shared/management/client/rest"
"github.com/netbirdio/netbird/shared/management/http/api"
"github.com/netbirdio/netbird/shared/management/http/util"
)
var (
testAgentNetworkProvider = api.AgentNetworkProvider{
Id: "ainp_test",
ProviderId: "openai_api",
Name: "OpenAI",
UpstreamUrl: "https://api.openai.com",
Models: []api.AgentNetworkProviderModel{},
Enabled: true,
}
testAgentNetworkPolicy = api.AgentNetworkPolicy{
Id: "ainpol_test",
Name: "Engineering → OpenAI",
Enabled: true,
SourceGroups: []string{"grp-eng"},
DestinationProviderIds: []string{"ainp_test"},
}
testAgentNetworkGuardrail = api.AgentNetworkGuardrail{
Id: "aingr_test",
Name: "No secrets",
}
testAgentNetworkBudgetRule = api.AgentNetworkBudgetRule{
Id: "ainbud_test",
Name: "Org monthly ceiling",
Enabled: true,
}
testAgentNetworkSettings = api.AgentNetworkSettings{
Endpoint: "violet.eu.proxy.netbird.io",
ProxyAddress: "eu.proxy.netbird.io",
Dedicated: false,
EnableLogCollection: true,
AccessLogRetentionDays: ptr(30),
}
)
func TestAgentNetwork_ListCatalogProviders_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/catalog/providers", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal([]api.AgentNetworkCatalogProvider{{Id: "openai_api", Name: "OpenAI"}})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.ListCatalogProviders(context.Background())
require.NoError(t, err)
assert.Len(t, ret, 1)
assert.Equal(t, "openai_api", ret[0].Id)
})
}
func TestAgentNetwork_ListProviders_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/providers", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal([]api.AgentNetworkProvider{testAgentNetworkProvider})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.ListProviders(context.Background())
require.NoError(t, err)
assert.Len(t, ret, 1)
assert.Equal(t, testAgentNetworkProvider, ret[0])
})
}
func TestAgentNetwork_GetProvider_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/providers/ainp_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "GET", r.Method)
retBytes, _ := json.Marshal(testAgentNetworkProvider)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.GetProvider(context.Background(), "ainp_test")
require.NoError(t, err)
assert.Equal(t, testAgentNetworkProvider, *ret)
})
}
func TestAgentNetwork_GetProvider_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/providers/ainp_test", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "not found", Code: 404})
w.WriteHeader(404)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
_, err := c.AgentNetwork.GetProvider(context.Background(), "ainp_test")
require.Error(t, err)
assert.True(t, rest.IsNotFound(err), "a 404 must be matchable via IsNotFound")
})
}
func TestAgentNetwork_CreateProvider_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/providers", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "POST", r.Method)
reqBytes, err := io.ReadAll(r.Body)
require.NoError(t, err)
var req api.PostApiAgentNetworkProvidersJSONRequestBody
require.NoError(t, json.Unmarshal(reqBytes, &req))
assert.Equal(t, "OpenAI", req.Name)
retBytes, _ := json.Marshal(testAgentNetworkProvider)
_, err = w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.CreateProvider(context.Background(), api.PostApiAgentNetworkProvidersJSONRequestBody{
ProviderId: "openai_api",
Name: "OpenAI",
UpstreamUrl: "https://api.openai.com",
ApiKey: ptr("sk-test"),
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkProvider, *ret)
})
}
func TestAgentNetwork_UpdateProvider_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/providers/ainp_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "PUT", r.Method)
reqBytes, err := io.ReadAll(r.Body)
require.NoError(t, err)
// Omitted optional fields must be absent from the wire (not
// zero-valued) so the server-side merge preserves them.
assert.NotContains(t, string(reqBytes), "api_key")
assert.NotContains(t, string(reqBytes), "models")
retBytes, _ := json.Marshal(testAgentNetworkProvider)
_, err = w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.UpdateProvider(context.Background(), "ainp_test", api.PutApiAgentNetworkProvidersProviderIdJSONRequestBody{
ProviderId: "openai_api",
Name: "OpenAI",
UpstreamUrl: "https://api.openai.com",
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkProvider, *ret)
})
}
func TestAgentNetwork_DeleteProvider_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/providers/ainp_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "DELETE", r.Method)
_, err := w.Write([]byte("{}"))
require.NoError(t, err)
})
err := c.AgentNetwork.DeleteProvider(context.Background(), "ainp_test")
require.NoError(t, err)
})
}
func TestAgentNetwork_ListPolicies_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/policies", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal([]api.AgentNetworkPolicy{testAgentNetworkPolicy})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.ListPolicies(context.Background())
require.NoError(t, err)
assert.Len(t, ret, 1)
assert.Equal(t, testAgentNetworkPolicy, ret[0])
})
}
func TestAgentNetwork_GetPolicy_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/policies/ainpol_test", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(testAgentNetworkPolicy)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.GetPolicy(context.Background(), "ainpol_test")
require.NoError(t, err)
assert.Equal(t, testAgentNetworkPolicy, *ret)
})
}
func TestAgentNetwork_CreatePolicy_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/policies", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "POST", r.Method)
retBytes, _ := json.Marshal(testAgentNetworkPolicy)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.CreatePolicy(context.Background(), api.PostApiAgentNetworkPoliciesJSONRequestBody{
Name: "Engineering → OpenAI",
SourceGroups: []string{"grp-eng"},
DestinationProviderIds: []string{"ainp_test"},
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkPolicy, *ret)
})
}
func TestAgentNetwork_UpdatePolicy_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/policies/ainpol_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "PUT", r.Method)
retBytes, _ := json.Marshal(testAgentNetworkPolicy)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.UpdatePolicy(context.Background(), "ainpol_test", api.PutApiAgentNetworkPoliciesPolicyIdJSONRequestBody{
Name: "Engineering → OpenAI",
SourceGroups: []string{"grp-eng"},
DestinationProviderIds: []string{"ainp_test"},
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkPolicy, *ret)
})
}
func TestAgentNetwork_DeletePolicy_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/policies/ainpol_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "DELETE", r.Method)
_, err := w.Write([]byte("{}"))
require.NoError(t, err)
})
err := c.AgentNetwork.DeletePolicy(context.Background(), "ainpol_test")
require.NoError(t, err)
})
}
func TestAgentNetwork_ListGuardrails_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/guardrails", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal([]api.AgentNetworkGuardrail{testAgentNetworkGuardrail})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.ListGuardrails(context.Background())
require.NoError(t, err)
assert.Len(t, ret, 1)
assert.Equal(t, testAgentNetworkGuardrail, ret[0])
})
}
func TestAgentNetwork_GetGuardrail_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/guardrails/aingr_test", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(testAgentNetworkGuardrail)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.GetGuardrail(context.Background(), "aingr_test")
require.NoError(t, err)
assert.Equal(t, testAgentNetworkGuardrail, *ret)
})
}
func TestAgentNetwork_CreateGuardrail_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/guardrails", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "POST", r.Method)
retBytes, _ := json.Marshal(testAgentNetworkGuardrail)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.CreateGuardrail(context.Background(), api.PostApiAgentNetworkGuardrailsJSONRequestBody{
Name: "No secrets",
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkGuardrail, *ret)
})
}
func TestAgentNetwork_UpdateGuardrail_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/guardrails/aingr_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "PUT", r.Method)
retBytes, _ := json.Marshal(testAgentNetworkGuardrail)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.UpdateGuardrail(context.Background(), "aingr_test", api.PutApiAgentNetworkGuardrailsGuardrailIdJSONRequestBody{
Name: "No secrets",
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkGuardrail, *ret)
})
}
func TestAgentNetwork_DeleteGuardrail_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/guardrails/aingr_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "DELETE", r.Method)
_, err := w.Write([]byte("{}"))
require.NoError(t, err)
})
err := c.AgentNetwork.DeleteGuardrail(context.Background(), "aingr_test")
require.NoError(t, err)
})
}
func TestAgentNetwork_ListBudgetRules_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/budget-rules", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal([]api.AgentNetworkBudgetRule{testAgentNetworkBudgetRule})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.ListBudgetRules(context.Background())
require.NoError(t, err)
assert.Len(t, ret, 1)
assert.Equal(t, testAgentNetworkBudgetRule, ret[0])
})
}
func TestAgentNetwork_GetBudgetRule_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/budget-rules/ainbud_test", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(testAgentNetworkBudgetRule)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.GetBudgetRule(context.Background(), "ainbud_test")
require.NoError(t, err)
assert.Equal(t, testAgentNetworkBudgetRule, *ret)
})
}
func TestAgentNetwork_CreateBudgetRule_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/budget-rules", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "POST", r.Method)
retBytes, _ := json.Marshal(testAgentNetworkBudgetRule)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.CreateBudgetRule(context.Background(), api.PostApiAgentNetworkBudgetRulesJSONRequestBody{
Name: "Org monthly ceiling",
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkBudgetRule, *ret)
})
}
func TestAgentNetwork_UpdateBudgetRule_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/budget-rules/ainbud_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "PUT", r.Method)
retBytes, _ := json.Marshal(testAgentNetworkBudgetRule)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.UpdateBudgetRule(context.Background(), "ainbud_test", api.PutApiAgentNetworkBudgetRulesRuleIdJSONRequestBody{
Name: "Org monthly ceiling",
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkBudgetRule, *ret)
})
}
func TestAgentNetwork_DeleteBudgetRule_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/budget-rules/ainbud_test", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "DELETE", r.Method)
_, err := w.Write([]byte("{}"))
require.NoError(t, err)
})
err := c.AgentNetwork.DeleteBudgetRule(context.Background(), "ainbud_test")
require.NoError(t, err)
})
}
func TestAgentNetwork_GetSettings_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(testAgentNetworkSettings)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.GetSettings(context.Background())
require.NoError(t, err)
assert.Equal(t, testAgentNetworkSettings, *ret)
})
}
// TestAgentNetwork_GetSettings_UnbootstrappedDefaults pins the settings-read
// contract: an unbootstrapped account answers 200 with the defaults and empty
// cluster/subdomain/endpoint, which the client passes through untouched.
func TestAgentNetwork_GetSettings_UnbootstrappedDefaults(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(api.AgentNetworkSettings{
EnableLogCollection: true,
AccessLogRetentionDays: ptr(30),
})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.GetSettings(context.Background())
require.NoError(t, err)
assert.Empty(t, ret.Endpoint, "empty endpoint is the not-bootstrapped signal")
assert.True(t, ret.EnableLogCollection, "defaults must pass through")
})
}
func TestAgentNetwork_GetSettings_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "no", Code: 403})
w.WriteHeader(403)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
_, err := c.AgentNetwork.GetSettings(context.Background())
require.Error(t, err)
assert.Equal(t, "no", err.Error())
})
}
// TestAgentNetwork_GetSettings_LegacyNullBody pins the compatibility shim for
// management servers that answered 200 with a JSON null body before the
// defaults contract: the client translates that shape into an IsNotFound
// error instead of returning a bogus zero-valued settings object or
// fabricating defaults the server never stated.
func TestAgentNetwork_GetSettings_LegacyNullBody(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
_, err := w.Write([]byte("null"))
require.NoError(t, err)
})
ret, err := c.AgentNetwork.GetSettings(context.Background())
require.Error(t, err)
assert.Nil(t, ret)
assert.True(t, rest.IsNotFound(err), "the legacy 200+null shape must surface as IsNotFound")
})
}
func TestAgentNetwork_CreateSettings_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "POST", r.Method)
reqBytes, err := io.ReadAll(r.Body)
require.NoError(t, err)
var req api.PostApiAgentNetworkSettingsJSONRequestBody
require.NoError(t, json.Unmarshal(reqBytes, &req))
require.NotNil(t, req.ProxyAddress, "proxy address must be on the wire")
assert.Equal(t, "eu.proxy.netbird.io", *req.ProxyAddress)
assert.Nil(t, req.Endpoint, "endpoint must stay off the wire for a labeled bootstrap")
retBytes, _ := json.Marshal(testAgentNetworkSettings)
_, err = w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.CreateSettings(context.Background(), api.PostApiAgentNetworkSettingsJSONRequestBody{
ProxyAddress: ptr("eu.proxy.netbird.io"),
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkSettings, *ret)
})
}
func TestAgentNetwork_CreateSettings_Conflict(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "agent network settings already bootstrapped for account acct1", Code: 409})
w.WriteHeader(409)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
_, err := c.AgentNetwork.CreateSettings(context.Background(), api.PostApiAgentNetworkSettingsJSONRequestBody{
Endpoint: ptr("gw.example.com"),
})
require.Error(t, err)
assert.Contains(t, err.Error(), "already bootstrapped")
})
}
func TestAgentNetwork_UpdateSettings_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "PUT", r.Method)
reqBytes, err := io.ReadAll(r.Body)
require.NoError(t, err)
var req api.PutApiAgentNetworkSettingsJSONRequestBody
require.NoError(t, json.Unmarshal(reqBytes, &req))
assert.True(t, req.EnableLogCollection)
assert.Equal(t, "brave-otter.eu.proxy.netbird.io", req.Endpoint,
"the identity echo must be on the wire")
assert.Equal(t, "eu.proxy.netbird.io", req.ProxyAddress,
"the identity echo must be on the wire")
retBytes, _ := json.Marshal(testAgentNetworkSettings)
_, err = w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.AgentNetwork.UpdateSettings(context.Background(), api.PutApiAgentNetworkSettingsJSONRequestBody{
Endpoint: "brave-otter.eu.proxy.netbird.io",
ProxyAddress: "eu.proxy.netbird.io",
EnableLogCollection: true,
})
require.NoError(t, err)
assert.Equal(t, testAgentNetworkSettings, *ret)
})
}
func TestAgentNetwork_UpdateSettings_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "agent network settings have not been bootstrapped yet; POST /api/agent-network/settings to bootstrap them", Code: 404})
w.WriteHeader(404)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
_, err := c.AgentNetwork.UpdateSettings(context.Background(), api.PutApiAgentNetworkSettingsJSONRequestBody{
EnableLogCollection: true,
})
require.Error(t, err)
assert.True(t, rest.IsNotFound(err), "an unbootstrapped account must surface as IsNotFound")
})
}
func TestAgentNetwork_DeleteSettings_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "DELETE", r.Method)
_, err := w.Write([]byte("{}"))
require.NoError(t, err)
})
require.NoError(t, c.AgentNetwork.DeleteSettings(context.Background()))
})
}
func TestAgentNetwork_DeleteSettings_Guarded(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/agent-network/settings", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "agent network settings cannot be deleted while 2 provider(s) exist; delete the providers first", Code: 412})
w.WriteHeader(412)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
err := c.AgentNetwork.DeleteSettings(context.Background())
require.Error(t, err)
assert.Contains(t, err.Error(), "cannot be deleted")
})
}
+10
View File
@@ -143,6 +143,14 @@ type Client struct {
// ReverseProxyDomains NetBird reverse proxy domains APIs
ReverseProxyDomains *ReverseProxyDomainsAPI
// ReverseProxyTokens account-scoped proxy access tokens used to register
// self-hosted (bring-your-own-proxy) `netbird proxy` instances.
ReverseProxyTokens *ReverseProxyTokensAPI
// AgentNetwork NetBird Agent Network (AI/LLM gateway) APIs: catalog,
// providers, policies, guardrails, budget rules and account settings.
AgentNetwork *AgentNetworkAPI
}
// New initialize new Client instance using PAT token
@@ -204,6 +212,8 @@ func (c *Client) initialize() {
c.ReverseProxyServices = &ReverseProxyServicesAPI{c}
c.ReverseProxyClusters = &ReverseProxyClustersAPI{c}
c.ReverseProxyDomains = &ReverseProxyDomainsAPI{c}
c.ReverseProxyTokens = &ReverseProxyTokensAPI{c}
c.AgentNetwork = &AgentNetworkAPI{c}
}
// NewRequest creates and executes new management API request
@@ -2,6 +2,8 @@ package rest
import (
"context"
"errors"
"net/url"
"github.com/netbirdio/netbird/shared/management/http/api"
)
@@ -11,7 +13,10 @@ type ReverseProxyClustersAPI struct {
c *Client
}
// List lists all available proxy clusters
// List lists all available proxy clusters. Each cluster is enriched with the
// capability flags reported by its connected proxies (supports_custom_ports,
// supports_crowdsec, private, etc.), so callers can render UX gates without
// a follow-up round-trip.
func (a *ReverseProxyClustersAPI) List(ctx context.Context) ([]api.ProxyCluster, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/reverse-proxies/clusters", nil, nil)
if err != nil {
@@ -23,3 +28,24 @@ func (a *ReverseProxyClustersAPI) List(ctx context.Context) ([]api.ProxyCluster,
ret, err := parseResponse[[]api.ProxyCluster](resp)
return ret, err
}
// Delete removes every self-hosted (BYOP) proxy registration for the given
// cluster address owned by the calling account. Shared clusters operated by
// NetBird cannot be deleted via this endpoint; the server returns 404 / 400
// for cluster addresses the account does not own.
func (a *ReverseProxyClustersAPI) Delete(ctx context.Context, clusterAddress string) error {
// Guard against the empty input: url.PathEscape("") returns "" which
// would collapse the request URL onto the collection endpoint and
// silently delete nothing (or 405 depending on routing).
if clusterAddress == "" {
return errors.New("clusterAddress is required")
}
resp, err := a.c.NewRequest(ctx, "DELETE", "/api/reverse-proxies/clusters/"+url.PathEscape(clusterAddress), nil, nil)
if err != nil {
return err
}
if resp.Body != nil {
defer resp.Body.Close()
}
return nil
}
@@ -0,0 +1,104 @@
//go:build integration
package rest_test
import (
"context"
"encoding/json"
"net/http"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/shared/management/client/rest"
"github.com/netbirdio/netbird/shared/management/http/api"
"github.com/netbirdio/netbird/shared/management/http/util"
)
func boolPtr(b bool) *bool { return &b }
var testCluster = api.ProxyCluster{
Id: "cluster-1",
Address: "proxy.netbird.local",
Type: "shared",
Online: true,
ConnectedProxies: 2,
SupportsCustomPorts: boolPtr(true),
RequireSubdomain: boolPtr(false),
SupportsCrowdsec: boolPtr(false),
Private: boolPtr(true),
}
func TestReverseProxyClusters_List_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/clusters", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "GET", r.Method, "List must use GET")
retBytes, _ := json.Marshal([]api.ProxyCluster{testCluster})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.ReverseProxyClusters.List(context.Background())
require.NoError(t, err)
require.Len(t, ret, 1)
assert.Equal(t, testCluster.Id, ret[0].Id)
assert.Equal(t, testCluster.Address, ret[0].Address)
require.NotNil(t, ret[0].Private, "private capability must round-trip through the client")
assert.True(t, *ret[0].Private, "private capability must reflect the server value")
})
}
func TestReverseProxyClusters_List_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/clusters", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "No", Code: 500})
w.WriteHeader(500)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.ReverseProxyClusters.List(context.Background())
assert.Error(t, err)
assert.Empty(t, ret)
})
}
func TestReverseProxyClusters_Delete_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
// PathEscape on "proxy.netbird.local" leaves it intact; the route mux
// matches the unescaped form. Sanity-check both the method and that
// path-escaping doesn't double-encode the dotted address.
mux.HandleFunc("/api/reverse-proxies/clusters/proxy.netbird.local", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "DELETE", r.Method, "Delete must use DELETE")
w.WriteHeader(200)
})
err := c.ReverseProxyClusters.Delete(context.Background(), "proxy.netbird.local")
require.NoError(t, err)
})
}
func TestReverseProxyClusters_Delete_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/clusters/proxy.netbird.local", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "Not found", Code: 404})
w.WriteHeader(404)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
err := c.ReverseProxyClusters.Delete(context.Background(), "proxy.netbird.local")
assert.Error(t, err)
})
}
// TestReverseProxyClusters_Delete_EmptyAddress guards against an empty
// clusterAddress reaching the wire — that would collapse the URL onto
// the collection endpoint instead of a specific cluster. The client
// must short-circuit with a typed error before any request is issued.
func TestReverseProxyClusters_Delete_EmptyAddress(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/clusters/", func(http.ResponseWriter, *http.Request) {
t.Fatal("empty clusterAddress must be rejected client-side; no request should reach the server")
})
err := c.ReverseProxyClusters.Delete(context.Background(), "")
assert.Error(t, err, "empty clusterAddress must surface as an error")
})
}
@@ -116,8 +116,8 @@ func TestReverseProxyServices_Create_200(t *testing.T) {
Name: "test-service",
Domain: "test.example.com",
Enabled: true,
Auth: api.ServiceAuthConfig{},
Targets: []api.ServiceTarget{testServiceTarget},
Auth: &api.ServiceAuthConfig{},
Targets: &[]api.ServiceTarget{testServiceTarget},
})
require.NoError(t, err)
assert.Equal(t, testService.Id, ret.Id)
@@ -136,8 +136,8 @@ func TestReverseProxyServices_Create_Err(t *testing.T) {
Name: "test-service",
Domain: "test.example.com",
Enabled: true,
Auth: api.ServiceAuthConfig{},
Targets: []api.ServiceTarget{testServiceTarget},
Auth: &api.ServiceAuthConfig{},
Targets: &[]api.ServiceTarget{testServiceTarget},
})
assert.Error(t, err)
assert.Equal(t, "No", err.Error())
@@ -154,8 +154,9 @@ func TestReverseProxyServices_Create_WithPerTargetOptions(t *testing.T) {
var req api.ServiceRequest
require.NoError(t, json.Unmarshal(reqBytes, &req))
require.Len(t, req.Targets, 1)
target := req.Targets[0]
require.NotNil(t, req.Targets, "targets must be set on the request")
require.Len(t, *req.Targets, 1)
target := (*req.Targets)[0]
require.NotNil(t, target.Options, "options should be present")
opts := target.Options
require.NotNil(t, opts.SkipTlsVerify, "skip_tls_verify should be present")
@@ -177,8 +178,8 @@ func TestReverseProxyServices_Create_WithPerTargetOptions(t *testing.T) {
Name: "test-service",
Domain: "test.example.com",
Enabled: true,
Auth: api.ServiceAuthConfig{},
Targets: []api.ServiceTarget{
Auth: &api.ServiceAuthConfig{},
Targets: &[]api.ServiceTarget{
{
TargetId: "peer-123",
TargetType: "peer",
@@ -216,8 +217,8 @@ func TestReverseProxyServices_Update_200(t *testing.T) {
Name: "updated-service",
Domain: "test.example.com",
Enabled: true,
Auth: api.ServiceAuthConfig{},
Targets: []api.ServiceTarget{testServiceTarget},
Auth: &api.ServiceAuthConfig{},
Targets: &[]api.ServiceTarget{testServiceTarget},
})
require.NoError(t, err)
assert.Equal(t, testService.Id, ret.Id)
@@ -236,8 +237,8 @@ func TestReverseProxyServices_Update_Err(t *testing.T) {
Name: "updated-service",
Domain: "test.example.com",
Enabled: true,
Auth: api.ServiceAuthConfig{},
Targets: []api.ServiceTarget{testServiceTarget},
Auth: &api.ServiceAuthConfig{},
Targets: &[]api.ServiceTarget{testServiceTarget},
})
assert.Error(t, err)
assert.Equal(t, "No", err.Error())
@@ -0,0 +1,79 @@
package rest
import (
"bytes"
"context"
"encoding/json"
"errors"
"net/url"
"github.com/netbirdio/netbird/shared/management/http/api"
)
// ReverseProxyTokensAPI exposes the account-scoped proxy access tokens that
// self-hosted (bring-your-own-proxy) deployments use to register a
// `netbird proxy` instance with management. Tokens are bound to the
// calling account; revoking a token disconnects every proxy that
// authenticated with it.
type ReverseProxyTokensAPI struct {
c *Client
}
// List returns every proxy token the calling account has minted, including
// already-revoked entries. The plain token is never returned — only the
// metadata (id, name, created_at, last_used, revoked).
func (a *ReverseProxyTokensAPI) List(ctx context.Context) ([]api.ProxyToken, error) {
resp, err := a.c.NewRequest(ctx, "GET", "/api/reverse-proxies/proxy-tokens", nil, nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[[]api.ProxyToken](resp)
return ret, err
}
// Create mints a fresh account-scoped proxy token. The returned
// ProxyTokenCreated.PlainToken is shown only once — callers must persist
// it immediately. Subsequent reads will only expose the token metadata,
// not the secret material.
func (a *ReverseProxyTokensAPI) Create(ctx context.Context, request api.ProxyTokenRequest) (*api.ProxyTokenCreated, error) {
requestBytes, err := json.Marshal(request)
if err != nil {
return nil, err
}
resp, err := a.c.NewRequest(ctx, "POST", "/api/reverse-proxies/proxy-tokens", bytes.NewReader(requestBytes), nil)
if err != nil {
return nil, err
}
if resp.Body != nil {
defer resp.Body.Close()
}
ret, err := parseResponse[api.ProxyTokenCreated](resp)
if err != nil {
return nil, err
}
return &ret, nil
}
// Delete revokes a previously-issued proxy token by ID. Revoked tokens
// remain in List output (with revoked=true) so operators can audit which
// credentials existed; the plain secret can no longer authenticate any
// new proxy registration.
func (a *ReverseProxyTokensAPI) Delete(ctx context.Context, tokenID string) error {
// Guard against the empty input: url.PathEscape("") returns "" which
// would collapse the request URL onto the collection endpoint and
// silently delete nothing (or 405 depending on routing).
if tokenID == "" {
return errors.New("tokenID is required")
}
resp, err := a.c.NewRequest(ctx, "DELETE", "/api/reverse-proxies/proxy-tokens/"+url.PathEscape(tokenID), nil, nil)
if err != nil {
return err
}
if resp.Body != nil {
defer resp.Body.Close()
}
return nil
}
@@ -0,0 +1,144 @@
//go:build integration
package rest_test
import (
"context"
"encoding/json"
"io"
"net/http"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/shared/management/client/rest"
"github.com/netbirdio/netbird/shared/management/http/api"
"github.com/netbirdio/netbird/shared/management/http/util"
)
func intPtr(v int) *int { return &v }
var testProxyToken = api.ProxyToken{
Id: "tok-1",
Name: "ci-runner",
CreatedAt: time.Date(2026, 5, 21, 9, 0, 0, 0, time.UTC),
Revoked: false,
}
var testProxyTokenCreated = api.ProxyTokenCreated{
Id: "tok-1",
Name: "ci-runner",
CreatedAt: time.Date(2026, 5, 21, 9, 0, 0, 0, time.UTC),
PlainToken: "nbproxy_abcdef0123456789",
Revoked: false,
}
func TestReverseProxyTokens_List_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/proxy-tokens", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "GET", r.Method, "List must use GET")
retBytes, _ := json.Marshal([]api.ProxyToken{testProxyToken})
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.ReverseProxyTokens.List(context.Background())
require.NoError(t, err)
require.Len(t, ret, 1)
assert.Equal(t, testProxyToken.Id, ret[0].Id)
assert.Equal(t, testProxyToken.Name, ret[0].Name)
})
}
func TestReverseProxyTokens_List_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/proxy-tokens", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "No", Code: 500})
w.WriteHeader(500)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.ReverseProxyTokens.List(context.Background())
assert.Error(t, err)
assert.Empty(t, ret)
})
}
func TestReverseProxyTokens_Create_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/proxy-tokens", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "POST", r.Method, "Create must use POST")
body, err := io.ReadAll(r.Body)
require.NoError(t, err)
var req api.ProxyTokenRequest
require.NoError(t, json.Unmarshal(body, &req), "server must receive a valid ProxyTokenRequest body")
assert.Equal(t, "ci-runner", req.Name, "name must round-trip through the client")
require.NotNil(t, req.ExpiresIn, "expires_in must be sent when provided")
assert.Equal(t, 3600, *req.ExpiresIn, "expires_in value must round-trip unchanged")
retBytes, _ := json.Marshal(testProxyTokenCreated)
_, err = w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.ReverseProxyTokens.Create(context.Background(), api.ProxyTokenRequest{
Name: "ci-runner",
ExpiresIn: intPtr(3600),
})
require.NoError(t, err)
assert.Equal(t, testProxyTokenCreated.Id, ret.Id)
assert.Equal(t, testProxyTokenCreated.PlainToken, ret.PlainToken,
"PlainToken must be returned to the caller — it's the one-shot secret")
})
}
func TestReverseProxyTokens_Create_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/proxy-tokens", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "Bad", Code: 400})
w.WriteHeader(400)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
ret, err := c.ReverseProxyTokens.Create(context.Background(), api.ProxyTokenRequest{Name: ""})
assert.Error(t, err)
assert.Nil(t, ret)
})
}
func TestReverseProxyTokens_Delete_200(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/proxy-tokens/tok-1", func(w http.ResponseWriter, r *http.Request) {
assert.Equal(t, "DELETE", r.Method, "Delete must use DELETE")
w.WriteHeader(200)
})
err := c.ReverseProxyTokens.Delete(context.Background(), "tok-1")
require.NoError(t, err)
})
}
func TestReverseProxyTokens_Delete_Err(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/proxy-tokens/tok-1", func(w http.ResponseWriter, r *http.Request) {
retBytes, _ := json.Marshal(util.ErrorResponse{Message: "Not found", Code: 404})
w.WriteHeader(404)
_, err := w.Write(retBytes)
require.NoError(t, err)
})
err := c.ReverseProxyTokens.Delete(context.Background(), "tok-1")
assert.Error(t, err)
})
}
// TestReverseProxyTokens_Delete_EmptyID guards against an empty tokenID
// reaching the wire — url.PathEscape("") would collapse the URL onto
// the collection endpoint.
func TestReverseProxyTokens_Delete_EmptyID(t *testing.T) {
withMockClient(func(c *rest.Client, mux *http.ServeMux) {
mux.HandleFunc("/api/reverse-proxies/proxy-tokens/", func(http.ResponseWriter, *http.Request) {
t.Fatal("empty tokenID must be rejected client-side; no request should reach the server")
})
err := c.ReverseProxyTokens.Delete(context.Background(), "")
assert.Error(t, err, "empty tokenID must surface as an error")
})
}
@@ -0,0 +1,67 @@
package grpc
import (
"errors"
"fmt"
)
type SyncMessageVersion uint16
const (
Base SyncMessageVersion = iota
ComponentNetworkMap
)
const DefaultSyncMessageVersion = Base
const HighestSyncMessageVersion = ComponentNetworkMap
var ErrorUnrecognizedSyncMessageVersion = errors.New("unrecognized SyncMessageVersion")
func ValidateSyncMessageVersion(v *int) error {
// empty list == we support all available versions
if v == nil {
return nil
}
if *v < 0 || *v > int(HighestSyncMessageVersion) {
return fmt.Errorf("sync message version must between 0 and %d, %w", HighestSyncMessageVersion, ErrorUnrecognizedSyncMessageVersion)
}
return nil
}
// returns SyncMessage version from config, or highest available version if the config is missing or
// base if it is invalid
// the assumption is ValidateSyncMessageVersion() has been called before using SyncMessageVersionFromConfig()
func SyncMessageVersionFromConfig(v *int) SyncMessageVersion {
if v == nil {
return DefaultSyncMessageVersion
}
if *v < 0 || *v > int(HighestSyncMessageVersion) {
return Base
}
return SyncMessageVersion(*v)
}
// convert per-account supported versions to SyncMessageVersion
// the assumption is ValidateSyncMessageVersion() has been called before using SyncMessageVersionsFromMap()
func SyncMessageVersionsFromMap(toconvert map[string]int) map[string]SyncMessageVersion {
// no per-account overrides
if len(toconvert) == 0 {
return nil
}
toret := make(map[string]SyncMessageVersion)
for account, version := range toconvert {
toret[account] = SyncMessageVersionFromConfig(&version)
}
return toret
}
// return highest common sync message version, or Default (which is always available)
func HighestCommonSyncMessageVersion(a SyncMessageVersion, b SyncMessageVersion) SyncMessageVersion {
if a > b {
return b
}
return a
}
@@ -0,0 +1,39 @@
package grpc
import (
"testing"
"github.com/stretchr/testify/assert"
)
func TestValidation(t *testing.T) {
assert.NoError(t, ValidateSyncMessageVersion(nil))
assert.NoError(t, ValidateSyncMessageVersion(toIntPtr(0)))
assert.NoError(t, ValidateSyncMessageVersion(toIntPtr(1)))
assert.ErrorIs(t, ValidateSyncMessageVersion(toIntPtr(int(^uint(0)>>1))), ErrorUnrecognizedSyncMessageVersion)
assert.ErrorIs(t, ValidateSyncMessageVersion(toIntPtr(-1)), ErrorUnrecognizedSyncMessageVersion)
}
func TestVersionFromConfig(t *testing.T) {
assert.Equal(t, DefaultSyncMessageVersion, SyncMessageVersionFromConfig(nil))
assert.Equal(t, Base, SyncMessageVersionFromConfig(toIntPtr(0)))
assert.Equal(t, ComponentNetworkMap, SyncMessageVersionFromConfig(toIntPtr(1)))
assert.Equal(t, DefaultSyncMessageVersion, SyncMessageVersionFromConfig(toIntPtr(-1)))
assert.Equal(t, DefaultSyncMessageVersion, SyncMessageVersionFromConfig(toIntPtr(int(^uint(0)>>1))))
}
func TestPerAccountConversionStringToEnum(t *testing.T) {
assert.Equal(t, map[string]SyncMessageVersion{"1": HighestSyncMessageVersion}, SyncMessageVersionsFromMap(map[string]int{"1": 1}))
assert.Equal(t, map[string]SyncMessageVersion{"2": DefaultSyncMessageVersion}, SyncMessageVersionsFromMap(map[string]int{"2": -1}))
}
func TestCommonVersions(t *testing.T) {
assert.Equal(t, Base, HighestCommonSyncMessageVersion(Base, HighestSyncMessageVersion))
assert.Equal(t, Base, HighestCommonSyncMessageVersion(HighestSyncMessageVersion, Base))
assert.Equal(t, Base, HighestCommonSyncMessageVersion(Base, Base))
assert.Equal(t, HighestSyncMessageVersion, HighestCommonSyncMessageVersion(HighestSyncMessageVersion, HighestSyncMessageVersion))
}
func toIntPtr(v int) *int {
return &v
}
+2 -2
View File
@@ -1,4 +1,4 @@
#!/bin/bash
#!/usr/bin/env bash
set -e
if ! which realpath > /dev/null 2>&1
@@ -11,6 +11,6 @@ fi
old_pwd=$(pwd)
script_path=$(dirname $(realpath "$0"))
cd "$script_path"
go install github.com/oapi-codegen/oapi-codegen/v2/cmd/oapi-codegen@latest
go install github.com/oapi-codegen/oapi-codegen/v2/cmd/oapi-codegen@v2.7.1
oapi-codegen --config cfg.yaml openapi.yml
cd "$old_pwd"
File diff suppressed because it is too large Load Diff
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,23 @@
package integration_reference
import (
"fmt"
"strings"
)
// IntegrationReference holds the reference to a particular integration
type IntegrationReference struct {
ID int
IntegrationType string
}
func (ir IntegrationReference) String() string {
return fmt.Sprintf("%s:%d", ir.IntegrationType, ir.ID)
}
func (ir IntegrationReference) CacheKey(path ...string) string {
if len(path) == 0 {
return ir.String()
}
return fmt.Sprintf("%s:%s", ir.String(), strings.Join(path, ":"))
}
+614
View File
@@ -0,0 +1,614 @@
package networkmap
import (
"context"
"encoding/base64"
"fmt"
"net"
"net/netip"
"slices"
"strconv"
"time"
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/shared/management/domain"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/shared/management/proto"
"github.com/netbirdio/netbird/shared/management/types"
)
// DecodeEnvelope converts a NetworkMapEnvelope into a NetworkMapComponents
// the client can run Calculate() over. Every ID-reference on the wire is a
// xid from corresponding public_id field.
//
// ID scheme on the client side:
//
// Peers base64(wg_pub_key) // stable across snapshots
func DecodeEnvelope(ctx context.Context, env *proto.NetworkMapEnvelope) (*types.NetworkMapComponents, error) {
full := env.GetFull()
if full == nil {
return nil, fmt.Errorf("envelope has no Full payload")
}
c := &types.NetworkMapComponents{
PeerID: "", // engine fills its own peer id from PeerConfig
Network: decodeAccountNetwork(full.Network),
AccountSettings: decodeAccountSettings(full.AccountSettings),
CustomZoneDomain: full.CustomZoneDomain,
Peers: make(map[string]*nmdata.Peer, len(full.Peers)),
Groups: make(map[string]*nmdata.Group, len(full.Groups)),
Policies: make([]*nmdata.Policy, 0, len(full.Policies)),
Routes: make([]*nmdata.Route, 0, len(full.Routes)),
NameServerGroups: make([]*nmdata.NameServerGroup, 0, len(full.NameserverGroups)),
AllDNSRecords: decodeSimpleRecords(full.AllDnsRecords),
AccountZones: decodeCustomZones(full.AccountZones),
ResourcePoliciesMap: make(map[string][]*nmdata.Policy),
RoutersMap: make(map[string]map[string]*nmdata.NetworkRouter),
NetworkResources: make([]*nmdata.NetworkResource, 0, len(full.NetworkResources)),
RouterPeers: make(map[string]*nmdata.Peer),
AllowedUserIDs: stringSliceToSet(full.AllowedUserIds),
PostureFailedPeers: make(map[string]map[string]struct{}, len(full.PostureFailedPeers)),
GroupIDToUserIDs: make(map[string][]string, len(full.GroupIdToUserIds)),
}
if full.DnsSettings != nil {
c.DNSSettings = &nmdata.DNSSettings{
DisabledManagementGroups: full.DnsSettings.DisabledManagementGroupIds,
}
} else {
c.DNSSettings = &nmdata.DNSSettings{}
}
// Phase 1: peers. The envelope's peers slice is index-addressed on the
// wire; we re-key by the peer's WireGuard public key (base64) so the
// in-memory components struct uses a stable identifier across
// snapshots. peerIDByIndex lets downstream phases resolve wire indexes
// back to that key. A peer with a missing or malformed wg_pub_key is
// skipped (and its index keeps "" so any cross-reference falls into the
// same missing-peer branch downstream) — matches legacy behaviour, which
// degrades gracefully rather than aborting the whole sync on a single
// bad row.
peerIDByIndex := make([]string, len(full.Peers))
for idx, pc := range full.Peers {
if pc == nil {
log.Warnf("envelope: peers[%d] is nil, skipping", idx)
continue
}
if len(pc.WgPubKey) != 32 {
log.Warnf("envelope: peers[%d] wg_pub_key length %d (want 32), skipping", idx, len(pc.WgPubKey))
continue
}
peerID := base64.StdEncoding.EncodeToString(pc.WgPubKey)
peer := decodePeerCompact(pc, peerID)
c.Peers[peerID] = peer
peerIDByIndex[idx] = peerID
}
// Phase 2: groups.
for i, gc := range full.Groups {
if gc == nil {
return nil, fmt.Errorf("invalid envelope: groups[%d] is nil", i)
}
groupID := gc.Id
peerIDs := make([]string, 0, len(gc.PeerIndexes))
for _, idx := range gc.PeerIndexes {
if int(idx) < len(peerIDByIndex) {
peerIDs = append(peerIDs, peerIDByIndex[idx])
} else {
log.WithField("peer idx", idx).Error("unrecognized peer idx during decoding")
}
}
fromCompactResources := func() []nmdata.Resource {
var toret []nmdata.Resource
for _, r := range gc.Resources {
res := resourceFromProto(r, peerIDByIndex)
if res == (nmdata.Resource{}) {
log.WithContext(ctx).Warnf("skipping invalid resource in group compact: %s", r.String())
continue
}
toret = append(toret, res)
}
return toret
}
group := &nmdata.Group{
PublicID: gc.Id,
Peers: peerIDs,
Resources: fromCompactResources(),
}
if gc.IsAll {
group.Name = nmdata.GroupAllName
}
c.Groups[groupID] = group
}
// Phase 3: policies (PolicyCompact = one rule per entry; current data
// model is 1 rule per policy).
policyByID := make(map[string]*nmdata.Policy, len(full.Policies))
for i, pc := range full.Policies {
if pc == nil {
return nil, fmt.Errorf("invalid envelope: policies[%d] is nil", i)
}
policy := decodePolicyCompact(pc, pc.Id, peerIDByIndex)
c.Policies = append(c.Policies, policy)
policyByID[pc.Id] = policy
}
// Phase 4: routes.
for i, rr := range full.Routes {
if rr == nil {
return nil, fmt.Errorf("invalid envelope: routes[%d] is nil", i)
}
c.Routes = append(c.Routes, decodeRouteRaw(rr, peerIDByIndex))
}
// Phase 5: NSGs.
for i, nsg := range full.NameserverGroups {
if nsg == nil {
return nil, fmt.Errorf("invalid envelope: nameserver_groups[%d] is nil", i)
}
c.NameServerGroups = append(c.NameServerGroups, decodeNameServerGroupRaw(nsg))
}
// Phase 6: network resources.
for i, nr := range full.NetworkResources {
if nr == nil {
return nil, fmt.Errorf("invalid envelope: network_resources[%d] is nil", i)
}
c.NetworkResources = append(c.NetworkResources, decodeNetworkResource(nr))
}
// Phase 7: routers_map (outer key = network seq id, inner key = peer-id
// reconstructed from peer_index). Synthesized network id is "net_<seq>".
for networkID, list := range full.RoutersMap {
inner := make(map[string]*nmdata.NetworkRouter, len(list.Entries))
for _, entry := range list.Entries {
if !entry.PeerIndexSet {
continue
}
if int(entry.PeerIndex) >= len(peerIDByIndex) {
log.WithField("peer idx", entry.PeerIndex).Error("unrecognized peer id when decoding router map")
continue
}
peerID := peerIDByIndex[entry.PeerIndex]
inner[peerID] = &nmdata.NetworkRouter{
PublicID: entry.Id,
PeerGroups: entry.PeerGroupIds,
Masquerade: entry.Masquerade,
Metric: int(entry.Metric),
Enabled: entry.Enabled,
}
}
if len(inner) > 0 {
c.RoutersMap[networkID] = inner
}
}
// Phase 8: resource_policies_map (resource seq id → list of *types.Policy
// pointers from the decoded policies slice). Resource ID is synthesized
// the same way as in decodeNetworkResource.
for resourceID, ids := range full.ResourcePoliciesMap {
if len(ids.Ids) == 0 {
continue
}
policies := make([]*nmdata.Policy, 0, len(ids.Ids))
for _, id := range ids.Ids {
if p, ok := policyByID[id]; ok {
policies = append(policies, p)
} else {
log.WithField("policy id", id).Error("unrecognized policy when decoding resource policies")
}
}
if len(policies) > 0 {
c.ResourcePoliciesMap[resourceID] = policies
}
}
// Phase 8: rebuild resource_policies_map
for _, r := range c.NetworkResources {
policies := policiesForNetworkResource(r.ID, c.Policies, c.Groups)
if len(policies) == 0 {
continue
}
c.ResourcePoliciesMap[r.ID] = policies
}
// Phase 9: group_id_to_user_ids — wire keys are seq ids, synth to strings.
for groupId, list := range full.GroupIdToUserIds {
c.GroupIDToUserIDs[groupId] = append([]string(nil), list.UserIds...)
}
// Phase 10: posture_failed_peers — wire keys are posture-check seq ids,
// values are peer indexes that need to be turned into peer ids. PolicyRule
// SourcePostureChecks (also synth ids) reference the same key space.
for checkID, set := range full.PostureFailedPeers {
failed := make(map[string]struct{}, len(set.PeerIndexes))
for _, idx := range set.PeerIndexes {
if int(idx) < len(peerIDByIndex) {
failed[peerIDByIndex[idx]] = struct{}{}
} else {
log.WithField("peer idx", idx).Error("unrecognized peer when decoding posture failed peers")
}
}
if len(failed) > 0 {
c.PostureFailedPeers[checkID] = failed
}
}
// Phase 11: router_peer_indexes — peers that act as routers. They're
// already in c.Peers (router peers are appended to the global peers
// list by the encoder); RouterPeers is the subset.
for _, idx := range full.RouterPeerIndexes {
if int(idx) < len(peerIDByIndex) {
peerID := peerIDByIndex[idx]
c.RouterPeers[peerID] = c.Peers[peerID]
}
}
return c, nil
}
func networkResourceGroups(resourceId string, groups map[string]*nmdata.Group) []string {
var toret []string
for _, group := range groups {
for _, resource := range group.Resources {
if resource.ID == resourceId {
toret = append(toret, group.PublicID)
}
}
}
return toret
}
func policiesForNetworkResource(resourceId string, allPolicies []*nmdata.Policy, groups map[string]*nmdata.Group) []*nmdata.Policy {
var toret []*nmdata.Policy
networkResourceGroups := networkResourceGroups(resourceId, groups)
for _, p := range allPolicies {
if p == nil || !p.Enabled || len(p.Rules) == 0 {
continue
}
// there's always only one rule in each policy
if p.Rules[0].DestinationResource.ID == resourceId {
toret = append(toret, p)
continue
}
for _, groupId := range networkResourceGroups {
if slices.Contains(p.Rules[0].Destinations, groupId) {
toret = append(toret, p)
break
}
}
}
return toret
}
// decodeAccountNetwork never returns nil — Calculate() dereferences
// c.Network unconditionally, and servers that predate the fix omit the field
// entirely from the empty-components envelope.
func decodeAccountNetwork(an *proto.AccountNetwork) *nmdata.Network {
n := &nmdata.Network{}
if an == nil {
return n
}
n.Identifier = an.Identifier
n.Dns = an.Dns
n.Serial = int64(an.Serial)
if an.NetCidr != "" {
if _, ipnet, err := net.ParseCIDR(an.NetCidr); err == nil && ipnet != nil {
n.Net = *ipnet
}
}
if an.NetV6Cidr != "" {
if _, ipnet, err := net.ParseCIDR(an.NetV6Cidr); err == nil && ipnet != nil {
n.NetV6 = *ipnet
}
}
return n
}
func decodeAccountSettings(as *proto.AccountSettingsCompact) *nmdata.AccountSettingsInfo {
if as == nil {
return &nmdata.AccountSettingsInfo{}
}
return &nmdata.AccountSettingsInfo{
PeerLoginExpirationEnabled: as.PeerLoginExpirationEnabled,
PeerLoginExpiration: time.Duration(as.PeerLoginExpirationNs),
}
}
func decodePeerCompact(pc *proto.PeerCompact, peerID string) *nmdata.Peer {
var caps []int32
if pc.SupportsSourcePrefixes {
caps = append(caps, nmdata.PeerCapabilitySourcePrefixes)
}
if pc.SupportsIpv6 {
caps = append(caps, nmdata.PeerCapabilityIPv6Overlay)
}
peer := &nmdata.Peer{
ID: peerID,
Key: peerID,
SSHKey: string(pc.SshPubKey),
SSHEnabled: pc.SshEnabled,
DNSLabel: pc.DnsLabel,
LoginExpirationEnabled: pc.LoginExpirationEnabled,
ProxyMeta: nmdata.ProxyMeta{Embedded: pc.ProxyEmbedded},
Meta: nmdata.PeerSystemMeta{
WtVersion: pc.AgentVersion,
Capabilities: caps,
Flags: nmdata.Flags{
ServerSSHAllowed: pc.ServerSshAllowed,
},
},
}
if pc.AddedWithSsoLogin {
// Set a non-empty UserID so (*Peer).AddedWithSSOLogin() returns true.
// The original UserID isn't on the wire; the value is intentionally
// visibly synthetic so any future consumer that mistakes UserID for a
// real account user xid won't silently match (or worse, write the
// sentinel into a downstream record).
peer.UserID = "<env-sso>"
}
if pc.LastLoginUnixNano != 0 {
t := time.Unix(0, pc.LastLoginUnixNano)
peer.LastLogin = &t
}
switch len(pc.Ip) {
case 4:
peer.IP = netip.AddrFrom4([4]byte{pc.Ip[0], pc.Ip[1], pc.Ip[2], pc.Ip[3]})
case 16:
var a [16]byte
copy(a[:], pc.Ip)
peer.IP = netip.AddrFrom16(a)
}
if len(pc.Ipv6) == 16 {
var a [16]byte
copy(a[:], pc.Ipv6)
peer.IPv6 = netip.AddrFrom16(a)
}
return peer
}
func decodePolicyCompact(pc *proto.PolicyCompact, policyID string, peerIDByIndex []string) *nmdata.Policy {
rule := &nmdata.PolicyRule{
ID: policyID, // 1 rule per policy → reuse synthesized id
PolicyID: policyID,
Enabled: true,
Action: string(actionFromProto(pc.Action)),
Protocol: string(protocolFromProto(pc.Protocol)),
Bidirectional: pc.Bidirectional,
Ports: uint32SliceToStrings(pc.Ports),
PortRanges: portRangesFromProto(pc.PortRanges),
Sources: pc.SourceGroupIds,
Destinations: pc.DestinationGroupIds,
AuthorizedUser: pc.AuthorizedUser,
AuthorizedGroups: authorizedGroupsFromProto(pc.AuthorizedGroups),
SourceResource: resourceFromProto(pc.SourceResource, peerIDByIndex),
DestinationResource: resourceFromProto(pc.DestinationResource, peerIDByIndex),
}
return &nmdata.Policy{
ID: policyID,
PublicID: pc.Id,
Enabled: true,
Rules: []*nmdata.PolicyRule{rule},
SourcePostureChecks: pc.SourcePostureCheckIds,
}
}
// resourceFromProto rebuilds types.Resource. For peer-typed resources the
// peer reference is reconstructed from the envelope's peer index — wire
// format ships no xid for peers, so we use the synthesized peer id.
func resourceFromProto(r *proto.ResourceCompact, peerIDByIndex []string) nmdata.Resource {
if r == nil || !types.ResourceType(r.Type).Valid() {
return nmdata.Resource{}
}
if r.Type == string(types.ResourceTypePeer) {
if !r.PeerIndexSet || int(r.PeerIndex) >= len(peerIDByIndex) {
return nmdata.Resource{}
}
return nmdata.Resource{Type: r.Type, ID: peerIDByIndex[int(r.PeerIndex)]}
}
return nmdata.Resource{Type: r.Type, ID: r.Id}
}
// authorizedGroupsFromProto inverts encodeAuthorizedGroups: the wire form
// keys by group account_seq_id, the typed PolicyRule field keys by group
// xid string. We rebuild using the same synthetic scheme the rest of the
// decoder uses ("g<seq>").
func authorizedGroupsFromProto(m map[string]*proto.UserNameList) map[string][]string {
if len(m) == 0 {
return nil
}
out := make(map[string][]string, len(m))
for id, list := range m {
if list == nil {
continue
}
out[id] = append([]string(nil), list.Names...)
}
return out
}
func decodeRouteRaw(rr *proto.RouteRaw, peerIDByIndex []string) *nmdata.Route {
r := &nmdata.Route{
ID: rr.Id,
PublicID: rr.Id,
NetID: rr.NetId,
Description: rr.Description,
Domains: domainsFromPunycode(rr.Domains),
KeepRoute: rr.KeepRoute,
NetworkType: int(rr.NetworkType),
Masquerade: rr.Masquerade,
Metric: int(rr.Metric),
Enabled: rr.Enabled,
Groups: rr.GroupIds,
AccessControlGroups: rr.AccessControlGroupIds,
PeerGroups: rr.PeerGroupIds,
SkipAutoApply: rr.SkipAutoApply,
}
if rr.NetworkCidr != "" {
if p, err := netip.ParsePrefix(rr.NetworkCidr); err == nil {
r.Network = p
}
}
if rr.PeerIndexSet && int(rr.PeerIndex) < len(peerIDByIndex) {
r.Peer = peerIDByIndex[rr.PeerIndex]
}
return r
}
func decodeNameServerGroupRaw(nsg *proto.NameServerGroupRaw) *nmdata.NameServerGroup {
out := &nmdata.NameServerGroup{
ID: nsg.Id,
PublicID: nsg.Id,
Groups: nsg.GroupIds,
Primary: nsg.Primary,
Domains: nsg.Domains,
Enabled: nsg.Enabled,
SearchDomainsEnabled: nsg.SearchDomainsEnabled,
NameServers: make([]nmdata.NameServer, 0, len(nsg.Nameservers)),
}
for _, ns := range nsg.Nameservers {
if addr, err := netip.ParseAddr(ns.IP); err == nil {
out.NameServers = append(out.NameServers, nmdata.NameServer{
IP: addr,
NSType: int(ns.NSType),
Port: int(ns.Port),
})
}
}
return out
}
func decodeNetworkResource(nr *proto.NetworkResourceRaw) *nmdata.NetworkResource {
out := &nmdata.NetworkResource{
ID: nr.Id,
PublicID: nr.Id,
NetworkID: nr.NetworkSeq,
Name: nr.Name,
Description: nr.Description,
Type: nr.Type,
Address: nr.Address,
Domain: nr.DomainValue,
Enabled: nr.Enabled,
}
if nr.PrefixCidr != "" {
if p, err := netip.ParsePrefix(nr.PrefixCidr); err == nil {
out.Prefix = p
}
}
return out
}
func decodeSimpleRecords(records []*proto.SimpleRecord) []nmdata.SimpleRecord {
out := make([]nmdata.SimpleRecord, 0, len(records))
for _, r := range records {
out = append(out, nmdata.SimpleRecord{
Name: r.Name,
Type: int(r.Type),
Class: r.Class,
TTL: int(r.TTL),
RData: r.RData,
})
}
return out
}
func decodeCustomZones(zones []*proto.CustomZone) []nmdata.CustomZone {
out := make([]nmdata.CustomZone, 0, len(zones))
for _, z := range zones {
out = append(out, nmdata.CustomZone{
Domain: z.Domain,
Records: decodeSimpleRecords(z.Records),
SearchDomainDisabled: z.SearchDomainDisabled,
NonAuthoritative: z.NonAuthoritative,
})
}
return out
}
func uint32SliceToStrings(ports []uint32) []string {
if len(ports) == 0 {
return nil
}
out := make([]string, len(ports))
for i, p := range ports {
out[i] = strconv.FormatUint(uint64(p), 10)
}
return out
}
func portRangesFromProto(ranges []*proto.PortInfo_Range) []nmdata.RulePortRange {
if len(ranges) == 0 {
return nil
}
out := make([]nmdata.RulePortRange, 0, len(ranges))
for _, r := range ranges {
if r == nil || r.Start > 65535 || r.End > 65535 {
continue
}
out = append(out, nmdata.RulePortRange{
Start: uint16(r.Start),
End: uint16(r.End),
})
}
return out
}
func actionFromProto(a proto.RuleAction) types.PolicyTrafficActionType {
if a == proto.RuleAction_DROP {
return types.PolicyTrafficActionDrop
}
return types.PolicyTrafficActionAccept
}
func protocolFromProto(p proto.RuleProtocol) types.PolicyRuleProtocolType {
switch p {
case proto.RuleProtocol_TCP:
return types.PolicyRuleProtocolTCP
case proto.RuleProtocol_UDP:
return types.PolicyRuleProtocolUDP
case proto.RuleProtocol_ICMP:
return types.PolicyRuleProtocolICMP
case proto.RuleProtocol_ALL:
return types.PolicyRuleProtocolALL
case proto.RuleProtocol_NETBIRD_SSH:
return types.PolicyRuleProtocolNetbirdSSH
default:
return types.PolicyRuleProtocolALL
}
}
func stringSliceToSet(s []string) map[string]struct{} {
if len(s) == 0 {
return nil
}
out := make(map[string]struct{}, len(s))
for _, v := range s {
out[v] = struct{}{}
}
return out
}
// domainsFromPunycode is a thin wrapper that converts a punycode list back to
// the domain.List type the route.Route struct expects. It accepts the
// punycode strings as-is (no extra decoding) — symmetric with
// route.Domains.ToPunycodeList() used in the encoder.
func domainsFromPunycode(punycoded []string) domain.List {
if len(punycoded) == 0 {
return nil
}
out := make(domain.List, 0, len(punycoded))
for _, d := range punycoded {
out = append(out, domain.Domain(d))
}
return out
}
@@ -0,0 +1,61 @@
package networkmap
import (
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
protobuf "google.golang.org/protobuf/proto"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/shared/management/proto"
)
func TestDecodePolicy(t *testing.T) {
assert.Equal(t,
nmdata.Resource{Type: "peer", ID: "valid-id"},
resourceFromProto(
&proto.ResourceCompact{Type: "peer", PeerIndexSet: true, PeerIndex: uint32(1)},
[]string{"invalid-id-0", "valid-id", "invalid-id-2"}))
// check invalid peer index returns an empty resource
assert.Equal(t,
nmdata.Resource{},
resourceFromProto(
&proto.ResourceCompact{Type: "peer", PeerIndexSet: true, PeerIndex: uint32(100)},
[]string{"invalid-id-0", "valid-id", "invalid-id-2"}))
assert.Equal(t,
nmdata.Resource{Type: "domain", ID: "domain"},
resourceFromProto(
&proto.ResourceCompact{Type: "domain", Id: "domain"}, []string{}))
assert.Equal(t,
nmdata.Resource{Type: "host", ID: "host"},
resourceFromProto(
&proto.ResourceCompact{Type: "host", Id: "host"}, []string{}))
assert.Equal(t,
nmdata.Resource{Type: "subnet", ID: "subnet"},
resourceFromProto(
&proto.ResourceCompact{Type: "subnet", Id: "subnet"}, []string{}))
// an unknown resource type return an empty resource
assert.Equal(t,
nmdata.Resource{},
resourceFromProto(
&proto.ResourceCompact{Type: "boom", Id: "boom"}, []string{}))
}
// ResourceCompact fields 1-3 are the v0.77 wire contract. Retyping any of them
// makes peers on either side of the change silently drop policy resources, so
// the encoding is pinned here as raw bytes: field 1 "peer" (bytes), field 2
// true (varint), field 3 7 (varint).
func TestResourceCompactLegacyWireFormat(t *testing.T) {
legacy := []byte{0x0a, 0x04, 'p', 'e', 'e', 'r', 0x10, 0x01, 0x18, 0x07}
var decoded proto.ResourceCompact
require.NoError(t, protobuf.Unmarshal(legacy, &decoded))
assert.Equal(t, "peer", decoded.Type)
assert.True(t, decoded.PeerIndexSet)
assert.Equal(t, uint32(7), decoded.PeerIndex)
encoded, err := protobuf.Marshal(&proto.ResourceCompact{Type: "peer", PeerIndexSet: true, PeerIndex: 7})
require.NoError(t, err)
assert.Equal(t, legacy, encoded)
}
+337
View File
@@ -0,0 +1,337 @@
// Package networkmap contains the shared NetworkMap helpers that both the
// management server and the client agent need.
//
// The proto-conversion helpers (types.NetworkMap → proto.NetworkMap) live
// here so the client can run the same conversion locally after deriving its
// NetworkMap from a NetworkMapEnvelope, without taking a dependency on the
// server-side conversion package (which pulls in cloud integrations and is
// otherwise an unwanted internal import on the client).
//
// The helpers are pure functions over inputs — no caches, no IO, no logging
// beyond a context-aware error log when an individual user-id hash fails.
package networkmap
import (
"context"
log "github.com/sirupsen/logrus"
goproto "google.golang.org/protobuf/proto"
"net/netip"
nbdns "github.com/netbirdio/netbird/dns"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/shared/management/proto"
"github.com/netbirdio/netbird/shared/management/types"
"github.com/netbirdio/netbird/shared/netiputil"
"github.com/netbirdio/netbird/shared/sshauth"
)
// ToProtocolRoutes converts a slice of typed routes to their proto form.
func ToProtocolRoutes(routes []*nmdata.Route) []*proto.Route {
protoRoutes := make([]*proto.Route, 0, len(routes))
for _, r := range routes {
protoRoutes = append(protoRoutes, ToProtocolRoute(r))
}
return protoRoutes
}
// ToProtocolRoute converts one typed route to its proto form.
func ToProtocolRoute(route *nmdata.Route) *proto.Route {
return &proto.Route{
ID: string(route.ID),
NetID: string(route.NetID),
Network: route.Network.String(),
Domains: route.Domains.ToPunycodeList(),
NetworkType: int64(route.NetworkType),
Peer: route.Peer,
Metric: int64(route.Metric),
Masquerade: route.Masquerade,
KeepRoute: route.KeepRoute,
SkipAutoApply: route.SkipAutoApply,
}
}
// ToProtocolFirewallRules converts the firewall rules to the protocol form.
// When useSourcePrefixes is true, the compact SourcePrefixes field is
// populated alongside the deprecated PeerIP for forward compatibility.
// Wildcard rules ("0.0.0.0") are expanded into separate v4/v6 SourcePrefixes
// when includeIPv6 is true.
func ToProtocolFirewallRules(rules []*types.FirewallRule, includeIPv6, useSourcePrefixes bool) []*proto.FirewallRule {
result := make([]*proto.FirewallRule, 0, len(rules))
for i := range rules {
rule := rules[i]
fwRule := &proto.FirewallRule{
PolicyID: []byte(rule.PolicyID),
PeerIP: rule.PeerIP, //nolint:staticcheck // populated for backward compatibility
Direction: GetProtoDirection(rule.Direction),
Action: GetProtoAction(rule.Action),
Protocol: GetProtoProtocol(rule.Protocol),
Port: rule.Port,
}
if useSourcePrefixes && rule.PeerIP != "" {
result = append(result, populateSourcePrefixes(fwRule, rule, includeIPv6)...)
}
if ShouldUsePortRange(fwRule) {
fwRule.PortInfo = rule.PortRange.ToProto()
}
result = append(result, fwRule)
}
return result
}
// populateSourcePrefixes sets SourcePrefixes on fwRule and returns any
// additional rules needed (e.g. a v6 wildcard clone when the peer IP is
// unspecified).
func populateSourcePrefixes(fwRule *proto.FirewallRule, rule *types.FirewallRule, includeIPv6 bool) []*proto.FirewallRule {
addr, err := netip.ParseAddr(rule.PeerIP)
if err != nil {
return nil
}
if !addr.IsUnspecified() {
fwRule.SourcePrefixes = [][]byte{netiputil.EncodeAddr(addr.Unmap())}
return nil
}
v4Wildcard, _ := netiputil.EncodePrefix(netip.PrefixFrom(netip.IPv4Unspecified(), 0))
fwRule.SourcePrefixes = [][]byte{v4Wildcard}
if !includeIPv6 {
return nil
}
v6Rule := goproto.Clone(fwRule).(*proto.FirewallRule)
v6Rule.PeerIP = "::" //nolint:staticcheck // populated for backward compatibility
v6Wildcard, _ := netiputil.EncodePrefix(netip.PrefixFrom(netip.IPv6Unspecified(), 0))
v6Rule.SourcePrefixes = [][]byte{v6Wildcard}
if ShouldUsePortRange(v6Rule) {
v6Rule.PortInfo = rule.PortRange.ToProto()
}
return []*proto.FirewallRule{v6Rule}
}
// GetProtoDirection converts the direction to proto.RuleDirection.
func GetProtoDirection(direction int) proto.RuleDirection {
if direction == types.FirewallRuleDirectionOUT {
return proto.RuleDirection_OUT
}
return proto.RuleDirection_IN
}
// GetProtoAction converts the action to proto.RuleAction.
func GetProtoAction(action string) proto.RuleAction {
if action == string(types.PolicyTrafficActionDrop) {
return proto.RuleAction_DROP
}
return proto.RuleAction_ACCEPT
}
// GetProtoProtocol converts the protocol to proto.RuleProtocol.
func GetProtoProtocol(protocol string) proto.RuleProtocol {
switch types.PolicyRuleProtocolType(protocol) {
case types.PolicyRuleProtocolALL:
return proto.RuleProtocol_ALL
case types.PolicyRuleProtocolTCP:
return proto.RuleProtocol_TCP
case types.PolicyRuleProtocolUDP:
return proto.RuleProtocol_UDP
case types.PolicyRuleProtocolICMP:
return proto.RuleProtocol_ICMP
case types.PolicyRuleProtocolNetbirdSSH:
return proto.RuleProtocol_NETBIRD_SSH
default:
return proto.RuleProtocol_UNKNOWN
}
}
// GetProtoPortInfo converts route-firewall-rule port info to proto.PortInfo.
func GetProtoPortInfo(rule *types.RouteFirewallRule) *proto.PortInfo {
var portInfo proto.PortInfo
if rule.Port != 0 {
portInfo.PortSelection = &proto.PortInfo_Port{Port: uint32(rule.Port)}
} else if portRange := rule.PortRange; portRange.Start != 0 && portRange.End != 0 {
portInfo.PortSelection = &proto.PortInfo_Range_{
Range: &proto.PortInfo_Range{
Start: uint32(portRange.Start),
End: uint32(portRange.End),
},
}
}
return &portInfo
}
// ShouldUsePortRange reports whether the firewall rule should use a port
// range rather than a single port (TCP/UDP without a single port).
func ShouldUsePortRange(rule *proto.FirewallRule) bool {
return rule.Port == "" && (rule.Protocol == proto.RuleProtocol_UDP || rule.Protocol == proto.RuleProtocol_TCP)
}
// ToProtocolRoutesFirewallRules converts a slice of typed route-firewall
// rules to proto.
func ToProtocolRoutesFirewallRules(rules []*types.RouteFirewallRule) []*proto.RouteFirewallRule {
result := make([]*proto.RouteFirewallRule, len(rules))
for i := range rules {
rule := rules[i]
result[i] = &proto.RouteFirewallRule{
SourceRanges: rule.SourceRanges,
Action: GetProtoAction(rule.Action),
Destination: rule.Destination,
Protocol: GetProtoProtocol(rule.Protocol),
PortInfo: GetProtoPortInfo(rule),
IsDynamic: rule.IsDynamic,
Domains: rule.Domains.ToPunycodeList(),
PolicyID: []byte(rule.PolicyID),
RouteID: string(rule.RouteID),
}
}
return result
}
// ConvertToProtoCustomZone converts an nbdns.CustomZone to its proto form.
func ConvertToProtoCustomZone(zone nbdns.CustomZone) *proto.CustomZone {
protoZone := &proto.CustomZone{
Domain: zone.Domain,
Records: make([]*proto.SimpleRecord, 0, len(zone.Records)),
SearchDomainDisabled: zone.SearchDomainDisabled,
NonAuthoritative: zone.NonAuthoritative,
}
for _, record := range zone.Records {
protoZone.Records = append(protoZone.Records, &proto.SimpleRecord{
Name: record.Name,
Type: int64(record.Type),
Class: record.Class,
TTL: int64(record.TTL),
RData: record.RData,
})
}
return protoZone
}
// ConvertToProtoNameServerGroup converts a NameServerGroup to its proto form.
func ConvertToProtoNameServerGroup(nsGroup *nbdns.NameServerGroup) *proto.NameServerGroup {
protoGroup := &proto.NameServerGroup{
Primary: nsGroup.Primary,
Domains: nsGroup.Domains,
SearchDomainsEnabled: nsGroup.SearchDomainsEnabled,
NameServers: make([]*proto.NameServer, 0, len(nsGroup.NameServers)),
}
for _, ns := range nsGroup.NameServers {
protoGroup.NameServers = append(protoGroup.NameServers, &proto.NameServer{
IP: ns.IP.String(),
Port: int64(ns.Port),
NSType: int64(ns.NSType),
})
}
return protoGroup
}
// DNSConfigCache is the cache contract for amortising NameServerGroup
// proto-conversion across peers in the same account. Server uses a concrete
// implementation; client passes nil (no cross-peer caching needed when
// rebuilding a single NetworkMap from an envelope).
type DNSConfigCache interface {
GetNameServerGroup(key string) (*proto.NameServerGroup, bool)
SetNameServerGroup(key string, value *proto.NameServerGroup)
}
// ToProtocolDNSConfig converts nbdns.Config to proto.DNSConfig. If cache is
// non-nil, NameServerGroup proto values are cached by NSG.ID across calls —
// the server amortises this across peers, the client passes nil.
func ToProtocolDNSConfig(update nbdns.Config, cache DNSConfigCache, forwardPort int64) *proto.DNSConfig {
protoUpdate := &proto.DNSConfig{
ServiceEnable: update.ServiceEnable,
CustomZones: make([]*proto.CustomZone, 0, len(update.CustomZones)),
NameServerGroups: make([]*proto.NameServerGroup, 0, len(update.NameServerGroups)),
ForwarderPort: forwardPort, //nolint:staticcheck
}
for _, zone := range update.CustomZones {
protoUpdate.CustomZones = append(protoUpdate.CustomZones, ConvertToProtoCustomZone(zone))
}
for _, nsGroup := range update.NameServerGroups {
if cache != nil {
if cachedGroup, exists := cache.GetNameServerGroup(nsGroup.ID); exists {
protoUpdate.NameServerGroups = append(protoUpdate.NameServerGroups, cachedGroup)
continue
}
}
protoGroup := ConvertToProtoNameServerGroup(nsGroup)
if cache != nil {
cache.SetNameServerGroup(nsGroup.ID, protoGroup)
}
protoUpdate.NameServerGroups = append(protoUpdate.NameServerGroups, protoGroup)
}
return protoUpdate
}
// AppendRemotePeerConfig appends typed peers as proto.RemotePeerConfig
// entries to dst and returns the result. localIsProxy reports whether the peer
// receiving this config is itself an embedded proxy.
func AppendRemotePeerConfig(dst []*proto.RemotePeerConfig, peers []*nmdata.Peer, dnsName string, includeIPv6 bool, localIsProxy bool) []*proto.RemotePeerConfig {
for _, rPeer := range peers {
allowedIPs := []string{rPeer.IP.String() + "/32"}
if includeIPv6 && rPeer.IPv6.IsValid() {
allowedIPs = append(allowedIPs, rPeer.IPv6.String()+"/128")
}
dst = append(dst, &proto.RemotePeerConfig{
WgPubKey: rPeer.Key,
AllowedIps: allowedIPs,
SshConfig: &proto.SSHConfig{SshPubKey: []byte(rPeer.SSHKey)},
Fqdn: rPeer.FQDN(dnsName),
AgentVersion: rPeer.Meta.WtVersion,
LazyState: lazyStateFor(localIsProxy, rPeer),
})
}
return dst
}
// lazyStateFor returns the per-peer lazy override for a remote peer. Connections
// involving an ephemeral proxy peer on either endpoint default to lazy so shared
// proxy infrastructure is not kept permanently connected to every peer. All
// other peers follow the account-wide flag. A future admin-facing per-peer
// setting can return LazyStateEager here to force a peer always-active.
func lazyStateFor(localIsProxy bool, rPeer *nmdata.Peer) proto.LazyState {
if localIsProxy || rPeer.ProxyMeta.Embedded {
return proto.LazyState_LazyStateLazy
}
return proto.LazyState_LazyStateDefault
}
// BuildAuthorizedUsersProto deduplicates user-IDs into a hashed list and
// builds per-machine-user index maps. Returns (hashedUsers, machineUsers).
// Errors from individual hash failures are logged via the provided context;
// they leave the offending user out of the result but don't abort the build.
func BuildAuthorizedUsersProto(ctx context.Context, authorizedUsers map[string]map[string]struct{}) ([][]byte, map[string]*proto.MachineUserIndexes) {
userIDToIndex := make(map[string]uint32)
var hashedUsers [][]byte
machineUsers := make(map[string]*proto.MachineUserIndexes, len(authorizedUsers))
for machineUser, users := range authorizedUsers {
indexes := make([]uint32, 0, len(users))
for userID := range users {
idx, exists := userIDToIndex[userID]
if !exists {
hash, err := sshauth.HashUserID(userID)
if err != nil {
log.WithContext(ctx).WithError(err).Error("failed to hash user id")
continue
}
idx = uint32(len(hashedUsers))
userIDToIndex[userID] = idx
hashedUsers = append(hashedUsers, hash[:])
}
indexes = append(indexes, idx)
}
machineUsers[machineUser] = &proto.MachineUserIndexes{Indexes: indexes}
}
return hashedUsers, machineUsers
}
+189
View File
@@ -0,0 +1,189 @@
package networkmap
import (
"context"
"encoding/base64"
"fmt"
"github.com/netbirdio/netbird/shared/management/proto"
"github.com/netbirdio/netbird/shared/management/types"
)
// EnvelopeResult is what the client engine consumes after receiving a
// component-format NetworkMap. Both fields are populated:
//
// - NetworkMap is the *proto.NetworkMap shape the engine reads today via
// update.GetNetworkMap() — built from the envelope's components by
// running Calculate() locally + converting back through the shared
// proto helpers + merging the optional ProxyPatch.
// - Components is the *types.NetworkMapComponents the engine retains so
// future incremental delta updates have a base to apply changes
// against. The client keeps it under its sync lock.
type EnvelopeResult struct {
NetworkMap *proto.NetworkMap
Components *types.NetworkMapComponents
}
// EnvelopeToNetworkMap is the full client-side pipeline: decode the
// component envelope back to a typed NetworkMapComponents, run Calculate()
// locally to produce the typed NetworkMap, convert it to the wire form the
// engine consumes, and fold in any ProxyPatch the server attached.
//
// localPeerKey is the receiving peer's WG pub key (used to derive
// includeIPv6 / useSourcePrefixes from the receiving peer's own record in
// the components struct, mirroring legacy ToSyncResponse behaviour).
//
// dnsName is the account's DNS domain ("netbird.cloud" etc.); used when
// rebuilding the per-peer FQDNs that proto.RemotePeerConfig carries.
func EnvelopeToNetworkMap(ctx context.Context, env *proto.NetworkMapEnvelope, localPeerKey, dnsName string) (*EnvelopeResult, error) {
components, err := DecodeEnvelope(ctx, env)
if err != nil {
return nil, fmt.Errorf("decode envelope: %w", err)
}
// Find the receiving peer in the decoded components by WG key.
// c.Peers is keyed by canonical base64 of the raw 32-byte pub key
// (decoder re-encodes the bytes off the wire). The caller may pass a
// non-canonical encoding (some persisted production keys carry
// non-zero trailing padding bits that survived a legacy import), so
// round-trip through raw bytes once to canonicalize before lookup.
canonicalKey := canonicalizeWgKey(localPeerKey)
localPeer := components.Peers[canonicalKey]
if localPeer == nil {
return nil, fmt.Errorf("receiving peer (wg_key prefix %q) not found among %d decoded peers — components have no PeerID, Calculate would return empty", trimKey(localPeerKey), len(components.Peers))
}
components.PeerID = canonicalKey
includeIPv6 := localPeer.SupportsIPv6() && localPeer.IPv6.IsValid()
useSourcePrefixes := localPeer.SupportsSourcePrefixes()
typedNM := components.Calculate(ctx)
full := env.GetFull()
dnsFwdPort := int64(0)
if full != nil {
dnsFwdPort = full.DnsForwarderPort
}
protoNM := &proto.NetworkMap{
Serial: typedNM.Network.CurrentSerial(),
}
if full != nil {
protoNM.PeerConfig = full.PeerConfig
}
protoNM.Routes = ToProtocolRoutes(typedNM.Routes)
protoNM.DNSConfig = ToProtocolDNSConfig(typedNM.DNSConfig, nil, dnsFwdPort)
remotePeers := AppendRemotePeerConfig(nil, typedNM.Peers, dnsName, includeIPv6, localPeer.ProxyMeta.Embedded)
protoNM.RemotePeers = remotePeers
protoNM.RemotePeersIsEmpty = len(remotePeers) == 0
protoNM.OfflinePeers = AppendRemotePeerConfig(nil, typedNM.OfflinePeers, dnsName, includeIPv6, localPeer.ProxyMeta.Embedded)
firewallRules := ToProtocolFirewallRules(typedNM.FirewallRules, includeIPv6, useSourcePrefixes)
protoNM.FirewallRules = firewallRules
protoNM.FirewallRulesIsEmpty = len(firewallRules) == 0
routesFirewallRules := ToProtocolRoutesFirewallRules(typedNM.RoutesFirewallRules)
protoNM.RoutesFirewallRules = routesFirewallRules
protoNM.RoutesFirewallRulesIsEmpty = len(routesFirewallRules) == 0
if typedNM.AuthorizedUsers != nil {
hashedUsers, machineUsers := BuildAuthorizedUsersProto(ctx, typedNM.AuthorizedUsers)
userIDClaim := ""
if full != nil {
userIDClaim = full.UserIdClaim
}
protoNM.SshAuth = &proto.SSHAuth{
AuthorizedUsers: hashedUsers,
MachineUsers: machineUsers,
UserIDClaim: userIDClaim,
}
}
if typedNM.ForwardingRules != nil {
forwardingRules := make([]*proto.ForwardingRule, 0, len(typedNM.ForwardingRules))
for _, rule := range typedNM.ForwardingRules {
forwardingRules = append(forwardingRules, rule.ToProto())
}
protoNM.ForwardingRules = forwardingRules
}
// Merge the proxy patch the server attached. Mirrors the legacy
// NetworkMap.Merge step that the server runs after Calculate().
if full != nil && full.ProxyPatch != nil {
mergeProxyPatch(protoNM, full.ProxyPatch)
}
return &EnvelopeResult{
NetworkMap: protoNM,
Components: components,
}, nil
}
// mergeProxyPatch folds a ProxyPatch's pre-expanded fragments into the
// proto.NetworkMap that Calculate() produced. Mirrors types.NetworkMap.Merge
// — same six collections, deduplicated where the legacy merge dedupes.
func mergeProxyPatch(nm *proto.NetworkMap, patch *proto.ProxyPatch) {
nm.RemotePeers = appendUniquePeers(nm.RemotePeers, patch.Peers)
nm.OfflinePeers = appendUniquePeers(nm.OfflinePeers, patch.OfflinePeers)
nm.FirewallRules = append(nm.FirewallRules, patch.FirewallRules...)
nm.Routes = append(nm.Routes, patch.Routes...)
nm.RoutesFirewallRules = append(nm.RoutesFirewallRules, patch.RouteFirewallRules...)
nm.ForwardingRules = append(nm.ForwardingRules, patch.ForwardingRules...)
if len(nm.RemotePeers) > 0 {
nm.RemotePeersIsEmpty = false
}
if len(nm.FirewallRules) > 0 {
nm.FirewallRulesIsEmpty = false
}
if len(nm.RoutesFirewallRules) > 0 {
nm.RoutesFirewallRulesIsEmpty = false
}
}
// appendUniquePeers dedupes by WgPubKey — mirrors legacy
// mergeUniquePeersByID's intent (legacy keyed off Peer.ID; in proto form the
// closest stable identifier is WgPubKey).
func appendUniquePeers(dst, extra []*proto.RemotePeerConfig) []*proto.RemotePeerConfig {
if len(extra) == 0 {
return dst
}
seen := make(map[string]struct{}, len(dst))
for _, p := range dst {
if p == nil {
continue
}
seen[p.WgPubKey] = struct{}{}
}
for _, p := range extra {
if p == nil {
continue
}
if _, ok := seen[p.WgPubKey]; ok {
continue
}
seen[p.WgPubKey] = struct{}{}
dst = append(dst, p)
}
return dst
}
func trimKey(s string) string {
if len(s) > 12 {
return s[:12]
}
return s
}
// canonicalizeWgKey normalises a base64-encoded WireGuard public key so it
// matches the canonical encoding emitted by the envelope decoder. Returns
// the input unchanged when it does not decode to 32 raw bytes (caller will
// hit a miss in the peer map and surface the error).
func canonicalizeWgKey(s string) string {
raw, err := base64.StdEncoding.DecodeString(s)
if err != nil || len(raw) != 32 {
return s
}
return base64.StdEncoding.EncodeToString(raw)
}
@@ -0,0 +1,355 @@
package networkmap_test
import (
"context"
"crypto/rand"
"encoding/base64"
"fmt"
"net"
"net/netip"
"testing"
"github.com/stretchr/testify/require"
goproto "google.golang.org/protobuf/proto"
mgmtgrpc "github.com/netbirdio/netbird/management/internals/shared/grpc"
"github.com/netbirdio/netbird/management/server/types"
nbnetworkmap "github.com/netbirdio/netbird/shared/management/networkmap"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/shared/management/proto"
)
// TestEnvelopeToNetworkMap_RoundTrip exercises the full client-side pipeline:
// build a small components struct, encode an envelope, marshal/unmarshal the
// wire bytes, decode back via EnvelopeToNetworkMap, and verify the result is
// non-empty and consistent.
func TestEnvelopeToNetworkMap_RoundTrip(t *testing.T) {
c, localPeerKey := buildSmokeComponents(t)
envelope := mgmtgrpc.EncodeNetworkMapEnvelope(mgmtgrpc.ComponentsEnvelopeInput{
Components: c,
DNSDomain: "netbird.cloud",
})
wire, err := goproto.Marshal(envelope)
require.NoError(t, err, "marshal envelope")
var decoded proto.NetworkMapEnvelope
require.NoError(t, goproto.Unmarshal(wire, &decoded), "unmarshal envelope")
result, err := nbnetworkmap.EnvelopeToNetworkMap(context.Background(), &decoded, localPeerKey, "netbird.cloud")
require.NoError(t, err, "EnvelopeToNetworkMap")
require.NotNil(t, result)
require.NotNil(t, result.NetworkMap, "decoded NetworkMap must be non-nil")
require.NotNil(t, result.Components, "Components must be retained for future delta updates")
require.NotNil(t, result.Components.AccountSettings)
require.NotEmpty(t, result.NetworkMap.RemotePeers, "two-peer allow policy should produce one remote peer")
require.NotEmpty(t, result.NetworkMap.FirewallRules, "two-peer allow policy should produce firewall rules")
}
// TestCalculate_FirewallRuleProtocol_NeverNetbirdSSH guards against the
// scenario where a rule with Protocol=NetbirdSSH leaks the enum value into
// proto.FirewallRule.Protocol. Calculate() must rewrite NetbirdSSH → TCP
// before forming firewall rules. Without that rewrite, agents fall into
// UNKNOWN-protocol handling, which on some platforms downgrades to
// allow-all — a real security regression.
func TestCalculate_FirewallRuleProtocol_NeverNetbirdSSH(t *testing.T) {
c, localPeerKey := buildSmokeComponents(t)
// Replace the smoke policy with a NetbirdSSH-protocol allow.
c.Policies = []*nmdata.Policy{{
ID: "pol-ssh", PublicID: "2", Enabled: true,
Rules: []*nmdata.PolicyRule{{
ID: "rule-ssh",
Enabled: true,
Action: string(types.PolicyTrafficActionAccept),
Protocol: string(types.PolicyRuleProtocolNetbirdSSH),
Bidirectional: true,
Sources: []string{"group-all"},
Destinations: []string{"group-all"},
}},
}}
envelope := mgmtgrpc.EncodeNetworkMapEnvelope(mgmtgrpc.ComponentsEnvelopeInput{
Components: c,
DNSDomain: "netbird.cloud",
})
wire, err := goproto.Marshal(envelope)
require.NoError(t, err)
var decoded proto.NetworkMapEnvelope
require.NoError(t, goproto.Unmarshal(wire, &decoded))
result, err := nbnetworkmap.EnvelopeToNetworkMap(context.Background(), &decoded, localPeerKey, "netbird.cloud")
require.NoError(t, err)
require.NotEmpty(t, result.NetworkMap.FirewallRules, "ssh policy should produce firewall rules")
for i, fr := range result.NetworkMap.FirewallRules {
require.NotEqualf(t, proto.RuleProtocol_NETBIRD_SSH, fr.Protocol,
"FirewallRules[%d].Protocol must be the rewritten TCP, not NETBIRD_SSH", i)
}
}
func TestEnvelopeToNetworkMap_NilEnvelope(t *testing.T) {
_, err := nbnetworkmap.EnvelopeToNetworkMap(context.Background(), nil, "key", "netbird.cloud")
require.Error(t, err, "nil envelope must produce an error rather than panic")
}
func TestEnvelopeToNetworkMap_FullPayloadMissing(t *testing.T) {
env := &proto.NetworkMapEnvelope{}
_, err := nbnetworkmap.EnvelopeToNetworkMap(context.Background(), env, "key", "netbird.cloud")
require.Error(t, err, "envelope with no Full payload must produce an error")
}
// TestDecodeEnvelope_MalformedWgKeyPeerSkipped feeds an envelope where one
// peer has a wg_pub_key that is not 32 bytes long. The decoder must skip
// that peer (keeping the rest of the snapshot usable) instead of aborting
// the whole sync — mirrors legacy behaviour that tolerates an occasional
// bad row.
func TestDecodeEnvelope_MalformedWgKeyPeerSkipped(t *testing.T) {
c, localPeerKey := buildSmokeComponents(t)
envelope := mgmtgrpc.EncodeNetworkMapEnvelope(mgmtgrpc.ComponentsEnvelopeInput{
Components: c,
DNSDomain: "netbird.cloud",
})
require.NotNil(t, envelope.GetFull())
full := envelope.GetFull()
require.Len(t, full.Peers, 2, "smoke fixture should have two peers")
// Truncate the second peer's wg_pub_key so it fails the length gate.
for _, p := range full.Peers {
if base64.StdEncoding.EncodeToString(p.WgPubKey) != localPeerKey {
p.WgPubKey = p.WgPubKey[:31]
}
}
wire, err := goproto.Marshal(envelope)
require.NoError(t, err, "marshal envelope")
var decoded proto.NetworkMapEnvelope
require.NoError(t, goproto.Unmarshal(wire, &decoded), "unmarshal envelope")
result, err := nbnetworkmap.EnvelopeToNetworkMap(context.Background(), &decoded, localPeerKey, "netbird.cloud")
require.NoError(t, err, "EnvelopeToNetworkMap must tolerate one bad peer key")
require.NotNil(t, result)
require.NotNil(t, result.Components)
require.Len(t, result.Components.Peers, 1, "the well-formed peer survives, the malformed one is dropped")
}
// TestEnvelopeRoundTrip_AllGroupShortCircuitParity reproduces prod accounts
// with several groups literally named "All" where the "All"-named group does
// not contain every peer. Server-side Calculate short-circuits destination
// expansion at the first group named "All" (getUniquePeerIDsFromGroupsIDs),
// ignoring the remaining destination groups. The wire must preserve enough
// group identity for the decoded components to short-circuit identically —
// otherwise the client unions all destination groups and emits extra
// firewall rules the server never produced.
func TestEnvelopeRoundTrip_AllGroupShortCircuitParity(t *testing.T) {
ctx := context.Background()
peers := map[string]*nmdata.Peer{}
for i, id := range []string{"peer-T", "peer-S", "peer-ALL", "peer-O"} {
peers[id] = &nmdata.Peer{
ID: id,
Key: randomWgKey(t),
IP: netip.AddrFrom4([4]byte{100, 64, 0, byte(i + 1)}),
DNSLabel: id,
Meta: nmdata.PeerSystemMeta{WtVersion: "0.40.0"},
}
}
c := &types.NetworkMapComponents{
PeerID: "peer-T",
Network: &nmdata.Network{
Identifier: "net-all-groups",
Net: net.IPNet{IP: net.IP{100, 64, 0, 0}, Mask: net.CIDRMask(10, 32)},
Serial: 1,
},
AccountSettings: &nmdata.AccountSettingsInfo{},
DNSSettings: &nmdata.DNSSettings{},
Peers: peers,
Groups: map[string]*nmdata.Group{
"g-src": {PublicID: "1", Name: "staff", Peers: []string{"peer-T", "peer-S"}},
"g-all": {PublicID: "2", Name: "All", Peers: []string{"peer-ALL"}},
"g-two": {PublicID: "3", Name: "second", Peers: []string{"peer-T", "peer-O"}},
},
Policies: []*nmdata.Policy{{
ID: "pol-multi-dest", PublicID: "10", Enabled: true,
Rules: []*nmdata.PolicyRule{{
ID: "rule-multi-dest",
Enabled: true,
Action: string(types.PolicyTrafficActionAccept),
Protocol: string(types.PolicyRuleProtocolALL),
Sources: []string{"g-src"},
Destinations: []string{"g-all", "g-two"},
}},
}},
}
serverNM := c.Calculate(ctx)
require.NotNil(t, serverNM)
envelope := mgmtgrpc.EncodeNetworkMapEnvelope(mgmtgrpc.ComponentsEnvelopeInput{
Components: c,
DNSDomain: "netbird.cloud",
})
wire, err := goproto.Marshal(envelope)
require.NoError(t, err, "marshal envelope")
var decodedEnv proto.NetworkMapEnvelope
require.NoError(t, goproto.Unmarshal(wire, &decodedEnv), "unmarshal envelope")
result, err := nbnetworkmap.EnvelopeToNetworkMap(ctx, &decodedEnv, peers["peer-T"].Key, "netbird.cloud")
require.NoError(t, err, "EnvelopeToNetworkMap")
clientNM := result.NetworkMap
serverRules := make([]string, 0, len(serverNM.FirewallRules))
for _, r := range serverNM.FirewallRules {
serverRules = append(serverRules, fmt.Sprintf("%s/%d", r.PeerIP, r.Direction))
}
clientRules := make([]string, 0, len(clientNM.FirewallRules))
for _, r := range clientNM.FirewallRules {
clientRules = append(clientRules, fmt.Sprintf("%s/%d", r.PeerIP, r.Direction)) // nolint:staticcheck
}
require.ElementsMatch(t, serverRules, clientRules,
"client-side Calculate must expand destination groups exactly like the server")
serverPeers := make([]string, 0, len(serverNM.Peers))
for _, p := range serverNM.Peers {
serverPeers = append(serverPeers, p.Key)
}
clientPeers := make([]string, 0, len(clientNM.RemotePeers))
for _, p := range clientNM.RemotePeers {
clientPeers = append(clientPeers, p.WgPubKey)
}
require.ElementsMatch(t, serverPeers, clientPeers,
"client-side Calculate must connect the same remote peers as the server")
}
// TestEnvelopeToNetworkMap_EmptyComponents covers the graceful-degrade path
// the server takes for a peer that is missing from the account or absent from
// the validated-peers map. The legacy server short-circuited before
// Calculate() and shipped a NetworkMap carrying only the account Network; the
// components path runs Calculate() on the client instead, so the envelope must
// carry Network or the client panics dereferencing a nil *types.Network.
func TestEnvelopeToNetworkMap_EmptyComponents(t *testing.T) {
localPeerKey := randomWgKey(t)
c := types.EmptyNetworkMapComponents(&types.NetworkMapComponents{
PeerID: "peer-A",
Network: &nmdata.Network{
Identifier: "net-empty",
Net: net.IPNet{IP: net.IP{100, 64, 0, 0}, Mask: net.CIDRMask(10, 32)},
Serial: 7,
},
Peers: map[string]*nmdata.Peer{
"peer-A": {ID: "peer-A", Key: localPeerKey, IP: netip.AddrFrom4([4]byte{100, 64, 0, 1})},
},
})
envelope := mgmtgrpc.EncodeNetworkMapEnvelope(mgmtgrpc.ComponentsEnvelopeInput{
Components: c,
DNSDomain: "netbird.cloud",
})
require.NotNil(t, envelope.GetFull().Network, "empty envelope must carry the account Network")
wire, err := goproto.Marshal(envelope)
require.NoError(t, err, "marshal envelope")
var decoded proto.NetworkMapEnvelope
require.NoError(t, goproto.Unmarshal(wire, &decoded), "unmarshal envelope")
result, err := nbnetworkmap.EnvelopeToNetworkMap(context.Background(), &decoded, localPeerKey, "netbird.cloud")
require.NoError(t, err, "EnvelopeToNetworkMap must degrade gracefully on empty components")
require.Equal(t, uint64(7), result.NetworkMap.Serial)
require.Empty(t, result.NetworkMap.RemotePeers, "unvalidated peer connects to nobody")
}
// TestEnvelopeToNetworkMap_MissingNetwork simulates a server that omits
// AccountNetwork from the envelope. Clients must degrade rather than panic, so
// they survive talking to a management server that predates the encoder fix.
func TestEnvelopeToNetworkMap_MissingNetwork(t *testing.T) {
c, localPeerKey := buildSmokeComponents(t)
envelope := mgmtgrpc.EncodeNetworkMapEnvelope(mgmtgrpc.ComponentsEnvelopeInput{
Components: c,
DNSDomain: "netbird.cloud",
})
envelope.GetFull().Network = nil
wire, err := goproto.Marshal(envelope)
require.NoError(t, err, "marshal envelope")
var decoded proto.NetworkMapEnvelope
require.NoError(t, goproto.Unmarshal(wire, &decoded), "unmarshal envelope")
result, err := nbnetworkmap.EnvelopeToNetworkMap(context.Background(), &decoded, localPeerKey, "netbird.cloud")
require.NoError(t, err, "a missing AccountNetwork must not panic the client")
require.NotNil(t, result.Components.Network)
require.NotEmpty(t, result.NetworkMap.RemotePeers, "the rest of the snapshot stays usable")
}
// buildSmokeComponents returns a minimal NetworkMapComponents (2 peers, 1
// group, 1 allow policy) plus the receiving peer's WG public key. Sufficient
// to validate the encode → marshal → decode → Calculate pipeline produces
// non-empty output.
func buildSmokeComponents(t *testing.T) (*types.NetworkMapComponents, string) {
t.Helper()
peerAKey := randomWgKey(t)
peerBKey := randomWgKey(t)
peerA := &nmdata.Peer{
ID: "peer-A",
Key: peerAKey,
IP: netip.AddrFrom4([4]byte{100, 64, 0, 1}),
DNSLabel: "peerA",
Meta: nmdata.PeerSystemMeta{WtVersion: "0.40.0"},
}
peerB := &nmdata.Peer{
ID: "peer-B",
Key: peerBKey,
IP: netip.AddrFrom4([4]byte{100, 64, 0, 2}),
DNSLabel: "peerB",
Meta: nmdata.PeerSystemMeta{WtVersion: "0.40.0"},
}
group := &nmdata.Group{
PublicID: "1", Name: "All",
Peers: []string{"peer-A", "peer-B"},
}
policy := &nmdata.Policy{
ID: "pol-allow", PublicID: "1", Enabled: true,
Rules: []*nmdata.PolicyRule{{
ID: "rule-allow",
Enabled: true,
Action: string(types.PolicyTrafficActionAccept),
Protocol: string(types.PolicyRuleProtocolALL),
Bidirectional: true,
Sources: []string{"group-all"},
Destinations: []string{"group-all"},
}},
}
c := &types.NetworkMapComponents{
PeerID: "peer-A",
Network: &nmdata.Network{
Identifier: "net-smoke",
Net: net.IPNet{IP: net.IP{100, 64, 0, 0}, Mask: net.CIDRMask(10, 32)},
Serial: 1,
},
AccountSettings: &nmdata.AccountSettingsInfo{},
DNSSettings: &nmdata.DNSSettings{},
Peers: map[string]*nmdata.Peer{
"peer-A": peerA,
"peer-B": peerB,
},
Groups: map[string]*nmdata.Group{
"group-all": group,
},
Policies: []*nmdata.Policy{policy},
}
return c, peerAKey
}
func randomWgKey(t *testing.T) string {
t.Helper()
var raw [32]byte
_, err := rand.Read(raw[:])
require.NoError(t, err)
return base64.StdEncoding.EncodeToString(raw[:])
}
@@ -0,0 +1,815 @@
package networkmap
import (
"slices"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/shared/management/types"
)
type sshRequirements struct {
neededGroupIDs map[string]struct{}
needAllowedUserIDs bool
}
// GetPeerNetworkMapComponents computes the peer's NetworkMapComponents from the
// slim twin store. It mirrors the former Account.GetPeerNetworkMapComponents
// exactly, operating on nmdata twins throughout — no Account reference and no
// twin↔real conversion, since the produced components hold twins.
func (nmd *NetworkMapData) GetPeerNetworkMapComponents(peerID string, peersCustomZone nmdata.CustomZone) *types.NetworkMapComponents {
nmd.InjectProxyPolicies()
forceRoutingPeerDNS := nmd.forcesRoutingPeerDNSResolution(peerID)
peer := nmd.Peers[peerID]
if peer == nil {
return types.EmptyNetworkMapComponents(&types.NetworkMapComponents{
PeerID: peerID,
Network: nmd.Network,
Peers: map[string]*nmdata.Peer{peerID: peer},
ForceRoutingPeerDNSResolution: forceRoutingPeerDNS,
})
}
if _, ok := nmd.ValidatedPeers[peerID]; !ok {
return types.EmptyNetworkMapComponents(&types.NetworkMapComponents{
PeerID: peerID,
Network: nmd.Network,
Peers: map[string]*nmdata.Peer{peerID: peer},
ForceRoutingPeerDNSResolution: forceRoutingPeerDNS,
})
}
components := &types.NetworkMapComponents{
PeerID: peerID,
Network: nmd.Network,
AccountSettings: nmd.AccountSettings,
DNSSettings: nmd.DNSSettings,
CustomZoneDomain: peersCustomZone.Domain,
NameServerGroups: make([]*nmdata.NameServerGroup, 0),
ResourcePoliciesMap: make(map[string][]*nmdata.Policy),
RoutersMap: make(map[string]map[string]*nmdata.NetworkRouter),
NetworkResources: make([]*nmdata.NetworkResource, 0),
PostureFailedPeers: make(map[string]map[string]struct{}, len(nmd.PostureChecks)),
RouterPeers: make(map[string]*nmdata.Peer),
NetworkXIDToPublicID: nmd.NetworkXIDToPublicID,
PostureCheckXIDToPublicID: nmd.PostureCheckXIDToPublicID,
ForceRoutingPeerDNSResolution: forceRoutingPeerDNS,
}
relevantPeers, relevantGroups, relevantPolicies, relevantRoutes, sshReqs := nmd.getPeersGroupsPoliciesRoutes(peerID, peer.SSHEnabled, &components.PostureFailedPeers)
if len(sshReqs.neededGroupIDs) > 0 {
components.GroupIDToUserIDs = filterGroupIDToUserIDs(nmd.GroupIDToUserIDs, sshReqs.neededGroupIDs)
}
if sshReqs.needAllowedUserIDs {
components.AllowedUserIDs = nmd.getAllowedUserIDs()
}
components.Peers = relevantPeers
components.Groups = relevantGroups
components.Policies = relevantPolicies
components.Routes = relevantRoutes
components.AllDNSRecords = filterDNSRecordsByPeers(peersCustomZone.Records, relevantPeers, peer.SupportsIPv6() && peer.IPv6.IsValid())
peerGroups := nmd.GetPeerGroups(peerID)
components.AccountZones = nmd.appliedZones(peerGroups)
components.AccountZones = append(components.AccountZones, nmd.privateServiceZones(peerGroups)...)
for _, nsGroup := range nmd.NameServerGroups {
if nsGroup != nil && nsGroup.Enabled {
for _, gID := range nsGroup.Groups {
if _, found := relevantGroups[gID]; found {
components.NameServerGroups = append(components.NameServerGroups, nsGroup)
break
}
}
}
}
for _, resource := range nmd.NetworkResources {
if resource == nil || !resource.Enabled {
continue
}
policies, exists := nmd.ResourcePolicies[resource.ID]
if !exists {
continue
}
addSourcePeers := false
networkRoutingPeers, routerExists := nmd.Routers[resource.NetworkID]
if routerExists {
if _, ok := networkRoutingPeers[peerID]; ok {
addSourcePeers = true
}
}
for _, policy := range policies {
if policy == nil || !policy.Enabled || len(policy.Rules) == 0 || policy.Rules[0] == nil {
continue
}
if addSourcePeers {
var peers []string
if policy.Rules[0].SourceResource.Type == string(types.ResourceTypePeer) && policy.Rules[0].SourceResource.ID != "" {
peers = []string{policy.Rules[0].SourceResource.ID}
} else {
peers = nmd.getUniquePeerIDsFromGroupsIDs(policy.SourceGroups())
}
for _, pID := range nmd.getPostureValidPeersSaveFailed(peers, policy.SourcePostureChecks, &components.PostureFailedPeers) {
if _, exists := components.Peers[pID]; !exists {
components.Peers[pID] = nmd.Peers[pID]
}
}
} else {
peerInSources := false
if policy.Rules[0].SourceResource.Type == string(types.ResourceTypePeer) && policy.Rules[0].SourceResource.ID != "" {
peerInSources = policy.Rules[0].SourceResource.ID == peerID
} else {
for _, groupID := range policy.SourceGroups() {
if group := nmd.Groups[groupID]; group != nil && slices.Contains(group.Peers, peerID) {
peerInSources = true
break
}
}
}
if !peerInSources {
continue
}
isValid, pname := nmd.validatePostureChecksOnPeerGetFailed(policy.SourcePostureChecks, peerID)
if !isValid && len(pname) > 0 {
if _, ok := components.PostureFailedPeers[pname]; !ok {
components.PostureFailedPeers[pname] = make(map[string]struct{})
}
components.PostureFailedPeers[pname][peer.ID] = struct{}{}
continue
}
addSourcePeers = true
}
for _, rule := range policy.Rules {
if rule == nil || !rule.Enabled {
continue
}
for _, srcGroupID := range rule.Sources {
if g := nmd.Groups[srcGroupID]; g != nil {
if _, exists := components.Groups[srcGroupID]; !exists {
components.Groups[srcGroupID] = g
}
}
}
for _, dstGroupID := range rule.Destinations {
if g := nmd.Groups[dstGroupID]; g != nil {
if _, exists := components.Groups[dstGroupID]; !exists {
components.Groups[dstGroupID] = g
}
}
}
}
components.ResourcePoliciesMap[resource.ID] = policies
}
if addSourcePeers {
components.RoutersMap[resource.NetworkID] = networkRoutingPeers
for peerIDKey := range networkRoutingPeers {
p := nmd.Peers[peerIDKey]
if p == nil {
continue
}
// An unapproved peer must not carry traffic, so it is kept out of
// RouterPeers as well: the envelope encoder indexes that map into
// the wire peer table, from which the client restores every entry.
if _, validated := nmd.ValidatedPeers[peerIDKey]; !validated {
continue
}
if _, exists := components.RouterPeers[peerIDKey]; !exists {
components.RouterPeers[peerIDKey] = p
}
if _, exists := components.Peers[peerIDKey]; !exists {
components.Peers[peerIDKey] = p
}
}
components.NetworkResources = append(components.NetworkResources, resource)
}
}
filterGroupPeers(&components.Groups, components.Peers)
filterPostureFailedPeers(&components.PostureFailedPeers, components.Policies, components.ResourcePoliciesMap, components.Peers)
return components
}
func (nmd *NetworkMapData) getPeersGroupsPoliciesRoutes(
peerID string,
peerSSHEnabled bool,
postureFailedPeers *map[string]map[string]struct{},
) (map[string]*nmdata.Peer, map[string]*nmdata.Group, []*nmdata.Policy, []*nmdata.Route, sshRequirements) {
relevantPeerIDs := make(map[string]*nmdata.Peer, len(nmd.Peers)/4)
relevantGroupIDs := make(map[string]*nmdata.Group, len(nmd.Groups)/4)
relevantPolicies := make([]*nmdata.Policy, 0, len(nmd.Policies))
relevantRoutes := make([]*nmdata.Route, 0, len(nmd.Routes))
sshReqs := sshRequirements{neededGroupIDs: make(map[string]struct{})}
relevantPeerIDs[peerID] = nmd.Peers[peerID]
peerGroupSet := nmd.GetPeerGroups(peerID)
for groupID := range peerGroupSet {
relevantGroupIDs[groupID] = nmd.Groups[groupID]
}
routeAccessControlGroups := make(map[string]struct{})
for _, r := range nmd.Routes {
if r == nil {
continue
}
relevant := r.Peer == peerID
if !relevant {
for _, groupID := range r.PeerGroups {
if _, ok := peerGroupSet[groupID]; ok {
relevant = true
break
}
}
}
if !relevant && r.Enabled {
for _, groupID := range r.Groups {
if _, ok := peerGroupSet[groupID]; ok {
relevant = true
break
}
}
}
if !relevant {
continue
}
for _, groupID := range r.PeerGroups {
if g := nmd.Groups[groupID]; g != nil {
relevantGroupIDs[groupID] = g
}
}
for _, groupID := range r.Groups {
if g := nmd.Groups[groupID]; g != nil {
relevantGroupIDs[groupID] = g
}
}
if r.Enabled {
for _, groupID := range r.AccessControlGroups {
if g := nmd.Groups[groupID]; g != nil {
relevantGroupIDs[groupID] = g
}
routeAccessControlGroups[groupID] = struct{}{}
}
}
if r.Peer != "" {
if _, ok := nmd.ValidatedPeers[r.Peer]; ok {
if p := nmd.Peers[r.Peer]; p != nil {
relevantPeerIDs[r.Peer] = p
}
}
}
for _, groupID := range r.PeerGroups {
g := nmd.Groups[groupID]
if g == nil {
continue
}
for _, pid := range g.Peers {
if _, exists := relevantPeerIDs[pid]; exists {
continue
}
if _, ok := nmd.ValidatedPeers[pid]; !ok {
continue
}
if p := nmd.Peers[pid]; p != nil {
relevantPeerIDs[pid] = p
}
}
}
relevantRoutes = append(relevantRoutes, r)
}
for _, policy := range nmd.Policies {
if policy == nil || !policy.Enabled {
continue
}
policyRelevant := false
for _, rule := range policy.Rules {
if rule == nil || !rule.Enabled {
continue
}
if len(routeAccessControlGroups) > 0 {
for _, destGroupID := range rule.Destinations {
if _, needed := routeAccessControlGroups[destGroupID]; needed {
policyRelevant = true
for _, srcGroupID := range rule.Sources {
if g := nmd.Groups[srcGroupID]; g != nil {
relevantGroupIDs[srcGroupID] = g
}
}
for _, dstGroupID := range rule.Destinations {
if g := nmd.Groups[dstGroupID]; g != nil {
relevantGroupIDs[dstGroupID] = g
}
}
break
}
}
}
var sourcePeers, destinationPeers []string
var peerInSources, peerInDestinations bool
if rule.SourceResource.Type == string(types.ResourceTypePeer) && rule.SourceResource.ID != "" {
sourcePeers, peerInSources = nmd.getPeerFromResource(rule.SourceResource, peerID, policy.SourcePostureChecks, postureFailedPeers)
} else {
sourcePeers, peerInSources = nmd.getPeersFromGroups(rule.Sources, peerID, policy.SourcePostureChecks, postureFailedPeers)
}
if rule.DestinationResource.Type == string(types.ResourceTypePeer) && rule.DestinationResource.ID != "" {
destinationPeers, peerInDestinations = nmd.getPeerFromResource(rule.DestinationResource, peerID, nil, postureFailedPeers)
} else {
destinationPeers, peerInDestinations = nmd.getPeersFromGroups(rule.Destinations, peerID, nil, postureFailedPeers)
}
if peerInSources {
policyRelevant = true
for _, pid := range destinationPeers {
relevantPeerIDs[pid] = nmd.Peers[pid]
}
for _, dstGroupID := range rule.Destinations {
if g := nmd.Groups[dstGroupID]; g != nil {
relevantGroupIDs[dstGroupID] = g
}
}
}
if peerInDestinations {
policyRelevant = true
for _, pid := range sourcePeers {
relevantPeerIDs[pid] = nmd.Peers[pid]
}
for _, srcGroupID := range rule.Sources {
if g := nmd.Groups[srcGroupID]; g != nil {
relevantGroupIDs[srcGroupID] = g
}
}
if rule.Protocol == string(types.PolicyRuleProtocolNetbirdSSH) {
switch {
case len(rule.AuthorizedGroups) > 0:
for groupID := range rule.AuthorizedGroups {
sshReqs.neededGroupIDs[groupID] = struct{}{}
}
case rule.AuthorizedUser != "":
default:
sshReqs.needAllowedUserIDs = true
}
} else if nmdata.PolicyRuleImpliesLegacySSH(rule) && peerSSHEnabled {
sshReqs.needAllowedUserIDs = true
}
}
}
if policyRelevant {
relevantPolicies = append(relevantPolicies, policy)
}
}
return relevantPeerIDs, relevantGroupIDs, relevantPolicies, relevantRoutes, sshReqs
}
func (nmd *NetworkMapData) getPeersFromGroups(groups []string, peerID string, sourcePostureChecksIDs []string,
postureFailedPeers *map[string]map[string]struct{}) ([]string, bool) {
peerInGroups := false
filteredPeerIDs := make([]string, 0, len(groups))
seenPeerIds := make(map[string]struct{}, len(groups))
for _, gid := range groups {
group := nmd.Groups[gid]
if group == nil {
continue
}
if group.IsGroupAll() || len(groups) == 1 {
filteredPeerIDs = make([]string, 0, len(group.Peers))
peerInGroups = false
for _, pid := range group.Peers {
if !nmd.admitPolicyPeer(pid, sourcePostureChecksIDs, postureFailedPeers) {
continue
}
if pid == peerID {
peerInGroups = true
continue
}
filteredPeerIDs = append(filteredPeerIDs, pid)
}
return filteredPeerIDs, peerInGroups
}
for _, pid := range group.Peers {
if _, seen := seenPeerIds[pid]; seen {
continue
}
seenPeerIds[pid] = struct{}{}
if !nmd.admitPolicyPeer(pid, sourcePostureChecksIDs, postureFailedPeers) {
continue
}
if pid == peerID {
peerInGroups = true
continue
}
filteredPeerIDs = append(filteredPeerIDs, pid)
}
}
return filteredPeerIDs, peerInGroups
}
// getPeerFromResource resolves a rule side that names a peer directly, admitting it
// like a member of a group holding only that peer.
func (nmd *NetworkMapData) getPeerFromResource(resource nmdata.Resource, peerID string, sourcePostureChecksIDs []string,
postureFailedPeers *map[string]map[string]struct{}) ([]string, bool) {
if !nmd.admitPolicyPeer(resource.ID, sourcePostureChecksIDs, postureFailedPeers) {
return nil, false
}
if resource.ID == peerID {
return nil, true
}
return []string{resource.ID}, false
}
// admitPolicyPeer applies the per-peer admission of a rule side: the peer must exist,
// be validated and pass the rule's posture checks. A failed check is recorded in
// postureFailedPeers.
func (nmd *NetworkMapData) admitPolicyPeer(pid string, sourcePostureChecksIDs []string, postureFailedPeers *map[string]map[string]struct{}) bool {
peer, ok := nmd.Peers[pid]
if !ok || peer == nil {
return false
}
if _, ok := nmd.ValidatedPeers[pid]; !ok {
return false
}
isValid, pname := nmd.validatePostureChecksOnPeerGetFailed(sourcePostureChecksIDs, pid)
if !isValid && len(pname) > 0 {
if _, ok := (*postureFailedPeers)[pname]; !ok {
(*postureFailedPeers)[pname] = make(map[string]struct{})
}
(*postureFailedPeers)[pname][pid] = struct{}{}
return false
}
return true
}
func (nmd *NetworkMapData) validatePostureChecksOnPeerGetFailed(sourcePostureChecksID []string, peerID string) (bool, string) {
peer, ok := nmd.Peers[peerID]
if !ok || peer == nil {
return false, ""
}
for _, postureChecksID := range sourcePostureChecksID {
if valid, cached := nmd.cachedPostureCheckResult(postureChecksID, peerID); cached {
if !valid {
return false, postureChecksID
}
continue
}
postureChecks := nmd.PostureChecks[postureChecksID]
if postureChecks == nil {
continue
}
if !postureChecks.Passes(peer) {
return false, postureChecksID
}
}
return true, ""
}
func (nmd *NetworkMapData) PrecomputePostureValidation() {
if len(nmd.PostureChecks) == 0 {
nmd.PostureValidation = nil
return
}
checkPeerIDs := make(map[string]map[string]struct{})
for _, policy := range nmd.Policies {
if policy == nil || !policy.Enabled || len(policy.SourcePostureChecks) == 0 {
continue
}
groupPeerIDs := nmd.getUniquePeerIDsFromGroupsIDs(policy.SourceGroups())
for _, postureChecksID := range policy.SourcePostureChecks {
set := checkPeerIDs[postureChecksID]
if set == nil {
set = make(map[string]struct{}, len(groupPeerIDs))
checkPeerIDs[postureChecksID] = set
}
for _, pid := range groupPeerIDs {
set[pid] = struct{}{}
}
for _, rule := range policy.Rules {
if rule == nil {
continue
}
if rule.SourceResource.Type == string(types.ResourceTypePeer) && rule.SourceResource.ID != "" {
set[rule.SourceResource.ID] = struct{}{}
}
}
}
}
results := make(map[string]map[string]bool, len(checkPeerIDs))
for postureChecksID, peerIDs := range checkPeerIDs {
results[postureChecksID] = nmd.evaluatePostureChecksForPeers(postureChecksID, peerIDs)
}
nmd.PostureValidation = results
}
func (nmd *NetworkMapData) evaluatePostureChecksForPeers(postureChecksID string, peerIDs map[string]struct{}) map[string]bool {
postureChecks := nmd.PostureChecks[postureChecksID]
if postureChecks == nil {
return nil
}
checks := postureChecks.GetChecks()
results := make(map[string]bool, len(peerIDs))
for peerID := range peerIDs {
peer := nmd.Peers[peerID]
if peer == nil {
continue
}
results[peerID] = nmdata.PassesChecks(checks, peer)
}
return results
}
func (nmd *NetworkMapData) cachedPostureCheckResult(postureChecksID, peerID string) (bool, bool) {
results, ok := nmd.PostureValidation[postureChecksID]
if !ok {
return false, false
}
if results == nil {
return true, true
}
valid, found := results[peerID]
return valid, found
}
func (nmd *NetworkMapData) getPostureValidPeersSaveFailed(inputPeers []string, postureChecksIDs []string, postureFailedPeers *map[string]map[string]struct{}) []string {
var dest []string
for _, peerID := range inputPeers {
if _, validated := nmd.ValidatedPeers[peerID]; !validated {
continue
}
valid, pname := nmd.validatePostureChecksOnPeerGetFailed(postureChecksIDs, peerID)
if valid {
dest = append(dest, peerID)
continue
}
if pname == "" {
continue
}
if _, ok := (*postureFailedPeers)[pname]; !ok {
(*postureFailedPeers)[pname] = make(map[string]struct{})
}
(*postureFailedPeers)[pname][peerID] = struct{}{}
}
return dest
}
// forcesRoutingPeerDNSResolution reports whether the given peer must run
// routing-peer DNS resolution regardless of the account-global
// RoutingPeerDNSResolutionEnabled setting: true when the peer routes a domain
// network resource targeted by an enabled reverse-proxy service, so the peer's
// DNS forwarder starts and can resolve the target for the embedded proxy peers.
func (nmd *NetworkMapData) forcesRoutingPeerDNSResolution(peerID string) bool {
if len(nmd.ProxyTargetedDomainResourceIDs) == 0 {
return false
}
for _, resource := range nmd.NetworkResources {
if resource == nil || !resource.Enabled || resource.Type != string(types.ResourceTypeDomain) {
continue
}
if _, ok := nmd.ProxyTargetedDomainResourceIDs[resource.ID]; !ok {
continue
}
if _, isRouter := nmd.Routers[resource.NetworkID][peerID]; isRouter {
return true
}
}
return false
}
// GetPeerGroups returns the set of group IDs the peer belongs to. The
// underlying peer→groups index is built once per NetworkMapData and the
// returned set is shared — callers must not mutate it.
func (nmd *NetworkMapData) GetPeerGroups(peerID string) map[string]struct{} {
nmd.peerGroupsOnce.Do(func() {
idx := make(map[string]map[string]struct{}, len(nmd.Peers))
for groupID, group := range nmd.Groups {
if group == nil {
continue
}
for _, pid := range group.Peers {
set, ok := idx[pid]
if !ok {
set = make(map[string]struct{})
idx[pid] = set
}
set[groupID] = struct{}{}
}
}
nmd.peerGroupsIdx = idx
})
if set, ok := nmd.peerGroupsIdx[peerID]; ok {
return set
}
return map[string]struct{}{}
}
func (nmd *NetworkMapData) getUniquePeerIDsFromGroupsIDs(groups []string) []string {
peerIDs := make(map[string]struct{}, len(groups))
for _, groupID := range groups {
group := nmd.Groups[groupID]
if group == nil {
continue
}
if group.IsGroupAll() || len(groups) == 1 {
return group.Peers
}
for _, peerID := range group.Peers {
peerIDs[peerID] = struct{}{}
}
}
ids := make([]string, 0, len(peerIDs))
for peerID := range peerIDs {
ids = append(ids, peerID)
}
return ids
}
func (nmd *NetworkMapData) getAllowedUserIDs() map[string]struct{} {
return nmd.AllowedUserIDs
}
func (nmd *NetworkMapData) appliedZones(peerGroups map[string]struct{}) []nmdata.CustomZone {
if len(peerGroups) == 0 {
return nil
}
var out []nmdata.CustomZone
for _, cand := range nmd.AppliedZoneCandidates {
if peerInDistributionGroups(peerGroups, cand.DistributionGroups) {
out = append(out, cand.Zone)
}
}
return out
}
func (nmd *NetworkMapData) privateServiceZones(peerGroups map[string]struct{}) []nmdata.CustomZone {
byApex := make(map[string]*nmdata.CustomZone)
var order []string
for _, cand := range nmd.PrivateServiceCandidates {
if !peerInDistributionGroups(peerGroups, cand.AccessGroups) {
continue
}
zone, exists := byApex[cand.Zone.Domain]
if !exists {
nz := nmdata.CustomZone{
Domain: cand.Zone.Domain,
SearchDomainDisabled: cand.Zone.SearchDomainDisabled,
NonAuthoritative: cand.Zone.NonAuthoritative,
}
byApex[cand.Zone.Domain] = &nz
zone = &nz
order = append(order, cand.Zone.Domain)
}
zone.Records = append(zone.Records, cand.Zone.Records...)
}
var out []nmdata.CustomZone
for _, apex := range order {
zone := byApex[apex]
if len(zone.Records) == 0 {
continue
}
out = append(out, *zone)
}
return out
}
func peerInDistributionGroups(peerGroups map[string]struct{}, groups []string) bool {
for _, g := range groups {
if _, ok := peerGroups[g]; ok {
return true
}
}
return false
}
func filterGroupPeers(groups *map[string]*nmdata.Group, peers map[string]*nmdata.Peer) {
for groupID, groupInfo := range *groups {
filteredPeers := make([]string, 0, len(groupInfo.Peers))
for _, pid := range groupInfo.Peers {
if _, exists := peers[pid]; exists {
filteredPeers = append(filteredPeers, pid)
}
}
if len(filteredPeers) != len(groupInfo.Peers) {
ng := groupInfo.Copy()
ng.Peers = filteredPeers
(*groups)[groupID] = ng
}
}
}
func filterPostureFailedPeers(postureFailedPeers *map[string]map[string]struct{}, policies []*nmdata.Policy, resourcePoliciesMap map[string][]*nmdata.Policy, peers map[string]*nmdata.Peer) {
if len(*postureFailedPeers) == 0 {
return
}
referencedPostureChecks := make(map[string]struct{})
for _, policy := range policies {
for _, checkID := range policy.SourcePostureChecks {
referencedPostureChecks[checkID] = struct{}{}
}
}
for _, resPolicies := range resourcePoliciesMap {
for _, policy := range resPolicies {
for _, checkID := range policy.SourcePostureChecks {
referencedPostureChecks[checkID] = struct{}{}
}
}
}
for checkID, failedPeers := range *postureFailedPeers {
if _, referenced := referencedPostureChecks[checkID]; !referenced {
delete(*postureFailedPeers, checkID)
continue
}
for peerID := range failedPeers {
if _, exists := peers[peerID]; !exists {
delete(failedPeers, peerID)
}
}
if len(failedPeers) == 0 {
delete(*postureFailedPeers, checkID)
}
}
}
func filterDNSRecordsByPeers(records []nmdata.SimpleRecord, peers map[string]*nmdata.Peer, includeIPv6 bool) []nmdata.SimpleRecord {
if len(records) == 0 || len(peers) == 0 {
return nil
}
peerIPs := make(map[string]struct{}, len(peers)*2)
for _, peer := range peers {
if peer == nil {
continue
}
peerIPs[peer.IP.String()] = struct{}{}
if includeIPv6 && peer.IPv6.IsValid() {
peerIPs[peer.IPv6.String()] = struct{}{}
}
}
filteredRecords := make([]nmdata.SimpleRecord, 0, len(records))
for _, record := range records {
if _, exists := peerIPs[record.RData]; exists {
filteredRecords = append(filteredRecords, record)
}
}
return filteredRecords
}
func filterGroupIDToUserIDs(fullMap map[string][]string, neededGroupIDs map[string]struct{}) map[string][]string {
if len(neededGroupIDs) == 0 {
return nil
}
filtered := make(map[string][]string, len(neededGroupIDs))
for groupID := range neededGroupIDs {
if users, ok := fullMap[groupID]; ok {
filtered[groupID] = users
}
}
return filtered
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,83 @@
package networkmap
import (
"sync"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
// NetworkMapData is a dependency-light, slim twin of the server Account. It
// carries only the state GetPeerNetworkMapComponents needs, expressed in the
// fresh nmdata twin types. A builder converts an Account into a NetworkMapData
// once per account; the per-peer components calculation then runs on this twin
// with no reference back to the Account.
type NetworkMapData struct { //nolint:revive // established name across the codebase
Peers map[string]*nmdata.Peer
Groups map[string]*nmdata.Group
Policies []*nmdata.Policy
Routes []*nmdata.Route
NameServerGroups []*nmdata.NameServerGroup
NetworkResources []*nmdata.NetworkResource
Network *nmdata.Network
DNSSettings *nmdata.DNSSettings
AccountSettings *nmdata.AccountSettingsInfo
PostureChecks map[string]*nmdata.PostureChecks
// PostureValidation holds the precomputed posture-check results, keyed by
// posture check ID then peer ID. Filled by PrecomputePostureValidation; a
// present but nil inner map marks a check ID that resolves to no posture
// check, which the calc treats as passing.
PostureValidation map[string]map[string]bool
AllowedUserIDs map[string]struct{}
NetworkXIDToPublicID map[string]string
PostureCheckXIDToPublicID map[string]string
ValidatedPeers map[string]struct{}
ResourcePolicies map[string][]*nmdata.Policy
Routers map[string]map[string]*nmdata.NetworkRouter
GroupIDToUserIDs map[string][]string
DNSDomain string
// ProxyTargetedDomainResourceIDs is the account-level half of
// forcesRoutingPeerDNSResolution: domain network resources targeted by an
// enabled reverse-proxy service.
ProxyTargetedDomainResourceIDs map[string]struct{}
AppliedZoneCandidates []AppliedZoneCandidate
PrivateServiceCandidates []PrivateServiceCandidate
// Services are the account's reverse-proxy services, persisted ones and
// the in-memory ones synthesised from agent-network state. They are the
// source of the proxy ACLs injectProxyPolicies synthesises, which no
// builder can load because they are never written to the database.
Services []*nmdata.Service
// Domains are the account's registered reverse-proxy domains, used to
// resolve the zone apex a private service's records hang under.
Domains []nmdata.ProxyDomain
peerGroupsOnce sync.Once
peerGroupsIdx map[string]map[string]struct{}
proxyPoliciesOnce sync.Once
}
// AppliedZoneCandidate is an account-level custom DNS zone reduced to the
// per-peer decision the components calc still makes: include the zone only when
// the peer belongs to one of its distribution groups. Record conversion is done
// once at build time.
type AppliedZoneCandidate struct {
DistributionGroups []string
Zone nmdata.CustomZone
}
// PrivateServiceCandidate is a single private service's synthesized records,
// carried per apex zone. The builder resolves proxy-cluster connectivity and
// domain-suffix matching once; the calc merges the candidates whose AccessGroups
// the peer belongs to, grouped by Zone.Domain.
type PrivateServiceCandidate struct {
AccessGroups []string
Zone nmdata.CustomZone
}
@@ -0,0 +1,18 @@
package nmdata
import "time"
// AccountSettingsInfo is the slim twin of types.AccountSettingsInfo.
type AccountSettingsInfo struct {
PeerLoginExpirationEnabled bool
PeerLoginExpiration time.Duration
PeerInactivityExpirationEnabled bool
PeerInactivityExpiration time.Duration
DNSDomain string
IPv6EnabledGroups []string
RoutingPeerDNSResolutionEnabled bool
LazyConnectionEnabled bool
AutoUpdateVersion string
AutoUpdateAlways bool
MetricsPushEnabled bool
}
@@ -0,0 +1,18 @@
package nmdata
// SimpleRecord is the slim twin of dns.SimpleRecord.
type SimpleRecord struct {
Name string
Type int
Class string
TTL int
RData string
}
// CustomZone is the slim twin of dns.CustomZone.
type CustomZone struct {
Domain string
Records []SimpleRecord
SearchDomainDisabled bool
NonAuthoritative bool
}
@@ -0,0 +1,6 @@
package nmdata
// DNSSettings is the slim twin of types.DNSSettings.
type DNSSettings struct {
DisabledManagementGroups []string
}
@@ -0,0 +1,30 @@
package nmdata
import "slices"
// GroupAllName is the reserved name of the default group that contains every
// peer in an account.
const GroupAllName = "All"
// Group is the slim twin of types.Group.
type Group struct {
ID string
Name string
PublicID string
Peers []string
Resources []Resource
}
func (g *Group) IsGroupAll() bool {
return g.Name == GroupAllName
}
func (g *Group) Copy() *Group {
return &Group{
ID: g.ID,
Name: g.Name,
PublicID: g.PublicID,
Peers: slices.Clone(g.Peers),
Resources: slices.Clone(g.Resources),
}
}
@@ -0,0 +1,84 @@
package nmdata
import (
"reflect"
"testing"
)
// TestGroupCopy_AllFieldsCopied fills every Group field with a unique non-zero
// value derived from its field path, so a field added to Group but forgotten
// in Copy fails here by name without the test needing an update. The unique
// per-path values also catch fields swapped inside Copy.
func TestGroupCopy_AllFieldsCopied(t *testing.T) {
src := &Group{}
seed := 0
fillValue(t, reflect.ValueOf(src).Elem(), "Group", &seed)
copied := src.Copy()
srcV := reflect.ValueOf(src).Elem()
copiedV := reflect.ValueOf(copied).Elem()
for i := 0; i < srcV.NumField(); i++ {
name := srcV.Type().Field(i).Name
if !reflect.DeepEqual(srcV.Field(i).Interface(), copiedV.Field(i).Interface()) {
t.Errorf("field %s not copied: src=%#v copy=%#v",
name, srcV.Field(i).Interface(), copiedV.Field(i).Interface())
}
}
for i := 0; i < srcV.NumField(); i++ {
f := srcV.Field(i)
if f.Kind() != reflect.Slice || f.Len() == 0 {
continue
}
name := srcV.Type().Field(i).Name
fillValue(t, f.Index(0), name+"-mutated", &seed)
if reflect.DeepEqual(f.Interface(), copiedV.Field(i).Interface()) {
t.Errorf("field %s shares memory with the copy", name)
}
}
}
// fillValue sets v to a deterministic non-zero value derived from its field
// path. Kinds it does not handle fail the test loudly, so the filler is
// extended together with the struct instead of silently under-testing new
// fields.
func fillValue(t *testing.T, v reflect.Value, path string, seed *int) {
t.Helper()
switch v.Kind() {
case reflect.String:
v.SetString(path)
case reflect.Bool:
v.SetBool(true)
case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
*seed++
v.SetInt(int64(*seed))
case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64:
*seed++
v.SetUint(uint64(*seed))
case reflect.Float32, reflect.Float64:
*seed++
v.SetFloat(float64(*seed))
case reflect.Slice:
s := reflect.MakeSlice(v.Type(), 2, 2)
fillValue(t, s.Index(0), path+"[0]", seed)
fillValue(t, s.Index(1), path+"[1]", seed)
v.Set(s)
case reflect.Struct:
settable := 0
for i := 0; i < v.NumField(); i++ {
f := v.Field(i)
if !f.CanSet() {
continue
}
settable++
fillValue(t, f, path+"."+v.Type().Field(i).Name, seed)
}
if settable == 0 {
t.Fatalf("struct %s at %s has no settable fields — extend fillValue to construct it", v.Type(), path)
}
default:
t.Fatalf("unsupported kind %s at %s — extend fillValue", v.Kind(), path)
}
}
@@ -0,0 +1,24 @@
package nmdata
import "net/netip"
// NameServerGroup is the slim twin of dns.NameServerGroup.
type NameServerGroup struct {
ID string
PublicID string
Name string
Description string
NameServers []NameServer
Groups []string
Primary bool
Domains []string
Enabled bool
SearchDomainsEnabled bool
}
// NameServer is the slim twin of dns.NameServer.
type NameServer struct {
IP netip.Addr
NSType int
Port int
}
@@ -0,0 +1,16 @@
package nmdata
import "net"
// Network is the slim twin of types.Network.
type Network struct {
Identifier string
Net net.IPNet
NetV6 net.IPNet
Dns string
Serial int64
}
func (n *Network) CurrentSerial() uint64 {
return uint64(n.Serial)
}
@@ -0,0 +1,18 @@
package nmdata
import "net/netip"
// NetworkResource is the slim twin of resources/types.NetworkResource.
type NetworkResource struct {
ID string
NetworkID string
AccountID string
PublicID string
Name string
Description string
Type string
Address string // TODO: isn't persisted in the DB
Domain string
Prefix netip.Prefix
Enabled bool
}
@@ -0,0 +1,10 @@
package nmdata
// NetworkRouter is the slim twin of routers/types.NetworkRouter.
type NetworkRouter struct {
PublicID string
PeerGroups []string
Masquerade bool
Metric int
Enabled bool
}
+138
View File
@@ -0,0 +1,138 @@
package nmdata
import (
"net"
"net/netip"
"slices"
"time"
)
// Peer capability constants mirror the proto enum values.
const (
PeerCapabilitySourcePrefixes int32 = 1
PeerCapabilityIPv6Overlay int32 = 2
PeerCapabilityComponentNetworkMap int32 = 3
)
// Peer is the slim twin of peer.Peer.
type Peer struct {
ID string
Key string
SSHKey string
DNSLabel string
UserID string
SSHEnabled bool
LoginExpirationEnabled bool
LastLogin *time.Time
IP netip.Addr
IPv6 netip.Addr
RequiresApproval bool
Connected bool
ExtraDNSLabels []string
Meta PeerSystemMeta
ProxyMeta ProxyMeta
Location PeerLocation
}
// ProxyMeta is the slim twin of peer.ProxyMeta.
type ProxyMeta struct {
Embedded bool
Cluster string
}
// ProxyDomain is the slim twin of a registered reverse-proxy domain, carrying
// what private-service zone resolution needs: the apex a service domain can sit
// under, and the cluster it is registered against.
type ProxyDomain struct {
Domain string
TargetCluster string
}
// PeerSystemMeta is the slim twin of peer.PeerSystemMeta.
type PeerSystemMeta struct {
WtVersion string
GoOS string
OSVersion string
KernelVersion string
NetworkAddresses []NetworkAddress
Files []File
Capabilities []int32
Flags Flags
SyncMessageVersion int
}
// Flags is the slim twin of peer.Flags.
type Flags struct {
ServerSSHAllowed bool
DisableIPv6 bool
}
// NetworkAddress is the slim twin of peer.NetworkAddress.
type NetworkAddress struct {
NetIP netip.Prefix
}
// File is the slim twin of peer.File.
type File struct {
Path string
ProcessIsRunning bool
}
// PeerLocation is the slim twin of peer.Location.
type PeerLocation struct {
CountryCode string
CityName string
ConnectionIP net.IP
}
func (p *Peer) HasCapability(capability int32) bool {
return slices.Contains(p.Meta.Capabilities, capability)
}
func (p *Peer) SupportsIPv6() bool {
return !p.Meta.Flags.DisableIPv6 && p.HasCapability(PeerCapabilityIPv6Overlay)
}
func (p *Peer) SupportsSourcePrefixes() bool {
return p.HasCapability(PeerCapabilitySourcePrefixes)
}
func (p *Peer) AddedWithSSOLogin() bool {
return p.UserID != ""
}
func (p *Peer) FQDN(dnsDomain string) string {
if dnsDomain == "" {
return ""
}
return p.DNSLabel + "." + dnsDomain
}
func (p *Peer) GetLastLogin() time.Time {
if p.LastLogin != nil {
return *p.LastLogin
}
return time.Time{}
}
// SessionExpiresAt mirrors peer.Peer.SessionExpiresAt.
func (p *Peer) SessionExpiresAt(accountExpirationEnabled bool, expiresIn time.Duration) time.Time {
if !accountExpirationEnabled || !p.AddedWithSSOLogin() || !p.LoginExpirationEnabled {
return time.Time{}
}
last := p.GetLastLogin()
if last.IsZero() {
return time.Time{}
}
return last.Add(expiresIn).UTC()
}
func (p *Peer) LoginExpired(expiresIn time.Duration) (bool, time.Duration) {
if !p.AddedWithSSOLogin() || !p.LoginExpirationEnabled {
return false, 0
}
expiresAt := p.GetLastLogin().Add(expiresIn)
now := time.Now()
timeLeft := expiresAt.Sub(now)
return timeLeft <= 0, timeLeft
}
@@ -0,0 +1,96 @@
package nmdata
const (
policyRuleProtocolALL = "all"
policyRuleProtocolTCP = "tcp"
defaultSSHPortString = "22"
nativeSSHPortString = "22022"
defaultSSHPortNumber uint16 = 22
nativeSSHPortNumber uint16 = 22022
)
// Policy is the slim twin of types.Policy.
type Policy struct {
ID string
PublicID string
Enabled bool
SourcePostureChecks []string
Rules []*PolicyRule
}
// PolicyRule is the slim twin of types.PolicyRule.
type PolicyRule struct {
ID string
PolicyID string
Enabled bool
Action string
Protocol string
Bidirectional bool
Sources []string
Destinations []string
SourceResource Resource
DestinationResource Resource
Ports []string
PortRanges []RulePortRange
AuthorizedGroups map[string][]string
AuthorizedUser string
}
// RulePortRange is the slim twin of types.RulePortRange.
type RulePortRange struct {
Start uint16
End uint16
}
// Resource is the slim twin of types.Resource.
type Resource struct {
ID string
Type string
}
func (p *Policy) SourceGroups() []string {
if len(p.Rules) == 1 && p.Rules[0] != nil {
return p.Rules[0].Sources
}
groups := make(map[string]struct{}, len(p.Rules))
for _, rule := range p.Rules {
if rule == nil {
continue
}
for _, source := range rule.Sources {
groups[source] = struct{}{}
}
}
groupIDs := make([]string, 0, len(groups))
for groupID := range groups {
groupIDs = append(groupIDs, groupID)
}
return groupIDs
}
// PolicyRuleImpliesLegacySSH is the twin-typed sibling of types.PolicyRuleImpliesLegacySSH.
func PolicyRuleImpliesLegacySSH(rule *PolicyRule) bool {
return rule.Protocol == policyRuleProtocolALL ||
(rule.Protocol == policyRuleProtocolTCP && (portsIncludesSSH(rule.Ports) || portRangeIncludesSSH(rule.PortRanges)))
}
func portRangeIncludesSSH(portRanges []RulePortRange) bool {
for _, pr := range portRanges {
if (pr.Start <= defaultSSHPortNumber && pr.End >= defaultSSHPortNumber) || (pr.Start <= nativeSSHPortNumber && pr.End >= nativeSSHPortNumber) {
return true
}
}
return false
}
func portsIncludesSSH(ports []string) bool {
for _, port := range ports {
if port == defaultSSHPortString || port == nativeSSHPortString {
return true
}
}
return false
}
@@ -0,0 +1,83 @@
package nmdata
const (
checkActionAllow = "allow"
checkActionDeny = "deny"
)
// PostureChecks is the slim twin of posture.Checks.
type PostureChecks struct {
ID string
Checks ChecksDefinition
}
// ChecksDefinition is the slim twin of posture.ChecksDefinition.
type ChecksDefinition struct {
NBVersionCheck *NBVersionCheck
OSVersionCheck *OSVersionCheck
GeoLocationCheck *GeoLocationCheck
PeerNetworkRangeCheck *PeerNetworkRangeCheck
ProcessCheck *ProcessCheck
}
// Check is the slim twin of posture.Check. It is sealed: only the check types
// in this package implement it.
type Check interface {
check(peer *Peer) (bool, error)
}
// Passes reports whether the peer satisfies every check in this bundle. It
// mirrors the server posture path: a check returning (false, _) — including on
// an evaluation error — fails the bundle.
func (pc *PostureChecks) Passes(peer *Peer) bool {
return PassesChecks(pc.GetChecks(), peer)
}
// PassesChecks is Passes over an already built check set, for callers that
// evaluate many peers against the same bundle.
func PassesChecks(checks []Check, peer *Peer) bool {
for _, c := range checks {
valid, _ := c.check(peer)
if !valid {
return false
}
}
return true
}
// PostureVerdictChanged reports whether any check in the bundles gives a different
// verdict for newPeer than for oldPeer. Checks are replayed one by one, so a change
// that moves a field but stays on the same side of a threshold does not count. An
// evaluation error is a deny, like in PassesChecks.
func PostureVerdictChanged(checks []*PostureChecks, oldPeer, newPeer *Peer) bool {
for _, pc := range checks {
for _, c := range pc.GetChecks() {
single := []Check{c}
if PassesChecks(single, oldPeer) != PassesChecks(single, newPeer) {
return true
}
}
}
return false
}
// GetChecks returns the initialized checks in the same order as posture.Checks.GetChecks.
func (pc *PostureChecks) GetChecks() []Check {
var checks []Check
if pc.Checks.NBVersionCheck != nil {
checks = append(checks, pc.Checks.NBVersionCheck)
}
if pc.Checks.OSVersionCheck != nil {
checks = append(checks, pc.Checks.OSVersionCheck)
}
if pc.Checks.GeoLocationCheck != nil {
checks = append(checks, pc.Checks.GeoLocationCheck)
}
if pc.Checks.PeerNetworkRangeCheck != nil {
checks = append(checks, pc.Checks.PeerNetworkRangeCheck)
}
if pc.Checks.ProcessCheck != nil {
checks = append(checks, pc.Checks.ProcessCheck)
}
return checks
}
@@ -0,0 +1,45 @@
package nmdata
import "fmt"
// GeoLocation is the slim twin of posture.Location.
type GeoLocation struct {
CountryCode string
CityName string
}
// GeoLocationCheck is the slim twin of posture.GeoLocationCheck.
type GeoLocationCheck struct {
Locations []GeoLocation
Action string
}
func (g *GeoLocationCheck) check(peer *Peer) (bool, error) {
if peer.Location.CountryCode == "" && peer.Location.CityName == "" {
return false, fmt.Errorf("peer's location is not set")
}
for _, loc := range g.Locations {
if loc.CountryCode == peer.Location.CountryCode {
if loc.CityName == "" || loc.CityName == peer.Location.CityName {
switch g.Action {
case checkActionDeny:
return false, nil
case checkActionAllow:
return true, nil
default:
return false, fmt.Errorf("invalid geo location action: %s", g.Action)
}
}
}
}
if g.Action == checkActionDeny {
return true, nil
}
if g.Action == checkActionAllow {
return false, nil
}
return false, fmt.Errorf("invalid geo location action: %s", g.Action)
}
@@ -0,0 +1,38 @@
package nmdata
import (
"strings"
"github.com/hashicorp/go-version"
)
// NBVersionCheck is the slim twin of posture.NBVersionCheck.
type NBVersionCheck struct {
MinVersion string
}
func (n *NBVersionCheck) check(peer *Peer) (bool, error) {
return meetsMinVersion(n.MinVersion, peer.Meta.WtVersion)
}
func meetsMinVersion(minVer, peerVer string) (bool, error) {
peerVer = sanitizeVersion(peerVer)
minVer = sanitizeVersion(minVer)
peerNBVer, err := version.NewVersion(peerVer)
if err != nil {
return false, err
}
constraints, err := version.NewConstraint(">= " + minVer)
if err != nil {
return false, err
}
return constraints.Check(peerNBVer), nil
}
func sanitizeVersion(v string) string {
parts := strings.Split(v, "-")
return parts[0]
}
@@ -0,0 +1,62 @@
package nmdata
import (
"fmt"
"net/netip"
)
// PeerNetworkRangeCheck is the slim twin of posture.PeerNetworkRangeCheck.
type PeerNetworkRangeCheck struct {
Action string
Ranges []netip.Prefix
}
func (p *PeerNetworkRangeCheck) check(peer *Peer) (bool, error) {
peerPrefixes := make([]netip.Prefix, 0, len(peer.Meta.NetworkAddresses)+1)
for _, peerNetAddr := range peer.Meta.NetworkAddresses {
peerPrefixes = append(peerPrefixes, peerNetAddr.NetIP)
}
if connIP := peer.Location.ConnectionIP; len(connIP) > 0 {
if addr, ok := netip.AddrFromSlice(connIP); ok {
addr = addr.Unmap()
peerPrefixes = append(peerPrefixes, netip.PrefixFrom(addr, addr.BitLen()))
}
}
if len(peerPrefixes) == 0 {
return false, fmt.Errorf("peer's does not contain peer network range addresses")
}
for _, peerPrefix := range peerPrefixes {
for _, rangePrefix := range p.Ranges {
if !prefixContains(rangePrefix, peerPrefix) {
continue
}
switch p.Action {
case checkActionDeny:
return false, nil
case checkActionAllow:
return true, nil
default:
return false, fmt.Errorf("invalid peer network range check action: %s", p.Action)
}
}
}
if p.Action == checkActionDeny {
return true, nil
}
if p.Action == checkActionAllow {
return false, nil
}
return false, fmt.Errorf("invalid peer network range check action: %s", p.Action)
}
func prefixContains(outer, inner netip.Prefix) bool {
outer = outer.Masked()
inner = inner.Masked()
return outer.Bits() <= inner.Bits() &&
outer.Addr().BitLen() == inner.Addr().BitLen() &&
outer.Contains(inner.Addr())
}
@@ -0,0 +1,79 @@
package nmdata
import (
"strings"
"github.com/hashicorp/go-version"
)
// MinVersionCheck is the slim twin of posture.MinVersionCheck.
type MinVersionCheck struct {
MinVersion string
}
// MinKernelVersionCheck is the slim twin of posture.MinKernelVersionCheck.
type MinKernelVersionCheck struct {
MinKernelVersion string
}
// OSVersionCheck is the slim twin of posture.OSVersionCheck.
type OSVersionCheck struct {
Android *MinVersionCheck
Darwin *MinVersionCheck
Ios *MinVersionCheck
Linux *MinKernelVersionCheck
Windows *MinKernelVersionCheck
}
func (c *OSVersionCheck) check(peer *Peer) (bool, error) {
switch peer.Meta.GoOS {
case "android":
return checkMinVersion(peer.Meta.OSVersion, c.Android)
case "darwin":
return checkMinVersion(peer.Meta.OSVersion, c.Darwin)
case "ios":
return checkMinVersion(peer.Meta.OSVersion, c.Ios)
case "linux":
kernelVersion := strings.Split(peer.Meta.KernelVersion, "-")[0]
return checkMinKernelVersion(kernelVersion, c.Linux)
case "windows":
return checkMinKernelVersion(peer.Meta.KernelVersion, c.Windows)
}
return true, nil
}
func checkMinVersion(peerVersion string, check *MinVersionCheck) (bool, error) {
if check == nil {
return false, nil
}
peerNBVersion, err := version.NewVersion(peerVersion)
if err != nil {
return false, err
}
constraints, err := version.NewConstraint(">= " + check.MinVersion)
if err != nil {
return false, err
}
return constraints.Check(peerNBVersion), nil
}
func checkMinKernelVersion(peerVersion string, check *MinKernelVersionCheck) (bool, error) {
if check == nil {
return false, nil
}
peerNBVersion, err := version.NewVersion(peerVersion)
if err != nil {
return false, err
}
constraints, err := version.NewConstraint(">= " + check.MinKernelVersion)
if err != nil {
return false, err
}
return constraints.Check(peerNBVersion), nil
}
@@ -0,0 +1,56 @@
package nmdata
import (
"fmt"
"slices"
)
// Process is the slim twin of posture.Process.
type Process struct {
LinuxPath string
MacPath string
WindowsPath string
}
// ProcessCheck is the slim twin of posture.ProcessCheck.
type ProcessCheck struct {
Processes []Process
}
func (p *ProcessCheck) check(peer *Peer) (bool, error) {
peerActiveProcesses := extractPeerActiveProcesses(peer.Meta.Files)
var pathSelector func(Process) string
switch peer.Meta.GoOS {
case "linux":
pathSelector = func(process Process) string { return process.LinuxPath }
case "darwin":
pathSelector = func(process Process) string { return process.MacPath }
case "windows":
pathSelector = func(process Process) string { return process.WindowsPath }
default:
return false, fmt.Errorf("unsupported peer's operating system: %s", peer.Meta.GoOS)
}
return p.areAllProcessesRunning(peerActiveProcesses, pathSelector), nil
}
func (p *ProcessCheck) areAllProcessesRunning(activeProcesses []string, pathSelector func(Process) string) bool {
for _, process := range p.Processes {
path := pathSelector(process)
if path == "" || !slices.Contains(activeProcesses, path) {
return false
}
}
return true
}
func extractPeerActiveProcesses(files []File) []string {
activeProcesses := make([]string, 0, len(files))
for _, file := range files {
if file.ProcessIsRunning {
activeProcesses = append(activeProcesses, file.Path)
}
}
return activeProcesses
}
@@ -0,0 +1,54 @@
package nmdata
import (
"testing"
"github.com/stretchr/testify/assert"
)
func bundle(def ChecksDefinition) []*PostureChecks {
return []*PostureChecks{{Checks: def}}
}
func TestPostureVerdictChanged_ErrorCountsAsDeny(t *testing.T) {
c := bundle(ChecksDefinition{NBVersionCheck: &NBVersionCheck{MinVersion: "1.2.0"}})
tests := []struct {
name string
oldVer, newVer string
want bool
}{
{"both above min, no flip", "1.3.0", "1.4.0", false},
{"crosses up below->above", "1.1.0", "1.3.0", true},
{"unparsable old only -> flip", "garbage", "1.3.0", true},
{"unparsable both -> no flip", "garbage", "junk", false},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
oldPeer := &Peer{Meta: PeerSystemMeta{WtVersion: tt.oldVer}}
newPeer := &Peer{Meta: PeerSystemMeta{WtVersion: tt.newVer}}
assert.Equal(t, tt.want, PostureVerdictChanged(c, oldPeer, newPeer))
})
}
}
func TestPostureVerdictChanged_ReplaysEachCheck(t *testing.T) {
// Old fails the version check, new fails the kernel check: the bundle denies on
// both sides, yet every single check flipped, so the posture must be re-evaluated.
c := bundle(ChecksDefinition{
NBVersionCheck: &NBVersionCheck{MinVersion: "1.0.0"},
OSVersionCheck: &OSVersionCheck{Linux: &MinKernelVersionCheck{MinKernelVersion: "5.0.0"}},
})
oldPeer := &Peer{Meta: PeerSystemMeta{WtVersion: "0.9.0", GoOS: "linux", KernelVersion: "6.0.0"}}
newPeer := &Peer{Meta: PeerSystemMeta{WtVersion: "1.1.0", GoOS: "linux", KernelVersion: "4.0.0"}}
assert.False(t, c[0].Passes(oldPeer))
assert.False(t, c[0].Passes(newPeer))
assert.True(t, PostureVerdictChanged(c, oldPeer, newPeer))
}
func TestPostureVerdictChanged_NoChecks(t *testing.T) {
oldPeer := &Peer{Meta: PeerSystemMeta{WtVersion: "1.0.0"}}
newPeer := &Peer{Meta: PeerSystemMeta{WtVersion: "2.0.0"}}
assert.False(t, PostureVerdictChanged(nil, oldPeer, newPeer))
}
@@ -0,0 +1,108 @@
package nmdata
import (
"net/netip"
"slices"
"strings"
"github.com/netbirdio/netbird/shared/management/domain"
)
// NetworkType mirrors route.NetworkType iota values.
const (
NetworkTypeInvalid = 0
NetworkTypeIPv4 = 1
NetworkTypeIPv6 = 2
NetworkTypeDomain = 3
haSeparator = "|"
)
// Route is the slim twin of route.Route.
type Route struct {
ID string
AccountID string
PublicID string
Network netip.Prefix
Domains domain.List
KeepRoute bool
NetID string
Description string
Peer string
PeerID string
PeerGroups []string
NetworkType int
Masquerade bool
Metric int
Enabled bool
Groups []string
AccessControlGroups []string
SkipAutoApply bool
}
func (r *Route) Equal(other *Route) bool {
if r == nil && other == nil {
return true
} else if r == nil || other == nil {
return false
}
return other.ID == r.ID &&
other.Description == r.Description &&
other.NetID == r.NetID &&
other.Network == r.Network &&
slices.Equal(r.Domains, other.Domains) &&
other.KeepRoute == r.KeepRoute &&
other.NetworkType == r.NetworkType &&
other.Peer == r.Peer &&
other.PeerID == r.PeerID &&
other.Metric == r.Metric &&
other.Masquerade == r.Masquerade &&
other.Enabled == r.Enabled &&
slices.Equal(r.Groups, other.Groups) &&
slices.Equal(r.PeerGroups, other.PeerGroups) &&
slices.Equal(r.AccessControlGroups, other.AccessControlGroups) &&
other.SkipAutoApply == r.SkipAutoApply
}
func (r *Route) IsDynamic() bool {
return r.NetworkType == NetworkTypeDomain
}
func (r *Route) NetString() string {
if r.IsDynamic() && r.Domains != nil {
return r.Domains.SafeString()
}
return r.Network.String()
}
func (r *Route) GetHAUniqueID() string {
return r.NetID + haSeparator + r.NetString()
}
func (r *Route) GetResourceID() string {
return strings.Split(r.ID, ":")[0]
}
func (r *Route) Copy() *Route {
return &Route{
ID: r.ID,
AccountID: r.AccountID,
PublicID: r.PublicID,
Network: r.Network,
Domains: slices.Clone(r.Domains),
KeepRoute: r.KeepRoute,
NetID: r.NetID,
Description: r.Description,
Peer: r.Peer,
PeerID: r.PeerID,
PeerGroups: slices.Clone(r.PeerGroups),
NetworkType: r.NetworkType,
Masquerade: r.Masquerade,
Metric: r.Metric,
Enabled: r.Enabled,
Groups: slices.Clone(r.Groups),
AccessControlGroups: slices.Clone(r.AccessControlGroups),
SkipAutoApply: r.SkipAutoApply,
}
}
@@ -0,0 +1,26 @@
package nmdata
// Service is the slim twin of the reverse-proxy service.Service. It carries
// only the state proxy-policy injection reads: the persisted reverse-proxy
// services and the in-memory ones synthesised from agent-network state, which
// are never written to the database.
type Service struct {
ID string
Enabled bool
Private bool
Mode string
Domain string
ProxyCluster string
AccessGroups []string
Targets []*ServiceTarget
}
// ServiceTarget is the slim twin of service.Target.
type ServiceTarget struct {
Enabled bool
Path string
Port uint16
Protocol string
TargetID string
TargetType string
}
@@ -0,0 +1,111 @@
package networkmap
import (
"context"
"fmt"
"strings"
"github.com/hashicorp/go-multierror"
"github.com/miekg/dns"
log "github.com/sirupsen/logrus"
nbdns "github.com/netbirdio/netbird/dns"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
const peersZoneRecordTTL = 300
// PeersCustomZone builds the peers DNS zone from twin peer rows. It is the
// single source of the zone-record logic; Account.GetPeersCustomZone delegates
// here via twins.
func PeersCustomZone(ctx context.Context, accountID string, dnsDomain string, peers map[string]*nmdata.Peer, ipv6AllowedPeers map[string]struct{}) nmdata.CustomZone {
var merr *multierror.Error
if dnsDomain == "" {
log.WithContext(ctx).Error("no dns domain is set, returning empty zone")
return nmdata.CustomZone{}
}
customZone := nmdata.CustomZone{
Domain: dns.Fqdn(dnsDomain),
Records: make([]nmdata.SimpleRecord, 0, len(peers)),
}
domainSuffix := "." + dnsDomain
var sb strings.Builder
for _, peer := range peers {
if peer == nil {
continue
}
if peer.DNSLabel == "" {
merr = multierror.Append(merr, fmt.Errorf("peer %s has an empty DNS label", peer.ID))
continue
}
sb.Grow(len(peer.DNSLabel) + len(domainSuffix))
sb.WriteString(peer.DNSLabel)
sb.WriteString(domainSuffix)
fqdn := sb.String()
customZone.Records = append(customZone.Records, nmdata.SimpleRecord{
Name: fqdn,
Type: int(dns.TypeA),
Class: nbdns.DefaultClass,
TTL: peersZoneRecordTTL,
RData: peer.IP.String(),
})
// Only advertise AAAA for peers that have a valid IPv6, whose client supports it,
// and that belong to an IPv6-enabled group. Old clients don't configure v6 on their
// WireGuard interface, so resolving their AAAA causes connections to hang.
// Capability changes (client upgrade/downgrade, --disable-ipv6 toggle) propagate
// to other peers via SyncPeer/LoginPeer regardless of version change, so AAAA
// records refresh when a peer first reports the IPv6 overlay capability.
_, peerAllowed := ipv6AllowedPeers[peer.ID]
hasIPv6 := peer.IPv6.IsValid() && peer.SupportsIPv6() && peerAllowed
if hasIPv6 {
customZone.Records = append(customZone.Records, nmdata.SimpleRecord{
Name: fqdn,
Type: int(dns.TypeAAAA),
Class: nbdns.DefaultClass,
TTL: peersZoneRecordTTL,
RData: peer.IPv6.String(),
})
}
sb.Reset()
for _, extraLabel := range peer.ExtraDNSLabels {
sb.Grow(len(extraLabel) + len(domainSuffix))
sb.WriteString(extraLabel)
sb.WriteString(domainSuffix)
extraFqdn := sb.String()
customZone.Records = append(customZone.Records, nmdata.SimpleRecord{
Name: extraFqdn,
Type: int(dns.TypeA),
Class: nbdns.DefaultClass,
TTL: peersZoneRecordTTL,
RData: peer.IP.String(),
})
if hasIPv6 {
customZone.Records = append(customZone.Records, nmdata.SimpleRecord{
Name: extraFqdn,
Type: int(dns.TypeAAAA),
Class: nbdns.DefaultClass,
TTL: peersZoneRecordTTL,
RData: peer.IPv6.String(),
})
}
sb.Reset()
}
}
go func() {
if merr != nil {
log.WithContext(ctx).Errorf("error generating custom zone for account %s: %v", accountID, merr)
}
}()
return customZone
}
@@ -0,0 +1,124 @@
package networkmap
import (
"slices"
"strings"
"github.com/miekg/dns"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
// privateServiceDNSRecordTTL is short so proxy-peer changes propagate quickly.
const privateServiceDNSRecordTTL = 5
// BuildPrivateServiceCandidates derives the per-service DNS records a private
// service publishes, from the twin's own services. It is the counterpart of
// InjectProxyPolicies: that one synthesises the ACL half of a private service,
// this one the DNS half, and both read nmd.Services so a service added to the
// twin after it was loaded — an agent-network service is synthesised in memory
// and never persisted — reaches the peer with both halves rather than one.
//
// The per-peer access-group gate and the merge by apex stay in the components
// calculation; this only precomputes what is account-wide.
func (nmd *NetworkMapData) BuildPrivateServiceCandidates() {
if len(nmd.Services) == 0 {
nmd.PrivateServiceCandidates = nil
return
}
proxyPeersByCluster := nmd.connectedProxyPeersByCluster()
if len(proxyPeersByCluster) == 0 {
nmd.PrivateServiceCandidates = nil
return
}
var out []PrivateServiceCandidate
for _, svc := range nmd.Services {
if svc == nil || !svc.Enabled || !svc.Private || len(svc.AccessGroups) == 0 || svc.Domain == "" {
continue
}
proxyPeers := proxyPeersByCluster[svc.ProxyCluster]
if len(proxyPeers) == 0 {
continue
}
apex := nmd.privateServiceApex(svc)
if apex == "" {
continue
}
records := make([]nmdata.SimpleRecord, 0, len(proxyPeers))
for _, p := range proxyPeers {
records = append(records, nmdata.SimpleRecord{
Name: dns.Fqdn(svc.Domain),
Type: int(dns.TypeA),
Class: "IN",
TTL: privateServiceDNSRecordTTL,
RData: p.IP.String(),
})
}
out = append(out, PrivateServiceCandidate{
AccessGroups: svc.AccessGroups,
Zone: nmdata.CustomZone{
// NonAuthoritative keeps the zone match-only, so names without
// an explicit record fall through to the upstream resolver
// instead of returning NXDOMAIN for the whole apex.
Domain: dns.Fqdn(apex),
Records: records,
NonAuthoritative: true,
SearchDomainDisabled: true,
},
})
}
nmd.PrivateServiceCandidates = out
}
// privateServiceApex resolves the zone a service's record hangs under: the
// cluster when the service sits directly beneath it, otherwise the longest
// registered custom domain pointing at that same cluster. A service whose
// domain matches no registered apex publishes nothing, since a zone the client
// never intercepts cannot answer the query.
func (nmd *NetworkMapData) privateServiceApex(svc *nmdata.Service) string {
if domainUnderSuffix(svc.Domain, svc.ProxyCluster) {
return svc.ProxyCluster
}
apex := ""
for _, d := range nmd.Domains {
if d.TargetCluster != svc.ProxyCluster {
continue
}
if domainUnderSuffix(svc.Domain, d.Domain) && len(d.Domain) > len(apex) {
apex = d.Domain
}
}
return apex
}
func domainUnderSuffix(domain, suffix string) bool {
if suffix == "" {
return false
}
return domain == suffix || strings.HasSuffix(domain, "."+suffix)
}
// connectedProxyPeersByCluster groups the account's embedded proxy peers by the
// cluster they serve, keeping only connected ones.
func (nmd *NetworkMapData) connectedProxyPeersByCluster() map[string][]*nmdata.Peer {
var out map[string][]*nmdata.Peer
for _, peer := range nmd.Peers {
if peer == nil || !peer.ProxyMeta.Embedded || !peer.Connected || !peer.IP.IsValid() {
continue
}
if out == nil {
out = make(map[string][]*nmdata.Peer)
}
out[peer.ProxyMeta.Cluster] = append(out[peer.ProxyMeta.Cluster], peer)
}
for _, peers := range out {
slices.SortFunc(peers, func(a, b *nmdata.Peer) int { return strings.Compare(a.ID, b.ID) })
}
return out
}
@@ -0,0 +1,148 @@
package networkmap
import (
"net/netip"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
func proxyPeer(id, ip, cluster string, connected bool) *nmdata.Peer {
return &nmdata.Peer{
ID: id, Key: id + "-key", IP: netip.MustParseAddr(ip), Connected: connected,
ProxyMeta: nmdata.ProxyMeta{Embedded: true, Cluster: cluster},
}
}
func privateService(id, domain, cluster string, groups ...string) *nmdata.Service {
return &nmdata.Service{
ID: id, Enabled: true, Private: true, Mode: "http",
Domain: domain, ProxyCluster: cluster, AccessGroups: groups,
}
}
func twinWithProxy(services ...*nmdata.Service) *NetworkMapData {
return &NetworkMapData{
Peers: map[string]*nmdata.Peer{
"proxy-1": proxyPeer("proxy-1", "100.64.0.99", "eu.proxy.netbird.io", true),
},
Services: services,
}
}
// An agent-network service is synthesised in memory and never persisted, so it
// only ever reaches the twin through nmd.Services. Deriving the zone from that
// same field is what stops it from arriving with an ACL and no name.
func TestBuildPrivateServiceCandidates_SynthesisedServiceGetsAZone(t *testing.T) {
nmd := twinWithProxy(privateService(
"agent-network-acct-1", "acct-1.agent.netbird.io", "eu.proxy.netbird.io", "grp-admins"))
nmd.BuildPrivateServiceCandidates()
require.Len(t, nmd.PrivateServiceCandidates, 0,
"a service whose domain sits under no registered apex publishes nothing")
nmd.Domains = []nmdata.ProxyDomain{
{Domain: "agent.netbird.io", TargetCluster: "eu.proxy.netbird.io"},
}
nmd.BuildPrivateServiceCandidates()
require.Len(t, nmd.PrivateServiceCandidates, 1)
got := nmd.PrivateServiceCandidates[0]
assert.Equal(t, []string{"grp-admins"}, got.AccessGroups)
assert.Equal(t, "agent.netbird.io.", got.Zone.Domain, "apex is the registered domain, not the service FQDN")
assert.True(t, got.Zone.NonAuthoritative, "zone stays match-only")
assert.True(t, got.Zone.SearchDomainDisabled)
require.Len(t, got.Zone.Records, 1)
assert.Equal(t, nmdata.SimpleRecord{
Name: "acct-1.agent.netbird.io.", Type: 1, Class: "IN", TTL: 5, RData: "100.64.0.99",
}, got.Zone.Records[0])
}
func TestBuildPrivateServiceCandidates_ClusterApexNeedsNoRegisteredDomain(t *testing.T) {
nmd := twinWithProxy(privateService("svc-1", "myapp.eu.proxy.netbird.io", "eu.proxy.netbird.io", "grp-admins"))
nmd.BuildPrivateServiceCandidates()
require.Len(t, nmd.PrivateServiceCandidates, 1)
assert.Equal(t, "eu.proxy.netbird.io.", nmd.PrivateServiceCandidates[0].Zone.Domain)
}
func TestBuildPrivateServiceCandidates_LongestRegisteredApexWins(t *testing.T) {
nmd := twinWithProxy(privateService("svc-1", "app.sub.example.com", "eu.proxy.netbird.io", "grp-admins"))
nmd.Domains = []nmdata.ProxyDomain{
{Domain: "example.com", TargetCluster: "eu.proxy.netbird.io"},
{Domain: "sub.example.com", TargetCluster: "eu.proxy.netbird.io"},
{Domain: "other.com", TargetCluster: "eu.proxy.netbird.io"},
}
nmd.BuildPrivateServiceCandidates()
require.Len(t, nmd.PrivateServiceCandidates, 1)
assert.Equal(t, "sub.example.com.", nmd.PrivateServiceCandidates[0].Zone.Domain)
}
func TestBuildPrivateServiceCandidates_RegisteredApexOfAnotherClusterIsIgnored(t *testing.T) {
nmd := twinWithProxy(privateService("svc-1", "app.example.com", "eu.proxy.netbird.io", "grp-admins"))
nmd.Domains = []nmdata.ProxyDomain{
{Domain: "example.com", TargetCluster: "us.proxy.netbird.io"},
}
nmd.BuildPrivateServiceCandidates()
assert.Empty(t, nmd.PrivateServiceCandidates)
}
// A disconnected proxy peer's tunnel IP does not answer, so publishing it
// black-holes the name for as long as a client caches the record.
func TestBuildPrivateServiceCandidates_OnlyConnectedProxyPeersSurface(t *testing.T) {
nmd := twinWithProxy(privateService("svc-1", "myapp.eu.proxy.netbird.io", "eu.proxy.netbird.io", "grp-admins"))
nmd.Peers["proxy-2"] = proxyPeer("proxy-2", "100.64.0.100", "eu.proxy.netbird.io", false)
nmd.BuildPrivateServiceCandidates()
require.Len(t, nmd.PrivateServiceCandidates, 1)
require.Len(t, nmd.PrivateServiceCandidates[0].Zone.Records, 1)
assert.Equal(t, "100.64.0.99", nmd.PrivateServiceCandidates[0].Zone.Records[0].RData)
nmd.Peers["proxy-1"].Connected = false
nmd.BuildPrivateServiceCandidates()
assert.Empty(t, nmd.PrivateServiceCandidates, "no connected proxy peer means no zone at all")
}
func TestBuildPrivateServiceCandidates_SkipsServicesThatGrantNothing(t *testing.T) {
cases := map[string]func(*nmdata.Service){
"disabled": func(s *nmdata.Service) { s.Enabled = false },
"not private": func(s *nmdata.Service) { s.Private = false },
"no access groups": func(s *nmdata.Service) { s.AccessGroups = nil },
"no domain": func(s *nmdata.Service) { s.Domain = "" },
"other cluster": func(s *nmdata.Service) { s.ProxyCluster = "us.proxy.netbird.io" },
}
for name, mutate := range cases {
t.Run(name, func(t *testing.T) {
svc := privateService("svc-1", "myapp.eu.proxy.netbird.io", "eu.proxy.netbird.io", "grp-admins")
mutate(svc)
nmd := twinWithProxy(svc)
nmd.BuildPrivateServiceCandidates()
assert.Empty(t, nmd.PrivateServiceCandidates)
})
}
}
func TestBuildPrivateServiceCandidates_MultipleConnectedProxyPeersEachGetARecord(t *testing.T) {
nmd := twinWithProxy(privateService("svc-1", "myapp.eu.proxy.netbird.io", "eu.proxy.netbird.io", "grp-admins"))
nmd.Peers["proxy-2"] = proxyPeer("proxy-2", "100.64.0.100", "eu.proxy.netbird.io", true)
nmd.BuildPrivateServiceCandidates()
require.Len(t, nmd.PrivateServiceCandidates, 1)
records := nmd.PrivateServiceCandidates[0].Zone.Records
require.Len(t, records, 2)
assert.Equal(t, "100.64.0.99", records[0].RData, "records are ordered by proxy peer id")
assert.Equal(t, "100.64.0.100", records[1].RData)
}
@@ -0,0 +1,209 @@
package networkmap
import (
"fmt"
"slices"
"strings"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/shared/management/types"
)
const (
serviceModeUDP = "udp"
privateServicePortHTTP = 80
privateServicePortHTTPS = 443
)
// InjectProxyPolicies synthesises the in-memory ACLs that carry reverse-proxy
// traffic and appends them to the twin's policies. They are never persisted,
// so no builder can load them: a proxy-access policy lets a cluster's proxy
// peers reach each enabled target of a service, and a private-access policy
// lets a private service's AccessGroups reach those proxy peers on HTTP(S).
//
// GetPeerNetworkMapComponents calls it, so every caller of the twin gets the
// same policy set no matter which builder produced it. It runs at most once
// per twin, and is safe to call again to force the synthesis early.
func (nmd *NetworkMapData) InjectProxyPolicies() {
nmd.proxyPoliciesOnce.Do(nmd.injectProxyPolicies)
}
func (nmd *NetworkMapData) injectProxyPolicies() {
if len(nmd.Services) == 0 {
return
}
proxyPeersByCluster := nmd.proxyPeersByCluster()
if len(proxyPeersByCluster) == 0 {
return
}
for _, svc := range nmd.Services {
if svc == nil || !svc.Enabled {
continue
}
proxyPeers := proxyPeersByCluster[svc.ProxyCluster]
for _, target := range svc.Targets {
if target == nil || !target.Enabled {
continue
}
port, ok := resolveTargetPort(target)
if !ok {
continue
}
for _, proxyPeer := range proxyPeers {
nmd.addInjectedPolicy(proxyAccessPolicy(svc, target, proxyPeer, port))
}
}
nmd.injectPrivateServicePolicies(svc, proxyPeers)
}
}
// injectPrivateServicePolicies synthesises AccessGroups → cluster proxy peers on TCP 80/443.
func (nmd *NetworkMapData) injectPrivateServicePolicies(svc *nmdata.Service, proxyPeers []*nmdata.Peer) {
if !svc.Private || len(svc.AccessGroups) == 0 || len(proxyPeers) == 0 {
return
}
// A service's AccessGroups can name groups that no longer exist — persisted
// services and the agent-network synthesiser both carry the ids verbatim from
// their own state. An unresolvable source authorises nothing, so drop it here
// rather than let the network-map assembly resolve it to a nil group.
sources := nmd.existingGroupIDs(svc.AccessGroups)
if len(sources) == 0 {
return
}
for _, proxyPeer := range proxyPeers {
nmd.addInjectedPolicy(privateAccessPolicy(svc, proxyPeer, sources))
}
}
// addInjectedPolicy appends the policy to the twin's policy set, and to the
// policies of the network resource it targets — mirroring the account path,
// where the resource-policy map was built after injection.
func (nmd *NetworkMapData) addInjectedPolicy(policy *nmdata.Policy) {
nmd.Policies = append(nmd.Policies, policy)
resourceID := policy.Rules[0].DestinationResource.ID
if resourceID == "" {
return
}
for _, resource := range nmd.NetworkResources {
if resource == nil || !resource.Enabled || resource.ID != resourceID {
continue
}
if nmd.ResourcePolicies == nil {
nmd.ResourcePolicies = make(map[string][]*nmdata.Policy)
}
nmd.ResourcePolicies[resourceID] = append(nmd.ResourcePolicies[resourceID], policy)
return
}
}
func proxyAccessPolicy(svc *nmdata.Service, target *nmdata.ServiceTarget, proxyPeer *nmdata.Peer, port uint16) *nmdata.Policy {
policyID := fmt.Sprintf("proxy-access-%s-%s-%s", svc.ID, proxyPeer.ID, target.Path)
protocol := types.PolicyRuleProtocolTCP
if svc.Mode == serviceModeUDP {
protocol = types.PolicyRuleProtocolUDP
}
return &nmdata.Policy{
ID: policyID,
// The envelope encoder puts public ids on the wire and degrades to an
// empty one when a policy has none. A synthesised policy has no
// persisted row to take a public id from, and its own id is already
// stable and unique, so it serves as both.
PublicID: policyID,
Enabled: true,
Rules: []*nmdata.PolicyRule{
{
ID: policyID,
PolicyID: policyID,
Enabled: true,
SourceResource: nmdata.Resource{ID: proxyPeer.ID, Type: string(types.ResourceTypePeer)},
DestinationResource: nmdata.Resource{ID: target.TargetID, Type: target.TargetType},
Bidirectional: false,
Protocol: string(protocol),
Action: string(types.PolicyTrafficActionAccept),
PortRanges: []nmdata.RulePortRange{{Start: port, End: port}},
},
},
}
}
func privateAccessPolicy(svc *nmdata.Service, proxyPeer *nmdata.Peer, accessGroups []string) *nmdata.Policy {
policyID := fmt.Sprintf("private-access-%s-%s", svc.ID, proxyPeer.ID)
return &nmdata.Policy{
ID: policyID,
PublicID: policyID,
Enabled: true,
Rules: []*nmdata.PolicyRule{
{
ID: policyID,
PolicyID: policyID,
Enabled: true,
Sources: slices.Clone(accessGroups),
DestinationResource: nmdata.Resource{ID: proxyPeer.ID, Type: string(types.ResourceTypePeer)},
Bidirectional: false,
Protocol: string(types.PolicyRuleProtocolTCP),
Action: string(types.PolicyTrafficActionAccept),
PortRanges: []nmdata.RulePortRange{
{Start: privateServicePortHTTP, End: privateServicePortHTTP},
{Start: privateServicePortHTTPS, End: privateServicePortHTTPS},
},
},
},
}
}
func resolveTargetPort(target *nmdata.ServiceTarget) (uint16, bool) {
if target.Port != 0 {
return target.Port, true
}
switch target.Protocol {
case "https", "tls":
return privateServicePortHTTPS, true
case "http":
return privateServicePortHTTP, true
default:
return 0, false
}
}
// proxyPeersByCluster groups the account's embedded proxy peers by the cluster
// they serve. Sorted by peer ID so the synthesised policy order is stable.
func (nmd *NetworkMapData) proxyPeersByCluster() map[string][]*nmdata.Peer {
var proxyPeers map[string][]*nmdata.Peer
for _, peer := range nmd.Peers {
if peer == nil || !peer.ProxyMeta.Embedded {
continue
}
if proxyPeers == nil {
proxyPeers = make(map[string][]*nmdata.Peer)
}
proxyPeers[peer.ProxyMeta.Cluster] = append(proxyPeers[peer.ProxyMeta.Cluster], peer)
}
for _, peers := range proxyPeers {
slices.SortFunc(peers, func(a, b *nmdata.Peer) int { return strings.Compare(a.ID, b.ID) })
}
return proxyPeers
}
// existingGroupIDs returns the subset of groupIDs that resolve to a group,
// preserving the input order.
func (nmd *NetworkMapData) existingGroupIDs(groupIDs []string) []string {
out := make([]string, 0, len(groupIDs))
for _, groupID := range groupIDs {
if _, ok := nmd.Groups[groupID]; ok {
out = append(out, groupID)
}
}
return out
}
+1 -1
View File
@@ -1,4 +1,4 @@
#!/bin/bash
#!/usr/bin/env bash
set -e
if ! which realpath > /dev/null 2>&1
File diff suppressed because it is too large Load Diff
+558 -1
View File
@@ -52,6 +52,14 @@ service ManagementService {
// Executes a job on a target peer (e.g., debug bundle)
rpc Job(stream EncryptedMessage) returns (stream EncryptedMessage) {}
// ExtendAuthSession refreshes the peer's session expiry deadline using a fresh JWT.
// Same JWT validation pipeline as Login (including jwt.UserID == peer.UserID check),
// but does not redo the network-map sync. Only valid for SSO-registered peers where
// login expiration is enabled. The tunnel remains up.
// EncryptedMessage of the request has a body of ExtendAuthSessionRequest.
// EncryptedMessage of the response has a body of ExtendAuthSessionResponse.
rpc ExtendAuthSession(EncryptedMessage) returns (EncryptedMessage) {}
// CreateExpose creates a temporary reverse proxy service for a peer
rpc CreateExpose(EncryptedMessage) returns (EncryptedMessage) {}
@@ -102,6 +110,13 @@ message BundleParameters {
int64 bundle_for_time = 2;
int32 log_file_count = 3;
bool anonymize = 4;
// anonymize_level selects how much the anonymizer redacts: "default"
// (or empty) keeps internal IP ranges, "strict" also anonymizes them.
// Unknown values are treated as "strict".
string anonymize_level = 5;
// upload_url is the service URL the client requests an upload URL from
// before uploading the bundle. Empty selects the default upload server.
string upload_url = 6;
}
message BundleResult {
@@ -133,6 +148,23 @@ message SyncResponse {
// Posture checks to be evaluated by client
repeated Checks Checks = 6;
// 3-state session deadline. Carried on every Sync snapshot so admin-side
// changes propagate live without a client reconnect.
// field unset (nil) → snapshot carries no info; client keeps the
// deadline it already had
// set, seconds=0 nanos=0 → explicit "expiry disabled" or peer is not
// SSO-registered; client clears its anchor
// set, valid timestamp → new absolute UTC deadline
google.protobuf.Timestamp sessionExpiresAt = 7;
// NetworkMapEnvelope carries the component-based wire format for peers that
// advertise PeerCapabilityComponentNetworkMap. When set, NetworkMap (field 5)
// is left empty: management ships components and the client runs Calculate()
// locally instead of receiving an expanded NetworkMap.
NetworkMapEnvelope NetworkMapEnvelope = 8;
int32 Version = 9;
}
message SyncMetaRequest {
@@ -200,6 +232,25 @@ message Flags {
bool enableSSHLocalPortForwarding = 13;
bool enableSSHRemotePortForwarding = 14;
bool disableSSHAuth = 15;
bool disableIPv6 = 16;
// remoteJobsAllowed mirrors the peer's local opt-in for management-requested
// remote jobs (e.g. debug bundles). Reported so the dashboard can surface
// peers that have opted out.
bool remoteJobsAllowed = 17;
}
// PeerCapability represents a feature the client binary supports.
// Reported in PeerSystemMeta.capabilities on every login/sync.
enum PeerCapability {
PeerCapabilityUnknown = 0;
// Client reads SourcePrefixes instead of the deprecated PeerIP string.
PeerCapabilitySourcePrefixes = 1;
// Client handles IPv6 overlay addresses and firewall rules.
PeerCapabilityIPv6Overlay = 2;
// Client receives NetworkMap as components and assembles it locally.
PeerCapabilityComponentNetworkMap = 3;
}
// PeerSystemMeta is machine meta data like OS and version.
@@ -221,6 +272,9 @@ message PeerSystemMeta {
Environment environment = 15;
repeated File files = 16;
Flags flags = 17;
repeated PeerCapability capabilities = 18;
int32 syncMessageVersion = 19;
}
message LoginResponse {
@@ -230,6 +284,31 @@ message LoginResponse {
PeerConfig peerConfig = 2;
// Posture checks to be evaluated by client
repeated Checks Checks = 3;
// 3-state session deadline; same encoding as SyncResponse.sessionExpiresAt.
// field unset (nil) → no info; client keeps any deadline it had
// set, seconds=0 nanos=0 → explicit "expiry disabled" / non-SSO peer
// set, valid timestamp → new absolute UTC deadline
google.protobuf.Timestamp sessionExpiresAt = 4;
}
// ExtendAuthSessionRequest carries a fresh JWT to refresh the peer's session deadline.
// The encrypted body of an EncryptedMessage with this payload is sent to the
// ExtendAuthSession RPC.
message ExtendAuthSessionRequest {
// SSO token (must be a fresh, valid JWT for the peer's owning user)
string jwtToken = 1;
// Meta data of the peer (used for IdP user info refresh consistent with Login)
PeerSystemMeta meta = 2;
}
// ExtendAuthSessionResponse contains the refreshed session deadline.
message ExtendAuthSessionResponse {
// 3-state session deadline; same encoding as SyncResponse.sessionExpiresAt.
// In practice ExtendAuthSession only succeeds for SSO peers with expiry
// enabled, so this carries a valid timestamp on the success path. The
// 3-state encoding is documented here for symmetry with Login/Sync.
google.protobuf.Timestamp sessionExpiresAt = 1;
}
message ServerKeyResponse {
@@ -256,6 +335,8 @@ message NetbirdConfig {
RelayConfig relay = 4;
FlowConfig flow = 5;
MetricsConfig metrics = 6;
}
// HostConfig describes connection properties of some server (e.g. STUN, Signal, Management)
@@ -294,6 +375,10 @@ message FlowConfig {
bool dnsCollection = 8;
}
message MetricsConfig {
bool enabled = 1;
}
// JWTConfig represents JWT authentication configuration for validating tokens.
message JWTConfig {
string issuer = 1;
@@ -335,6 +420,9 @@ message PeerConfig {
// Auto-update config
AutoUpdateSettings autoUpdate = 8;
// IPv6 overlay address as compact bytes: 16 bytes IP + 1 byte prefix length.
bytes address_v6 = 9;
}
message AutoUpdateSettings {
@@ -421,6 +509,22 @@ message RemotePeerConfig {
string fqdn = 4;
string agentVersion = 5;
// lazyState is the management per-peer override for lazy (on-demand)
// connections to this remote peer. LazyStateDefault follows the account-wide
// flag; LazyStateLazy forces lazy; LazyStateEager forces an always-active
// connection. A local NB_LAZY_CONN/MDM override still wins over this.
LazyState lazyState = 6;
}
// LazyState is the management per-peer override for lazy connections.
enum LazyState {
// Follow the account-wide lazy connection flag.
LazyStateDefault = 0;
// Force a lazy (on-demand) connection regardless of the account flag.
LazyStateLazy = 1;
// Force an always-active connection regardless of the account flag.
LazyStateEager = 2;
}
// SSHConfig represents SSH configurations of a peer.
@@ -552,6 +656,13 @@ enum RuleProtocol {
UDP = 3;
ICMP = 4;
CUSTOM = 5;
// NETBIRD_SSH (types.PolicyRuleProtocolType "netbird-ssh") is the marker
// policy rule that drives SSH-server activation in Calculate(). The legacy
// proto.FirewallRule path doesn't ship this value (Calculate already
// expands SSH rules into TCP/22 before encoding), but the components path
// ships RAW policies — the client must see this protocol to derive
// AuthorizedUsers locally.
NETBIRD_SSH = 6;
}
enum RuleDirection {
@@ -567,7 +678,8 @@ enum RuleAction {
// FirewallRule represents a firewall rule
message FirewallRule {
string PeerIP = 1;
// Use sourcePrefixes instead.
string PeerIP = 1 [deprecated = true];
RuleDirection Direction = 2;
RuleAction Action = 3;
RuleProtocol Protocol = 4;
@@ -576,6 +688,13 @@ message FirewallRule {
// PolicyID is the ID of the policy that this rule belongs to
bytes PolicyID = 7;
// CustomProtocol is a custom protocol ID when Protocol is CUSTOM.
uint32 customProtocol = 8;
// Compact source IP prefixes for this rule, supersedes PeerIP.
// Each entry is 5 bytes (v4) or 17 bytes (v6): [IP bytes][1 byte prefix_len].
repeated bytes sourcePrefixes = 9;
}
message NetworkAddress {
@@ -684,3 +803,441 @@ message StopExposeRequest {
}
message StopExposeResponse {}
// =====================================================================
// Component-based NetworkMap wire format (PeerCapabilityComponentNetworkMap).
//
// Peers that advertise this capability receive NetworkMap building blocks
// (peers + groups + policies + routes + dns + ssh + forwarding) and run the
// expansion (Calculate) locally instead of receiving a fully-expanded
// NetworkMap from the server.
// =====================================================================
// NetworkMapEnvelope wraps either a full snapshot or a delta. Only Full is
// emitted today; Delta is reserved for the incremental-update work.
message NetworkMapEnvelope {
oneof payload {
NetworkMapComponentsFull full = 1;
NetworkMapComponentsDelta delta = 2;
}
}
// NetworkMapComponentsFull is the full per-peer component snapshot. The
// client decodes it into a types.NetworkMapComponents and runs Calculate()
// locally to produce the same NetworkMap the legacy server path would have
// produced. Every field carries RAW component data — no server-side
// expansion (firewall rules, DNS config, SSH auth, route firewall rules,
// forwarding rules) is shipped; the client computes those itself.
message NetworkMapComponentsFull {
uint64 serial = 1;
// Peer config for the receiving peer (legacy proto.PeerConfig kept as-is —
// it carries the receiving peer's own overlay address, FQDN, SSH config).
PeerConfig peer_config = 2;
// Account-level network metadata (id, IPv4/IPv6 overlay subnets, DNS,
// serial). Mirrors types.Network.
AccountNetwork network = 3;
// Account-level settings the client needs for its local Calculate().
AccountSettingsCompact account_settings = 4;
// Account DNS settings (mirrors types.DNSSettings).
DNSSettingsCompact dns_settings = 5;
// Domain shared across all peers in this account, e.g. "netbird.cloud".
// Each peer's FQDN is dns_label + "." + dns_domain.
string dns_domain = 6;
// Custom-zone domain for this peer's view (c.CustomZoneDomain). Empty when
// the peer has no custom zone records.
string custom_zone_domain = 7;
// Deduplicated agent versions; PeerCompact.agent_version_idx indexes here.
// Empty string at index 0 if any peer has no version.
repeated string agent_versions = 8;
// All peers (deduplicated). The client splits peers into online / offline
// locally using account_settings.peer_login_expiration on receive.
repeated PeerCompact peers = 9;
// Indexes into peers for the subset that may act as routers.
repeated uint32 router_peer_indexes = 10;
// Policies that affect the receiving peer.
repeated PolicyCompact policies = 11;
// Groups in unspecified order — clients key off id (public_id).
repeated GroupCompact groups = 12;
// Routes relevant to this peer, raw shape (mirrors []*route.Route).
repeated RouteRaw routes = 13;
// Nameserver groups (mirrors []*nbdns.NameServerGroup).
repeated NameServerGroupRaw nameserver_groups = 14;
// All DNS records the client needs to assemble its custom zone. Reuses
// the existing SimpleRecord wire shape.
repeated SimpleRecord all_dns_records = 15;
// Custom zones (typically the peer's own zone). Reuses the existing
// CustomZone wire shape.
repeated CustomZone account_zones = 16;
// Network resources (mirrors []*resourceTypes.NetworkResource).
repeated NetworkResourceRaw network_resources = 17;
// Routers per network. Outer key: network public_id. Each entry is
// the set of routers backing that network for this peer's view.
map<string, NetworkRouterList> routers_map = 18;
// For each NetworkResource public_id, the indexes into policies[]
// that apply to it.
map<string, PolicyIds> resource_policies_map = 19;
// Group-id (public_id) → user ids authorized for SSH on members.
map<string, UserIDList> group_id_to_user_ids = 20;
// Account-level allowed user ids (used by Calculate() when assembling SSH
// authorized users for the receiving peer).
repeated string allowed_user_ids = 21;
// Per posture-check public_id, the set of peer indexes that failed
// the check. Server-side evaluation result; clients do not re-evaluate.
map<string, PeerIndexSet> posture_failed_peers = 22;
// Account-level DNS forwarder port (mirrors the legacy
// proto.DNSConfig.ForwarderPort). Computed by the controller from peer
// versions; clients fold it into their Calculate() DNS output.
int64 dns_forwarder_port = 23;
// Pre-expanded NetworkMap fragments injected post-Calculate by external
// controllers (BYOP / port-forwarding proxies). The receiving client
// merges these into its locally-computed NetworkMap the same way the
// legacy server does via NetworkMap.Merge — so downstream consumers see
// a unified merged result regardless of source.
ProxyPatch proxy_patch = 24;
// SSH UserIDClaim — server-side HttpServerConfig.AuthUserIDClaim, or
// "sub" by default. Populated in proto.SSHAuth.UserIDClaim when the
// client rebuilds the NetworkMap from this envelope. Empty when the
// account has no AuthorizedUsers (and thus no SshAuth to populate).
string user_id_claim = 25;
// Reserved for future component additions (incremental_serial, parent_seq,
// etc.) without forcing a renumber.
reserved 26 to 50;
}
// ProxyPatch carries NetworkMap fragments that don't fit the component-graph
// model — they're pre-expanded by external controllers (BYOP /
// port-forwarding proxies) and injected post-Calculate. Fields use the
// legacy wire types because the proxy delivers them pre-formed; there is
// no raw component shape to convert from. Empty when no proxy is active.
message ProxyPatch {
repeated RemotePeerConfig peers = 1;
repeated RemotePeerConfig offline_peers = 2;
repeated FirewallRule firewall_rules = 3;
repeated Route routes = 4;
repeated RouteFirewallRule route_firewall_rules = 5;
repeated ForwardingRule forwarding_rules = 6;
}
// AccountSettingsCompact carries the account-level settings the client needs
// to evaluate locally. Mirrors the subset of types.AccountSettingsInfo that
// Calculate() actually reads — login-expiration (used to filter expired
// peers). Inactivity expiration is purely server-side bookkeeping and is not
// shipped.
message AccountSettingsCompact {
bool peer_login_expiration_enabled = 1;
// Login expiration window. Unit is nanoseconds (matches time.Duration).
int64 peer_login_expiration_ns = 2;
}
// AccountNetwork is the account-level overlay metadata. Mirrors types.Network
// so the client can populate NetworkMap.Network without a server round-trip.
message AccountNetwork {
string identifier = 1;
// IPv4 overlay subnet in CIDR form (e.g. "100.64.0.0/16").
string net_cidr = 2;
// IPv6 ULA overlay subnet in CIDR form (e.g. "fd00:4e42::/64"). Empty when
// the account has no IPv6 overlay yet.
string net_v6_cidr = 3;
string dns = 4;
uint64 serial = 5;
}
// NetworkMapComponentsDelta is reserved for the incremental update
// protocol. Field numbers 1100 are pre-allocated to keep room for the
// planned event types without needing a renumber.
message NetworkMapComponentsDelta {
reserved 1 to 100;
}
// PeerCompact is the wire-shape of a remote peer used by the component
// format. It carries every field of types.Peer that the client's local
// Calculate() reads — including the trio needed to evaluate
// LoginExpired() (added_with_sso_login + login_expiration_enabled +
// last_login_unix_nano). Fields the client does not consume (Status,
// CreatedAt, etc.) are not shipped.
message PeerCompact {
// Raw 32-byte WireGuard public key (no base64 wrapping).
bytes wg_pub_key = 1;
// Raw 4-byte IPv4 overlay address. Always a /32 host route, so no prefix
// byte is needed.
bytes ip = 2;
// Raw 16-byte IPv6 overlay address; always a /128 host route. Empty when
// the peer has no IPv6 overlay address.
bytes ipv6 = 3;
// Raw SSH public key bytes (or empty).
bytes ssh_pub_key = 4;
// DNS label without the account's domain suffix. Full FQDN is
// dns_label + "." + NetworkMapComponentsFull.dns_domain.
string dns_label = 5;
string agent_version = 6;
// True iff the peer was added via SSO login (i.e., types.Peer.UserID is
// non-empty). Combined with login_expiration_enabled and
// last_login_unix_nano this lets the client reproduce
// (*Peer).LoginExpired() locally.
bool added_with_sso_login = 7;
// True when the peer's login can expire — mirrors
// types.Peer.LoginExpirationEnabled.
bool login_expiration_enabled = 8;
// Unix-nanosecond timestamp of the peer's last login. 0 when the peer has
// never logged in (server stores nil; client treats 0 as "epoch", which
// makes a fresh peer immediately expired iff login_expiration_enabled is
// true — the same semantics as types.Peer.GetLastLogin).
int64 last_login_unix_nano = 9;
// True when the peer has an SSH server enabled locally. Used by the
// legacy SSH path in Calculate() (`policyRuleImpliesLegacySSH`): a rule
// with protocol ALL/TCP-with-SSH-ports activates SSH for the receiving
// peer when this bit is set, even without an explicit NetbirdSSH rule.
bool ssh_enabled = 10;
// Mirror of types.Peer.SupportsIPv6() — !Meta.Flags.DisableIPv6 &&
// HasCapability(PeerCapabilityIPv6Overlay). Used by the local peer's
// Calculate() when deciding whether to emit IPv6 firewall rules
// (appendIPv6FirewallRule) against this peer's IPv6 address.
bool supports_ipv6 = 11;
// Mirror of types.Peer.SupportsSourcePrefixes() —
// HasCapability(PeerCapabilitySourcePrefixes). Determines whether the
// local peer's Calculate() emits SourcePrefixes alongside legacy PeerIP
// fields in proto.FirewallRule.
bool supports_source_prefixes = 12;
// Mirror of types.Peer.Meta.Flags.ServerSSHAllowed. Read by Calculate()
// when expanding TCP port-22 firewall rules — the native SSH companion
// (port 22022) is only added when this flag is set and the peer agent
// version supports it.
bool server_ssh_allowed = 13;
// Mirror of types.Peer.ProxyMeta.Embedded. Connections involving an
// ephemeral proxy peer on either endpoint default to lazy, so this bit
// feeds the per-peer lazyState emitted in RemotePeerConfig.
bool proxy_embedded = 14;
}
// PolicyCompact is the compact form of a policy rule. Group references use
// the public_ids; the client resolves
// them against NetworkMapComponentsFull.groups. Direction is derived per-peer
// on the client (ingress when the peer is in destination_group_ids, egress
// when in source_group_ids; both when bidirectional).
message PolicyCompact {
// public_id. Used as a stable reference for
// ResourcePoliciesMap.indexes and future delta updates.
string id = 1;
RuleAction action = 2;
RuleProtocol protocol = 3;
bool bidirectional = 4;
// Single ports referenced by the rule.
repeated uint32 ports = 5;
// Port ranges (start..end) referenced by the rule.
repeated PortInfo.Range port_ranges = 6;
// Group ids (public_ids) of source / destination groups.
repeated string source_group_ids = 7;
repeated string destination_group_ids = 8;
// SSH authorization fields. PolicyRule.AuthorizedGroups maps the rule's
// applicable group ids (public_ids) to a list of local-user names —
// when a peer in one of those groups is the SSH destination, the named
// local users gain access. AuthorizedUser is the single-user form
// (legacy: rule scopes SSH to one specific user id).
//
// Both fields are only consumed by Calculate() when the rule's protocol
// is NetbirdSSH (or the legacy implicit-SSH heuristic).
map<string, UserNameList> authorized_groups = 9;
string authorized_user = 10;
// Resource-typed rule sources/destinations. When a rule targets a specific
// peer (rather than groups), Calculate() reads SourceResource /
// DestinationResource — without these the rule's connection resources
// can't be produced on the client. ResourceCompact's peer_index refers to
// NetworkMapComponentsFull.peers; type is the raw ResourceType string
// ("peer", "host", "subnet", "domain"). Only "peer" is meaningful for
// Calculate's resource-typed rule path today.
ResourceCompact source_resource = 11;
ResourceCompact destination_resource = 12;
// Posture-check ids gating this policy's source peers. Calculate()
// reads them when filtering rule peers (peers that fail any listed check
// are dropped from sourcePeers). Match keys in
// NetworkMapComponentsFull.posture_failed_peers.
repeated string source_posture_check_ids = 13;
}
// ResourceCompact mirrors types.Resource. Used by PolicyCompact to carry
// rule.SourceResource / rule.DestinationResource when the rule targets a
// specific resource (typically a peer) rather than groups.
message ResourceCompact {
string type = 1;
bool peer_index_set = 2;
uint32 peer_index = 3;
reserved 4;
string id = 5; // public id for domain/host/subnet resources
}
// UserNameList is a list of local-user names — used as the value type in
// PolicyCompact.authorized_groups.
message UserNameList {
repeated string names = 1;
}
// GroupCompact is the wire-shape of a group: public id, optional
// name, and indexes into NetworkMapComponentsFull.peers identifying members.
message GroupCompact {
// id comes from PublicID. Used by PolicyCompact.source_group_ids / destination_group_ids.
string id = 1;
// Indexes into NetworkMapComponentsFull.peers.
repeated uint32 peer_indexes = 2;
// True when the group is named "All" (types.Group.IsGroupAll). The
// client-side Calculate short-circuits group→peer expansion on such
// groups exactly like the server does; without this bit the decoded
// groups lose that property and the two sides expand policy
// destinations differently.
bool is_all = 3;
repeated ResourceCompact resources = 4;
}
// DNSSettingsCompact mirrors types.DNSSettings.
message DNSSettingsCompact {
// Group ids (public_id) whose DNS management is disabled.
repeated string disabled_management_group_ids = 1;
}
// RouteRaw mirrors *route.Route (the domain type), trimmed to fields that
// types.NetworkMapComponents.Calculate() reads. Group references are
// public_ids; the routing peer (when set) is referenced by index into
// NetworkMapComponentsFull.peers.
message RouteRaw {
string id = 1; // public_id
string net_id = 2;
string description = 3;
// Either network_cidr (e.g. "10.0.0.0/16") or domains is set, not both.
string network_cidr = 4;
repeated string domains = 5;
bool keep_route = 6;
// Routing peer reference: peer_index_set tells whether peer_index is valid
// (proto3 uint32 cannot disambiguate "0" from "unset"). Mutually exclusive
// with peer_group_ids.
//
// peer_index decodes back to types.Peer.ID (the peer's xid string), NOT
// to its WireGuard public key. This matches the server-side data flow:
// c.Routes carry route.Peer = peer.ID, and getRoutingPeerRoutes mutates
// it to peer.Key only after the route has been admitted to the network
// map. Decoders MUST set Route.Peer = peer.ID; the legacy Calculate()
// path will substitute the WG key downstream.
bool peer_index_set = 7;
uint32 peer_index = 8;
repeated string peer_group_ids = 9;
int32 network_type = 10;
bool masquerade = 11;
int32 metric = 12;
bool enabled = 13;
repeated string group_ids = 14;
repeated string access_control_group_ids = 15;
bool skip_auto_apply = 16;
}
// NameServerGroupRaw mirrors *nbdns.NameServerGroup. Distinct from the
// legacy NameServerGroup (which is the wire-trimmed shape consumed by
// proto.DNSConfig and lacks the Name/Description/Groups/Enabled fields).
message NameServerGroupRaw {
string id = 1;
// Reuses the legacy NameServer wire shape (IP as string).
repeated NameServer nameservers = 2;
// Group ids the NSG distributes nameservers to.
repeated string group_ids = 3;
bool primary = 4;
repeated string domains = 5;
bool enabled = 6;
bool search_domains_enabled = 7;
}
// NetworkResourceRaw mirrors *resourceTypes.NetworkResource.
//
message NetworkResourceRaw {
string id = 1;
string network_seq = 2;
string name = 3;
string description = 4;
// Resource type: "host" / "subnet" / "domain".
string type = 5;
string address = 6;
string domain_value = 7; // resource.Domain
string prefix_cidr = 8;
bool enabled = 9;
}
// NetworkRouterList carries the routers backing one network.
message NetworkRouterList {
// Routers in this network, keyed by peer_index (the routing peer).
repeated NetworkRouterEntry entries = 1;
}
// NetworkRouterEntry mirrors a single *routerTypes.NetworkRouter; the routing
// peer is referenced by index into NetworkMapComponentsFull.peers.
message NetworkRouterEntry {
string id = 1;
uint32 peer_index = 2;
bool peer_index_set = 3;
repeated string peer_group_ids = 4;
bool masquerade = 5;
int32 metric = 6;
bool enabled = 7;
}
message PolicyIds {
repeated string ids = 1;
}
// UserIDList is a list of user ids — used as the value type in
// NetworkMapComponentsFull.group_id_to_user_ids.
message UserIDList {
repeated string user_ids = 1;
}
// PeerIndexSet is a set of peer indexes — used as the value type in
// NetworkMapComponentsFull.posture_failed_peers.
message PeerIndexSet {
repeated uint32 peer_indexes = 1;
}
@@ -52,6 +52,13 @@ type ManagementServiceClient interface {
Logout(ctx context.Context, in *EncryptedMessage, opts ...grpc.CallOption) (*Empty, error)
// Executes a job on a target peer (e.g., debug bundle)
Job(ctx context.Context, opts ...grpc.CallOption) (ManagementService_JobClient, error)
// ExtendAuthSession refreshes the peer's session expiry deadline using a fresh JWT.
// Same JWT validation pipeline as Login (including jwt.UserID == peer.UserID check),
// but does not redo the network-map sync. Only valid for SSO-registered peers where
// login expiration is enabled. The tunnel remains up.
// EncryptedMessage of the request has a body of ExtendAuthSessionRequest.
// EncryptedMessage of the response has a body of ExtendAuthSessionResponse.
ExtendAuthSession(ctx context.Context, in *EncryptedMessage, opts ...grpc.CallOption) (*EncryptedMessage, error)
// CreateExpose creates a temporary reverse proxy service for a peer
CreateExpose(ctx context.Context, in *EncryptedMessage, opts ...grpc.CallOption) (*EncryptedMessage, error)
// RenewExpose extends the TTL of an active expose session
@@ -194,6 +201,15 @@ func (x *managementServiceJobClient) Recv() (*EncryptedMessage, error) {
return m, nil
}
func (c *managementServiceClient) ExtendAuthSession(ctx context.Context, in *EncryptedMessage, opts ...grpc.CallOption) (*EncryptedMessage, error) {
out := new(EncryptedMessage)
err := c.cc.Invoke(ctx, "/management.ManagementService/ExtendAuthSession", in, out, opts...)
if err != nil {
return nil, err
}
return out, nil
}
func (c *managementServiceClient) CreateExpose(ctx context.Context, in *EncryptedMessage, opts ...grpc.CallOption) (*EncryptedMessage, error) {
out := new(EncryptedMessage)
err := c.cc.Invoke(ctx, "/management.ManagementService/CreateExpose", in, out, opts...)
@@ -259,6 +275,13 @@ type ManagementServiceServer interface {
Logout(context.Context, *EncryptedMessage) (*Empty, error)
// Executes a job on a target peer (e.g., debug bundle)
Job(ManagementService_JobServer) error
// ExtendAuthSession refreshes the peer's session expiry deadline using a fresh JWT.
// Same JWT validation pipeline as Login (including jwt.UserID == peer.UserID check),
// but does not redo the network-map sync. Only valid for SSO-registered peers where
// login expiration is enabled. The tunnel remains up.
// EncryptedMessage of the request has a body of ExtendAuthSessionRequest.
// EncryptedMessage of the response has a body of ExtendAuthSessionResponse.
ExtendAuthSession(context.Context, *EncryptedMessage) (*EncryptedMessage, error)
// CreateExpose creates a temporary reverse proxy service for a peer
CreateExpose(context.Context, *EncryptedMessage) (*EncryptedMessage, error)
// RenewExpose extends the TTL of an active expose session
@@ -299,6 +322,9 @@ func (UnimplementedManagementServiceServer) Logout(context.Context, *EncryptedMe
func (UnimplementedManagementServiceServer) Job(ManagementService_JobServer) error {
return status.Errorf(codes.Unimplemented, "method Job not implemented")
}
func (UnimplementedManagementServiceServer) ExtendAuthSession(context.Context, *EncryptedMessage) (*EncryptedMessage, error) {
return nil, status.Errorf(codes.Unimplemented, "method ExtendAuthSession not implemented")
}
func (UnimplementedManagementServiceServer) CreateExpose(context.Context, *EncryptedMessage) (*EncryptedMessage, error) {
return nil, status.Errorf(codes.Unimplemented, "method CreateExpose not implemented")
}
@@ -494,6 +520,24 @@ func (x *managementServiceJobServer) Recv() (*EncryptedMessage, error) {
return m, nil
}
func _ManagementService_ExtendAuthSession_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(EncryptedMessage)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(ManagementServiceServer).ExtendAuthSession(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: "/management.ManagementService/ExtendAuthSession",
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(ManagementServiceServer).ExtendAuthSession(ctx, req.(*EncryptedMessage))
}
return interceptor(ctx, in, info, handler)
}
func _ManagementService_CreateExpose_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(EncryptedMessage)
if err := dec(in); err != nil {
@@ -583,6 +627,10 @@ var ManagementService_ServiceDesc = grpc.ServiceDesc{
MethodName: "Logout",
Handler: _ManagementService_Logout_Handler,
},
{
MethodName: "ExtendAuthSession",
Handler: _ManagementService_ExtendAuthSession_Handler,
},
{
MethodName: "CreateExpose",
Handler: _ManagementService_CreateExpose_Handler,
File diff suppressed because it is too large Load Diff
+261
View File
@@ -12,6 +12,15 @@ import "google/protobuf/timestamp.proto";
service ProxyService {
rpc GetMappingUpdate(GetMappingUpdateRequest) returns (stream GetMappingUpdateResponse);
// SyncMappings is a bidirectional stream that replaces GetMappingUpdate for
// new proxies. The proxy sends an initial SyncMappingsRequest to start the
// stream and then sends an ack after each batch is fully processed.
// Management waits for the ack before sending the next batch, providing
// application-level back-pressure during large initial syncs.
// Old proxies continue using GetMappingUpdate; old management servers
// return Unimplemented for this RPC and proxies fall back.
rpc SyncMappings(stream SyncMappingsRequest) returns (stream SyncMappingsResponse);
rpc SendAccessLog(SendAccessLogRequest) returns (SendAccessLogResponse);
rpc Authenticate(AuthenticateRequest) returns (AuthenticateResponse);
@@ -25,6 +34,27 @@ service ProxyService {
// ValidateSession validates a session token and checks user access permissions.
// Called by the proxy after receiving a session token from OIDC callback.
rpc ValidateSession(ValidateSessionRequest) returns (ValidateSessionResponse);
// ValidateTunnelPeer resolves an inbound peer by its WireGuard tunnel IP and
// checks the resolved user's access against the service's access_groups.
// Acts as a fast-path equivalent of OIDC for requests originating on the
// netbird mesh: when the source IP maps to a known peer in the calling
// account and that peer is in the service's access_groups, the proxy can
// issue a session cookie without redirecting through the OIDC flow.
// Mirrors ValidateSession's response shape.
rpc ValidateTunnelPeer(ValidateTunnelPeerRequest) returns (ValidateTunnelPeerResponse);
// CheckLLMPolicyLimits is the pre-flight RPC the proxy calls before each
// LLM request. Management runs the per-policy headroom selection across
// every policy authorising the caller's user / groups for the resolved
// provider and returns the chosen attribution policy + group, or a deny
// when no applicable policy has headroom > 0.
rpc CheckLLMPolicyLimits(CheckLLMPolicyLimitsRequest) returns (CheckLLMPolicyLimitsResponse);
// RecordLLMUsage is the post-flight RPC the proxy calls after the upstream
// returns. Increments the per-(dimension, window) counters for the
// attribution policy chosen by CheckLLMPolicyLimits.
rpc RecordLLMUsage(RecordLLMUsageRequest) returns (RecordLLMUsageResponse);
}
// ProxyCapabilities describes what a proxy can handle.
@@ -36,6 +66,13 @@ message ProxyCapabilities {
optional bool require_subdomain = 2;
// Whether the proxy has CrowdSec configured and can enforce IP reputation checks.
optional bool supports_crowdsec = 3;
// Whether the proxy is running embedded in the netbird client and serving
// exclusively over the WireGuard tunnel (i.e. `netbird proxy` rather than
// the standalone netbird-proxy binary). Surfaces upstream so dashboards can
// distinguish per-peer / private clusters from centralised ones.
optional bool private = 4;
// Whether the proxy enforces ProxyMapping.private (fails closed on ValidateTunnelPeer failure). Management MUST NOT stream private mappings to proxies that don't claim this.
optional bool supports_private_service = 5;
}
// GetMappingUpdateRequest is sent to initialise a mapping stream.
@@ -77,6 +114,64 @@ message PathTargetOptions {
bool proxy_protocol = 5;
// Idle timeout before a UDP session is reaped.
google.protobuf.Duration session_idle_timeout = 6;
// When true, the proxy dials this target via the host's network stack
// instead of through the embedded NetBird client. Useful for upstreams
// reachable without WireGuard (public APIs, LAN services, localhost
// sidecars). Defaults to false — embedded client is the standard path.
bool direct_upstream = 7;
// Proxy clamps to [0, proxy-wide max (1 MiB)] at apply time. Agent-network
// synthesized targets only; private services leave these zero.
int64 capture_max_request_bytes = 8;
// Proxy clamps to [0, proxy-wide max (1 MiB)] at apply time.
int64 capture_max_response_bytes = 9;
// Content types eligible for body capture (e.g. "application/json").
repeated string capture_content_types = 10;
// Per-target middleware configurations populated by the agent-network
// synthesizer. Validated and clamped by the proxy at apply time.
repeated MiddlewareConfig middlewares = 11;
// When true, the proxy stamps agent_network=true on access-log entries
// for this target so management routes them to the agent-network log
// surface.
bool agent_network = 12;
// When true, the proxy suppresses the per-request access-log emission for
// this target. Defaults false to preserve existing access-log behavior for
// every non-agent-network target. The agent-network synth target sets this
// true only when the account's EnableLogCollection toggle is off.
bool disable_access_log = 13;
}
// MiddlewareSlot identifies where in the request lifecycle a middleware
// runs. Mirrors proxy/internal/middleware.Slot.
enum MiddlewareSlot {
MIDDLEWARE_SLOT_UNSPECIFIED = 0;
MIDDLEWARE_SLOT_ON_REQUEST = 1;
MIDDLEWARE_SLOT_ON_RESPONSE = 2;
MIDDLEWARE_SLOT_TERMINAL = 3;
}
// MiddlewareConfig is the per-target configuration for a single middleware.
// The proxy validates every incoming MiddlewareConfig at apply time:
// unknown ids are rejected, timeout is clamped to [10ms, 5s], and the
// declared slot must match the registered middleware's slot.
message MiddlewareConfig {
// Middleware id; must match the proxy-local compiled-in registry.
string id = 1;
bool enabled = 2;
MiddlewareSlot slot = 3;
// Free-form JSON unmarshalled by the middleware factory into its own typed
// config struct. Empty / null / {} are valid (zero-value config).
bytes config_json = 4;
enum FailMode {
FAIL_OPEN = 0;
FAIL_CLOSED = 1;
}
FailMode fail_mode = 5;
// Clamped to [10ms, 5s] at apply time; zero → 500ms default.
google.protobuf.Duration timeout = 6;
// When true, the middleware may mutate request headers or body (subject to
// policy). Honoured only when the implementation also declares
// MutationsSupported.
bool can_mutate = 7;
}
message PathMapping {
@@ -99,6 +194,12 @@ message Authentication {
bool pin = 4;
bool oidc = 5;
repeated HeaderAuth header_auths = 6;
// Group ids allowed to reach the service through an OIDC identity. When
// non-empty the proxy requires the session token's groups claim to
// intersect this list before honouring the cookie, so a token minted for
// an identity outside these groups is not a bearer credential for the
// service. Empty means group membership does not restrict access.
repeated string allowed_group_ids = 7;
}
message AccessRestrictions {
@@ -129,6 +230,8 @@ message ProxyMapping {
// For L4/TLS: the port the proxy listens on.
int32 listen_port = 11;
AccessRestrictions access_restrictions = 12;
// NetBird-only: the proxy MUST call ValidateTunnelPeer and fail closed; operator auth schemes are bypassed.
bool private = 13;
}
// SendAccessLogRequest consists of one or more AccessLogs from a Proxy.
@@ -158,6 +261,10 @@ message AccessLog {
string protocol = 16;
// Extra key-value metadata for the access log entry (e.g. crowdsec_verdict, scenario).
map<string, string> metadata = 17;
// When true, the entry was emitted by an agent-network synth service.
// Management routes these to the agent-network access-log surface instead
// of the standard service log.
bool agent_network = 18;
}
message AuthenticateRequest {
@@ -204,6 +311,25 @@ message SendStatusUpdateRequest {
ProxyStatus status = 3;
bool certificate_issued = 4;
optional string error_message = 5;
// Per-account inbound listener state for the account that owns
// service_id. Populated only when --private is enabled and the
// embedded client for the account is up. Field numbers >=50 reserved
// for observability extensions.
optional ProxyInboundListener inbound_listener = 50;
}
// ProxyInboundListener describes a per-account inbound listener that the
// proxy has bound on the embedded netstack of the account's WireGuard
// client. Surfaced so dashboards can render "this account is reachable
// at <tunnel_ip>:<https_port> on this proxy".
message ProxyInboundListener {
// Tunnel IP the embedded netstack listens on. Same address other peers
// in the account see for the proxy peer.
string tunnel_ip = 1;
// TLS port served on tunnel_ip (auto-detected, default 443).
uint32 https_port = 2;
// Plain-HTTP port served on tunnel_ip (auto-detected, default 80).
uint32 http_port = 3;
}
// SendStatusUpdateResponse is intentionally empty to allow for future expansion
@@ -245,4 +371,139 @@ message ValidateSessionResponse {
string user_id = 2;
string user_email = 3;
string denied_reason = 4;
// peer_group_ids carries the calling user's group memberships so the
// proxy can authorise policy-aware middlewares without an additional
// management round-trip.
repeated string peer_group_ids = 5;
// peer_group_names carries the human-readable display names for the
// ids in peer_group_ids, ordered identically (positional pairing).
// Stamped onto upstream requests as X-NetBird-Groups so downstream
// services can read names rather than opaque ids.
repeated string peer_group_names = 6;
}
// ValidateTunnelPeerRequest carries the inbound peer's tunnel IP and the
// service domain whose group requirements should gate access. The calling
// account is inferred from the proxy's gRPC metadata (ProxyToken).
message ValidateTunnelPeerRequest {
string tunnel_ip = 1;
string domain = 2;
}
// ValidateTunnelPeerResponse mirrors ValidateSessionResponse plus a freshly
// minted session_token: when valid is true, the proxy installs the token as
// a session cookie so subsequent requests skip the management round-trip,
// matching the OIDC flow's UX. denied_reason values:
// "peer_not_found" — no peer with that tunnel IP in the calling account
// "no_user" — peer exists but is not bound to a user
// "service_not_found"
// "account_mismatch"
// "not_in_group" — peer resolved but not in service.access_groups
message ValidateTunnelPeerResponse {
bool valid = 1;
string user_id = 2;
string user_email = 3;
string denied_reason = 4;
// session_token is set only when valid is true. Same shape as the JWT
// the OIDC flow produces — proxy installs it via setSessionCookie so the
// tunnel fast-path is indistinguishable from OIDC for subsequent requests.
string session_token = 5;
// peer_group_ids carries the resolved peer's user group memberships so
// the proxy can authorise policy-aware middlewares without an additional
// management round-trip.
repeated string peer_group_ids = 6;
// peer_group_names carries the human-readable display names for the
// ids in peer_group_ids, ordered identically (positional pairing).
// Stamped onto upstream requests as X-NetBird-Groups so downstream
// services can read names rather than opaque ids.
repeated string peer_group_names = 7;
}
// SyncMappingsRequest is sent by the proxy on the bidirectional SyncMappings
// stream. The first message MUST be an init; all subsequent messages MUST be
// acks.
message SyncMappingsRequest {
oneof msg {
SyncMappingsInit init = 1;
SyncMappingsAck ack = 2;
}
}
// SyncMappingsInit is the first message on the stream, carrying the same
// identification fields as GetMappingUpdateRequest.
message SyncMappingsInit {
string proxy_id = 1;
string version = 2;
google.protobuf.Timestamp started_at = 3;
string address = 4;
ProxyCapabilities capabilities = 5;
}
// SyncMappingsAck is sent by the proxy after it has fully processed a batch.
// Management waits for this before sending the next batch.
message SyncMappingsAck {}
// SyncMappingsResponse is a batch of mappings sent by management.
// Identical semantics to GetMappingUpdateResponse.
message SyncMappingsResponse {
repeated ProxyMapping mapping = 1;
// initial_sync_complete is set on the last message of the initial snapshot.
bool initial_sync_complete = 2;
}
// CheckLLMPolicyLimitsRequest carries the resolved caller identity and the
// upstream provider already chosen by llm_router. Management computes which
// policies authorise the request, picks the one with the most remaining
// headroom, and returns the attribution decision.
message CheckLLMPolicyLimitsRequest {
// account_id is the netbird account the request belongs to.
string account_id = 1;
// user_id is the netbird user id of the caller. May be empty when the
// principal is a tunnel-peer that isn't bound to a user; group membership
// still gates the request in that case.
string user_id = 2;
// group_ids is the caller's full group membership at request time.
repeated string group_ids = 3;
// provider_id is the agent-network provider record id chosen by llm_router.
string provider_id = 4;
// model is the upstream model identifier extracted from the request body.
string model = 5;
}
// CheckLLMPolicyLimitsResponse is management's allow-or-deny decision for a
// pre-flight check.
message CheckLLMPolicyLimitsResponse {
// decision is "allow" or "deny".
string decision = 1;
// selected_policy_id names the policy that paid for this request.
string selected_policy_id = 2;
// attribution_group_id is the source group the request booked against.
string attribution_group_id = 3;
// window_seconds is the cap window length the selected policy uses.
int64 window_seconds = 4;
// deny_code is set on decision="deny" with a stable label.
string deny_code = 5;
// deny_reason is a short human-readable explanation paired with deny_code.
string deny_reason = 6;
}
// RecordLLMUsageRequest is the post-flight increment the proxy posts after
// the upstream call. Counters are keyed on (account, dimension, window).
message RecordLLMUsageRequest {
string account_id = 1;
string user_id = 2;
// group_id is the selected policy's attribution group, recorded against the
// policy window (window_seconds).
string group_id = 3;
int64 window_seconds = 4;
int64 tokens_input = 5;
int64 tokens_output = 6;
double cost_usd = 7;
// group_ids is the caller's full group membership, used to fan the same
// usage out to every applicable account-level budget rule's own window.
repeated string group_ids = 8;
}
message RecordLLMUsageResponse {
}
@@ -19,6 +19,14 @@ const _ = grpc.SupportPackageIsVersion7
// For semantics around ctx use and closing/ending streaming RPCs, please refer to https://pkg.go.dev/google.golang.org/grpc/?tab=doc#ClientConn.NewStream.
type ProxyServiceClient interface {
GetMappingUpdate(ctx context.Context, in *GetMappingUpdateRequest, opts ...grpc.CallOption) (ProxyService_GetMappingUpdateClient, error)
// SyncMappings is a bidirectional stream that replaces GetMappingUpdate for
// new proxies. The proxy sends an initial SyncMappingsRequest to start the
// stream and then sends an ack after each batch is fully processed.
// Management waits for the ack before sending the next batch, providing
// application-level back-pressure during large initial syncs.
// Old proxies continue using GetMappingUpdate; old management servers
// return Unimplemented for this RPC and proxies fall back.
SyncMappings(ctx context.Context, opts ...grpc.CallOption) (ProxyService_SyncMappingsClient, error)
SendAccessLog(ctx context.Context, in *SendAccessLogRequest, opts ...grpc.CallOption) (*SendAccessLogResponse, error)
Authenticate(ctx context.Context, in *AuthenticateRequest, opts ...grpc.CallOption) (*AuthenticateResponse, error)
SendStatusUpdate(ctx context.Context, in *SendStatusUpdateRequest, opts ...grpc.CallOption) (*SendStatusUpdateResponse, error)
@@ -27,6 +35,24 @@ type ProxyServiceClient interface {
// ValidateSession validates a session token and checks user access permissions.
// Called by the proxy after receiving a session token from OIDC callback.
ValidateSession(ctx context.Context, in *ValidateSessionRequest, opts ...grpc.CallOption) (*ValidateSessionResponse, error)
// ValidateTunnelPeer resolves an inbound peer by its WireGuard tunnel IP and
// checks the resolved user's access against the service's access_groups.
// Acts as a fast-path equivalent of OIDC for requests originating on the
// netbird mesh: when the source IP maps to a known peer in the calling
// account and that peer is in the service's access_groups, the proxy can
// issue a session cookie without redirecting through the OIDC flow.
// Mirrors ValidateSession's response shape.
ValidateTunnelPeer(ctx context.Context, in *ValidateTunnelPeerRequest, opts ...grpc.CallOption) (*ValidateTunnelPeerResponse, error)
// CheckLLMPolicyLimits is the pre-flight RPC the proxy calls before each
// LLM request. Management runs the per-policy headroom selection across
// every policy authorising the caller's user / groups for the resolved
// provider and returns the chosen attribution policy + group, or a deny
// when no applicable policy has headroom > 0.
CheckLLMPolicyLimits(ctx context.Context, in *CheckLLMPolicyLimitsRequest, opts ...grpc.CallOption) (*CheckLLMPolicyLimitsResponse, error)
// RecordLLMUsage is the post-flight RPC the proxy calls after the upstream
// returns. Increments the per-(dimension, window) counters for the
// attribution policy chosen by CheckLLMPolicyLimits.
RecordLLMUsage(ctx context.Context, in *RecordLLMUsageRequest, opts ...grpc.CallOption) (*RecordLLMUsageResponse, error)
}
type proxyServiceClient struct {
@@ -69,6 +95,37 @@ func (x *proxyServiceGetMappingUpdateClient) Recv() (*GetMappingUpdateResponse,
return m, nil
}
func (c *proxyServiceClient) SyncMappings(ctx context.Context, opts ...grpc.CallOption) (ProxyService_SyncMappingsClient, error) {
stream, err := c.cc.NewStream(ctx, &ProxyService_ServiceDesc.Streams[1], "/management.ProxyService/SyncMappings", opts...)
if err != nil {
return nil, err
}
x := &proxyServiceSyncMappingsClient{stream}
return x, nil
}
type ProxyService_SyncMappingsClient interface {
Send(*SyncMappingsRequest) error
Recv() (*SyncMappingsResponse, error)
grpc.ClientStream
}
type proxyServiceSyncMappingsClient struct {
grpc.ClientStream
}
func (x *proxyServiceSyncMappingsClient) Send(m *SyncMappingsRequest) error {
return x.ClientStream.SendMsg(m)
}
func (x *proxyServiceSyncMappingsClient) Recv() (*SyncMappingsResponse, error) {
m := new(SyncMappingsResponse)
if err := x.ClientStream.RecvMsg(m); err != nil {
return nil, err
}
return m, nil
}
func (c *proxyServiceClient) SendAccessLog(ctx context.Context, in *SendAccessLogRequest, opts ...grpc.CallOption) (*SendAccessLogResponse, error) {
out := new(SendAccessLogResponse)
err := c.cc.Invoke(ctx, "/management.ProxyService/SendAccessLog", in, out, opts...)
@@ -123,11 +180,46 @@ func (c *proxyServiceClient) ValidateSession(ctx context.Context, in *ValidateSe
return out, nil
}
func (c *proxyServiceClient) ValidateTunnelPeer(ctx context.Context, in *ValidateTunnelPeerRequest, opts ...grpc.CallOption) (*ValidateTunnelPeerResponse, error) {
out := new(ValidateTunnelPeerResponse)
err := c.cc.Invoke(ctx, "/management.ProxyService/ValidateTunnelPeer", in, out, opts...)
if err != nil {
return nil, err
}
return out, nil
}
func (c *proxyServiceClient) CheckLLMPolicyLimits(ctx context.Context, in *CheckLLMPolicyLimitsRequest, opts ...grpc.CallOption) (*CheckLLMPolicyLimitsResponse, error) {
out := new(CheckLLMPolicyLimitsResponse)
err := c.cc.Invoke(ctx, "/management.ProxyService/CheckLLMPolicyLimits", in, out, opts...)
if err != nil {
return nil, err
}
return out, nil
}
func (c *proxyServiceClient) RecordLLMUsage(ctx context.Context, in *RecordLLMUsageRequest, opts ...grpc.CallOption) (*RecordLLMUsageResponse, error) {
out := new(RecordLLMUsageResponse)
err := c.cc.Invoke(ctx, "/management.ProxyService/RecordLLMUsage", in, out, opts...)
if err != nil {
return nil, err
}
return out, nil
}
// ProxyServiceServer is the server API for ProxyService service.
// All implementations must embed UnimplementedProxyServiceServer
// for forward compatibility
type ProxyServiceServer interface {
GetMappingUpdate(*GetMappingUpdateRequest, ProxyService_GetMappingUpdateServer) error
// SyncMappings is a bidirectional stream that replaces GetMappingUpdate for
// new proxies. The proxy sends an initial SyncMappingsRequest to start the
// stream and then sends an ack after each batch is fully processed.
// Management waits for the ack before sending the next batch, providing
// application-level back-pressure during large initial syncs.
// Old proxies continue using GetMappingUpdate; old management servers
// return Unimplemented for this RPC and proxies fall back.
SyncMappings(ProxyService_SyncMappingsServer) error
SendAccessLog(context.Context, *SendAccessLogRequest) (*SendAccessLogResponse, error)
Authenticate(context.Context, *AuthenticateRequest) (*AuthenticateResponse, error)
SendStatusUpdate(context.Context, *SendStatusUpdateRequest) (*SendStatusUpdateResponse, error)
@@ -136,6 +228,24 @@ type ProxyServiceServer interface {
// ValidateSession validates a session token and checks user access permissions.
// Called by the proxy after receiving a session token from OIDC callback.
ValidateSession(context.Context, *ValidateSessionRequest) (*ValidateSessionResponse, error)
// ValidateTunnelPeer resolves an inbound peer by its WireGuard tunnel IP and
// checks the resolved user's access against the service's access_groups.
// Acts as a fast-path equivalent of OIDC for requests originating on the
// netbird mesh: when the source IP maps to a known peer in the calling
// account and that peer is in the service's access_groups, the proxy can
// issue a session cookie without redirecting through the OIDC flow.
// Mirrors ValidateSession's response shape.
ValidateTunnelPeer(context.Context, *ValidateTunnelPeerRequest) (*ValidateTunnelPeerResponse, error)
// CheckLLMPolicyLimits is the pre-flight RPC the proxy calls before each
// LLM request. Management runs the per-policy headroom selection across
// every policy authorising the caller's user / groups for the resolved
// provider and returns the chosen attribution policy + group, or a deny
// when no applicable policy has headroom > 0.
CheckLLMPolicyLimits(context.Context, *CheckLLMPolicyLimitsRequest) (*CheckLLMPolicyLimitsResponse, error)
// RecordLLMUsage is the post-flight RPC the proxy calls after the upstream
// returns. Increments the per-(dimension, window) counters for the
// attribution policy chosen by CheckLLMPolicyLimits.
RecordLLMUsage(context.Context, *RecordLLMUsageRequest) (*RecordLLMUsageResponse, error)
mustEmbedUnimplementedProxyServiceServer()
}
@@ -146,6 +256,9 @@ type UnimplementedProxyServiceServer struct {
func (UnimplementedProxyServiceServer) GetMappingUpdate(*GetMappingUpdateRequest, ProxyService_GetMappingUpdateServer) error {
return status.Errorf(codes.Unimplemented, "method GetMappingUpdate not implemented")
}
func (UnimplementedProxyServiceServer) SyncMappings(ProxyService_SyncMappingsServer) error {
return status.Errorf(codes.Unimplemented, "method SyncMappings not implemented")
}
func (UnimplementedProxyServiceServer) SendAccessLog(context.Context, *SendAccessLogRequest) (*SendAccessLogResponse, error) {
return nil, status.Errorf(codes.Unimplemented, "method SendAccessLog not implemented")
}
@@ -164,6 +277,15 @@ func (UnimplementedProxyServiceServer) GetOIDCURL(context.Context, *GetOIDCURLRe
func (UnimplementedProxyServiceServer) ValidateSession(context.Context, *ValidateSessionRequest) (*ValidateSessionResponse, error) {
return nil, status.Errorf(codes.Unimplemented, "method ValidateSession not implemented")
}
func (UnimplementedProxyServiceServer) ValidateTunnelPeer(context.Context, *ValidateTunnelPeerRequest) (*ValidateTunnelPeerResponse, error) {
return nil, status.Errorf(codes.Unimplemented, "method ValidateTunnelPeer not implemented")
}
func (UnimplementedProxyServiceServer) CheckLLMPolicyLimits(context.Context, *CheckLLMPolicyLimitsRequest) (*CheckLLMPolicyLimitsResponse, error) {
return nil, status.Errorf(codes.Unimplemented, "method CheckLLMPolicyLimits not implemented")
}
func (UnimplementedProxyServiceServer) RecordLLMUsage(context.Context, *RecordLLMUsageRequest) (*RecordLLMUsageResponse, error) {
return nil, status.Errorf(codes.Unimplemented, "method RecordLLMUsage not implemented")
}
func (UnimplementedProxyServiceServer) mustEmbedUnimplementedProxyServiceServer() {}
// UnsafeProxyServiceServer may be embedded to opt out of forward compatibility for this service.
@@ -198,6 +320,32 @@ func (x *proxyServiceGetMappingUpdateServer) Send(m *GetMappingUpdateResponse) e
return x.ServerStream.SendMsg(m)
}
func _ProxyService_SyncMappings_Handler(srv interface{}, stream grpc.ServerStream) error {
return srv.(ProxyServiceServer).SyncMappings(&proxyServiceSyncMappingsServer{stream})
}
type ProxyService_SyncMappingsServer interface {
Send(*SyncMappingsResponse) error
Recv() (*SyncMappingsRequest, error)
grpc.ServerStream
}
type proxyServiceSyncMappingsServer struct {
grpc.ServerStream
}
func (x *proxyServiceSyncMappingsServer) Send(m *SyncMappingsResponse) error {
return x.ServerStream.SendMsg(m)
}
func (x *proxyServiceSyncMappingsServer) Recv() (*SyncMappingsRequest, error) {
m := new(SyncMappingsRequest)
if err := x.ServerStream.RecvMsg(m); err != nil {
return nil, err
}
return m, nil
}
func _ProxyService_SendAccessLog_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(SendAccessLogRequest)
if err := dec(in); err != nil {
@@ -306,6 +454,60 @@ func _ProxyService_ValidateSession_Handler(srv interface{}, ctx context.Context,
return interceptor(ctx, in, info, handler)
}
func _ProxyService_ValidateTunnelPeer_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(ValidateTunnelPeerRequest)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(ProxyServiceServer).ValidateTunnelPeer(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: "/management.ProxyService/ValidateTunnelPeer",
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(ProxyServiceServer).ValidateTunnelPeer(ctx, req.(*ValidateTunnelPeerRequest))
}
return interceptor(ctx, in, info, handler)
}
func _ProxyService_CheckLLMPolicyLimits_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(CheckLLMPolicyLimitsRequest)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(ProxyServiceServer).CheckLLMPolicyLimits(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: "/management.ProxyService/CheckLLMPolicyLimits",
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(ProxyServiceServer).CheckLLMPolicyLimits(ctx, req.(*CheckLLMPolicyLimitsRequest))
}
return interceptor(ctx, in, info, handler)
}
func _ProxyService_RecordLLMUsage_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(RecordLLMUsageRequest)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(ProxyServiceServer).RecordLLMUsage(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: "/management.ProxyService/RecordLLMUsage",
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(ProxyServiceServer).RecordLLMUsage(ctx, req.(*RecordLLMUsageRequest))
}
return interceptor(ctx, in, info, handler)
}
// ProxyService_ServiceDesc is the grpc.ServiceDesc for ProxyService service.
// It's only intended for direct use with grpc.RegisterService,
// and not to be introspected or modified (even as a copy)
@@ -337,6 +539,18 @@ var ProxyService_ServiceDesc = grpc.ServiceDesc{
MethodName: "ValidateSession",
Handler: _ProxyService_ValidateSession_Handler,
},
{
MethodName: "ValidateTunnelPeer",
Handler: _ProxyService_ValidateTunnelPeer_Handler,
},
{
MethodName: "CheckLLMPolicyLimits",
Handler: _ProxyService_CheckLLMPolicyLimits_Handler,
},
{
MethodName: "RecordLLMUsage",
Handler: _ProxyService_RecordLLMUsage_Handler,
},
},
Streams: []grpc.StreamDesc{
{
@@ -344,6 +558,12 @@ var ProxyService_ServiceDesc = grpc.ServiceDesc{
Handler: _ProxyService_GetMappingUpdate_Handler,
ServerStreams: true,
},
{
StreamName: "SyncMappings",
Handler: _ProxyService_SyncMappings_Handler,
ServerStreams: true,
ClientStreams: true,
},
},
Metadata: "proxy_service.proto",
}
+34
View File
@@ -48,6 +48,10 @@ type Type int32
var (
ErrExtraSettingsNotFound = errors.New("extra settings not found")
ErrPeerAlreadyLoggedIn = errors.New("peer with the same public key is already logged in")
// ErrNoAuthMethodProvided is returned when a peer login attempt carries neither a
// setup key nor an SSO token. Match it with errors.Is.
ErrNoAuthMethodProvided = Errorf(Unauthenticated, "no peer auth method provided, please use a setup key or interactive SSO login")
)
// Error is an internal error
@@ -66,6 +70,16 @@ func (e *Error) Error() string {
return e.Message
}
// Is reports whether target is an *Error with the same type and message,
// enabling matching with errors.Is against sentinel errors.
func (e *Error) Is(target error) bool {
var t *Error
if !errors.As(target, &t) {
return false
}
return e.ErrorType == t.ErrorType && e.Message == t.Message
}
// Errorf returns Error(ErrorType, fmt.Sprintf(format, a...)).
func Errorf(errorType Type, format string, a ...interface{}) error {
return &Error{
@@ -205,6 +219,26 @@ func NewNetworkResourceNotFoundError(resourceID string) error {
return Errorf(NotFound, "network resource: %s not found", resourceID)
}
// NewAgentNetworkProviderNotFoundError creates a new Error with NotFound type for a missing Agent Network provider.
func NewAgentNetworkProviderNotFoundError(providerID string) error {
return Errorf(NotFound, "agent network provider: %s not found", providerID)
}
// NewAgentNetworkPolicyNotFoundError creates a new Error with NotFound type for a missing Agent Network policy.
func NewAgentNetworkPolicyNotFoundError(policyID string) error {
return Errorf(NotFound, "agent network policy: %s not found", policyID)
}
// NewAgentNetworkGuardrailNotFoundError creates a new Error with NotFound type for a missing Agent Network guardrail.
func NewAgentNetworkGuardrailNotFoundError(guardrailID string) error {
return Errorf(NotFound, "agent network guardrail: %s not found", guardrailID)
}
// NewAgentNetworkBudgetRuleNotFoundError creates a new Error with NotFound type for a missing Agent Network budget rule.
func NewAgentNetworkBudgetRuleNotFoundError(ruleID string) error {
return Errorf(NotFound, "agent network budget rule: %s not found", ruleID)
}
// NewPermissionDeniedError creates a new Error with PermissionDenied type for a permission denied error.
func NewPermissionDeniedError() error {
return Errorf(PermissionDenied, "permission denied")
+108
View File
@@ -0,0 +1,108 @@
package types
import (
"strconv"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/version"
)
const (
firewallRuleMinPortRangesVer = "0.48.0"
firewallRuleMinNativeSSHVer = "0.60.0"
nativeSSHPortString = "22022"
nativeSSHPortNumber = 22022
defaultSSHPortString = "22"
defaultSSHPortNumber = 22
)
type supportedFeatures struct {
nativeSSH bool
portRanges bool
}
type LookupMap map[string]struct{}
// ExpandPortsAndRanges expands Ports and PortRanges of a rule into individual firewall rules.
func ExpandPortsAndRanges(base FirewallRule, rule *nmdata.PolicyRule, peer *nmdata.Peer) []*FirewallRule {
features := peerSupportedFirewallFeatures(peer.Meta.WtVersion)
var expanded []*FirewallRule
for _, port := range rule.Ports {
fr := base
fr.Port = port
expanded = append(expanded, &fr)
}
for _, portRange := range rule.PortRanges {
if len(rule.Ports) > 0 {
break
}
fr := base
if features.portRanges {
fr.PortRange = RulePortRange{Start: portRange.Start, End: portRange.End}
} else {
if portRange.Start != portRange.End {
continue
}
fr.Port = strconv.FormatUint(uint64(portRange.Start), 10)
}
expanded = append(expanded, &fr)
}
if shouldCheckRulesForNativeSSH(features.nativeSSH, rule, peer) || rule.Protocol == string(PolicyRuleProtocolNetbirdSSH) {
expanded = addNativeSSHRule(base, expanded)
}
return expanded
}
func addNativeSSHRule(base FirewallRule, expanded []*FirewallRule) []*FirewallRule {
shouldAdd := false
for _, fr := range expanded {
if isPortInRule(nativeSSHPortString, 22022, fr) {
return expanded
}
if isPortInRule(defaultSSHPortString, 22, fr) {
shouldAdd = true
}
}
if !shouldAdd {
return expanded
}
fr := base
fr.Port = nativeSSHPortString
return append(expanded, &fr)
}
func isPortInRule(portString string, portInt uint16, rule *FirewallRule) bool {
return rule.Port == portString || (rule.PortRange.Start <= portInt && portInt <= rule.PortRange.End)
}
func shouldCheckRulesForNativeSSH(supportsNative bool, rule *nmdata.PolicyRule, peer *nmdata.Peer) bool {
return supportsNative && peer.SSHEnabled && peer.Meta.Flags.ServerSSHAllowed && rule.Protocol == string(PolicyRuleProtocolTCP)
}
func peerSupportedFirewallFeatures(peerVer string) supportedFeatures {
if version.IsDevelopmentVersion(peerVer) {
return supportedFeatures{true, true}
}
var features supportedFeatures
meetMinVer, err := version.MeetsMinVersion(firewallRuleMinNativeSSHVer, peerVer)
features.nativeSSH = err == nil && meetMinVer
if features.nativeSSH {
features.portRanges = true
} else {
meetMinVer, err = version.MeetsMinVersion(firewallRuleMinPortRangesVer, peerVer)
features.portRanges = err == nil && meetMinVer
}
return features
}
+182
View File
@@ -0,0 +1,182 @@
package types
import (
"context"
"fmt"
"reflect"
"strconv"
"strings"
log "github.com/sirupsen/logrus"
nbroute "github.com/netbirdio/netbird/route"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
const (
FirewallRuleDirectionIN = 0
FirewallRuleDirectionOUT = 1
)
// FirewallRule is a rule of the firewall.
type FirewallRule struct {
// PolicyID is the ID of the policy this rule is derived from
PolicyID string
// PeerIP of the peer
PeerIP string
// Direction of the traffic
Direction int
// Action of the traffic
Action string
// Protocol of the traffic
Protocol string
// Port of the traffic
Port string
// PortRange represents the range of ports for a firewall rule
PortRange RulePortRange
}
// Equal checks if two firewall rules are equal.
func (r *FirewallRule) Equal(other *FirewallRule) bool {
return reflect.DeepEqual(r, other)
}
// GenerateRouteFirewallRules generates a list of firewall rules for a given route.
// For static routes, source ranges match the destination family (v4 or v6).
// For dynamic routes (domain-based), separate v4 and v6 rules are generated
// so the routing peer's forwarding chain allows both address families.
func GenerateRouteFirewallRules(ctx context.Context, route *nmdata.Route, rule *nmdata.PolicyRule, groupPeers []*nmdata.Peer, direction int, includeIPv6 bool) []*RouteFirewallRule {
rulesExists := make(map[string]struct{})
rules := make([]*RouteFirewallRule, 0)
v4Sources, v6Sources := splitPeerSourcesByFamily(groupPeers)
isV6Route := route.Network.Addr().Is6()
// Skip v6 destination routes entirely for peers without IPv6 support
if isV6Route && !includeIPv6 {
return rules
}
// Pick sources matching the destination family
sourceRanges := v4Sources
if isV6Route {
sourceRanges = v6Sources
}
baseRule := RouteFirewallRule{
PolicyID: rule.PolicyID,
RouteID: nbroute.ID(route.ID),
SourceRanges: sourceRanges,
Action: rule.Action,
Destination: route.Network.String(),
Protocol: rule.Protocol,
Domains: route.Domains,
IsDynamic: route.IsDynamic(),
}
if len(rule.Ports) == 0 {
rules = append(rules, generateRulesWithPortRanges(baseRule, rule, rulesExists)...)
} else {
rules = append(rules, generateRulesWithPorts(ctx, baseRule, rule, rulesExists)...)
}
// Generate v6 counterpart for dynamic routes and 0.0.0.0/0 exit node routes.
isDefaultV4 := !isV6Route && route.Network.Bits() == 0
if includeIPv6 && (route.IsDynamic() || isDefaultV4) && len(v6Sources) > 0 {
v6Rule := baseRule
v6Rule.SourceRanges = v6Sources
if isDefaultV4 {
v6Rule.Destination = "::/0"
v6Rule.RouteID = nbroute.ID(route.ID + "-v6-default")
}
if len(rule.Ports) == 0 {
rules = append(rules, generateRulesWithPortRanges(v6Rule, rule, rulesExists)...)
} else {
rules = append(rules, generateRulesWithPorts(ctx, v6Rule, rule, rulesExists)...)
}
}
return rules
}
// splitPeerSourcesByFamily separates peer IPs into v4 (/32) and v6 (/128) source ranges.
func splitPeerSourcesByFamily(groupPeers []*nmdata.Peer) (v4, v6 []string) {
v4 = make([]string, 0, len(groupPeers))
v6 = make([]string, 0, len(groupPeers))
for _, peer := range groupPeers {
if peer == nil {
continue
}
v4 = append(v4, fmt.Sprintf(AllowedIPsFormat, peer.IP))
if peer.IPv6.IsValid() {
v6 = append(v6, fmt.Sprintf(AllowedIPsV6Format, peer.IPv6))
}
}
return
}
// generateRulesForPeer generates rules for a given peer based on ports and port ranges.
func generateRulesWithPortRanges(baseRule RouteFirewallRule, rule *nmdata.PolicyRule, rulesExists map[string]struct{}) []*RouteFirewallRule {
rules := make([]*RouteFirewallRule, 0)
ruleIDBase := generateRuleIDBase(rule, baseRule)
if len(rule.Ports) == 0 {
if len(rule.PortRanges) == 0 {
if _, ok := rulesExists[ruleIDBase]; !ok {
rulesExists[ruleIDBase] = struct{}{}
rules = append(rules, &baseRule)
}
} else {
for _, portRange := range rule.PortRanges {
ruleID := fmt.Sprintf("%s%d-%d", ruleIDBase, portRange.Start, portRange.End)
if _, ok := rulesExists[ruleID]; !ok {
rulesExists[ruleID] = struct{}{}
pr := baseRule
pr.PortRange = RulePortRange{Start: portRange.Start, End: portRange.End}
rules = append(rules, &pr)
}
}
}
return rules
}
return rules
}
// generateRulesWithPorts generates rules when specific ports are provided.
func generateRulesWithPorts(ctx context.Context, baseRule RouteFirewallRule, rule *nmdata.PolicyRule, rulesExists map[string]struct{}) []*RouteFirewallRule {
rules := make([]*RouteFirewallRule, 0)
ruleIDBase := generateRuleIDBase(rule, baseRule)
for _, port := range rule.Ports {
ruleID := ruleIDBase + port
if _, ok := rulesExists[ruleID]; ok {
continue
}
rulesExists[ruleID] = struct{}{}
pr := baseRule
p, err := strconv.ParseUint(port, 10, 16)
if err != nil {
log.WithContext(ctx).Errorf("failed to parse port %s for rule: %s", port, rule.ID)
continue
}
pr.Port = uint16(p)
rules = append(rules, &pr)
}
return rules
}
// generateRuleIDBase generates the base rule ID for checking duplicates.
func generateRuleIDBase(rule *nmdata.PolicyRule, baseRule RouteFirewallRule) string {
return rule.ID + strings.Join(baseRule.SourceRanges, ",") + strconv.Itoa(FirewallRuleDirectionIN) + baseRule.Protocol + baseRule.Action
}
@@ -0,0 +1,196 @@
package types
import (
"context"
"net/netip"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/shared/management/domain"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
func TestSplitPeerSourcesByFamily(t *testing.T) {
peers := []*nmdata.Peer{
{
IP: netip.MustParseAddr("100.64.0.1"),
IPv6: netip.MustParseAddr("fd00::1"),
},
{
IP: netip.MustParseAddr("100.64.0.2"),
},
{
IP: netip.MustParseAddr("100.64.0.3"),
IPv6: netip.MustParseAddr("fd00::3"),
},
nil,
}
v4, v6 := splitPeerSourcesByFamily(peers)
assert.Equal(t, []string{"100.64.0.1/32", "100.64.0.2/32", "100.64.0.3/32"}, v4)
assert.Equal(t, []string{"fd00::1/128", "fd00::3/128"}, v6)
}
func TestGenerateRouteFirewallRules_V4Route(t *testing.T) {
peers := []*nmdata.Peer{
{
IP: netip.MustParseAddr("100.64.0.1"),
IPv6: netip.MustParseAddr("fd00::1"),
},
{
IP: netip.MustParseAddr("100.64.0.2"),
},
}
r := &nmdata.Route{
ID: "route1",
Network: netip.MustParsePrefix("10.0.0.0/24"),
}
rule := &nmdata.PolicyRule{
PolicyID: "policy1",
ID: "rule1",
Action: string(PolicyTrafficActionAccept),
Protocol: string(PolicyRuleProtocolALL),
}
rules := GenerateRouteFirewallRules(context.Background(), r, rule, peers, FirewallRuleDirectionIN, true)
require.Len(t, rules, 1)
assert.Equal(t, []string{"100.64.0.1/32", "100.64.0.2/32"}, rules[0].SourceRanges, "v4 route should only have v4 sources")
assert.Equal(t, "10.0.0.0/24", rules[0].Destination)
}
func TestGenerateRouteFirewallRules_V6Route(t *testing.T) {
peers := []*nmdata.Peer{
{
IP: netip.MustParseAddr("100.64.0.1"),
IPv6: netip.MustParseAddr("fd00::1"),
},
{
IP: netip.MustParseAddr("100.64.0.2"),
},
}
r := &nmdata.Route{
ID: "route1",
Network: netip.MustParsePrefix("2001:db8::/32"),
}
rule := &nmdata.PolicyRule{
PolicyID: "policy1",
ID: "rule1",
Action: string(PolicyTrafficActionAccept),
Protocol: string(PolicyRuleProtocolALL),
}
rules := GenerateRouteFirewallRules(context.Background(), r, rule, peers, FirewallRuleDirectionIN, true)
require.Len(t, rules, 1)
assert.Equal(t, []string{"fd00::1/128"}, rules[0].SourceRanges, "v6 route should only have v6 sources")
}
func TestGenerateRouteFirewallRules_DynamicRoute_DualStack(t *testing.T) {
peers := []*nmdata.Peer{
{
IP: netip.MustParseAddr("100.64.0.1"),
IPv6: netip.MustParseAddr("fd00::1"),
},
{
IP: netip.MustParseAddr("100.64.0.2"),
},
}
r := &nmdata.Route{
ID: "route1",
NetworkType: nmdata.NetworkTypeDomain,
Domains: domain.List{"example.com"},
}
rule := &nmdata.PolicyRule{
PolicyID: "policy1",
ID: "rule1",
Action: string(PolicyTrafficActionAccept),
Protocol: string(PolicyRuleProtocolALL),
}
rules := GenerateRouteFirewallRules(context.Background(), r, rule, peers, FirewallRuleDirectionIN, true)
require.Len(t, rules, 2, "dynamic route should produce both v4 and v6 rules")
assert.Equal(t, []string{"100.64.0.1/32", "100.64.0.2/32"}, rules[0].SourceRanges)
assert.Equal(t, []string{"fd00::1/128"}, rules[1].SourceRanges)
assert.Equal(t, rules[0].Domains, rules[1].Domains)
assert.True(t, rules[0].IsDynamic)
assert.True(t, rules[1].IsDynamic)
}
func TestGenerateRouteFirewallRules_DynamicRoute_NoV6Peers(t *testing.T) {
peers := []*nmdata.Peer{
{IP: netip.MustParseAddr("100.64.0.1")},
{IP: netip.MustParseAddr("100.64.0.2")},
}
r := &nmdata.Route{
ID: "route1",
NetworkType: nmdata.NetworkTypeDomain,
Domains: domain.List{"example.com"},
}
rule := &nmdata.PolicyRule{
PolicyID: "policy1",
ID: "rule1",
Action: string(PolicyTrafficActionAccept),
Protocol: string(PolicyRuleProtocolALL),
}
rules := GenerateRouteFirewallRules(context.Background(), r, rule, peers, FirewallRuleDirectionIN, true)
require.Len(t, rules, 1, "no v6 peers means only v4 rule")
assert.Equal(t, []string{"100.64.0.1/32", "100.64.0.2/32"}, rules[0].SourceRanges)
}
func TestGenerateRouteFirewallRules_IncludeIPv6False(t *testing.T) {
peers := []*nmdata.Peer{
{
IP: netip.MustParseAddr("100.64.0.1"),
IPv6: netip.MustParseAddr("fd00::1"),
},
{
IP: netip.MustParseAddr("100.64.0.2"),
IPv6: netip.MustParseAddr("fd00::2"),
},
}
t.Run("v6 route excluded", func(t *testing.T) {
r := &nmdata.Route{
ID: "route1",
Network: netip.MustParsePrefix("2001:db8::/32"),
}
rule := &nmdata.PolicyRule{
PolicyID: "policy1",
ID: "rule1",
Action: string(PolicyTrafficActionAccept),
Protocol: string(PolicyRuleProtocolALL),
}
rules := GenerateRouteFirewallRules(context.Background(), r, rule, peers, FirewallRuleDirectionIN, false)
assert.Empty(t, rules, "v6 route should produce no rules when includeIPv6 is false")
})
t.Run("dynamic route only v4", func(t *testing.T) {
r := &nmdata.Route{
ID: "route1",
NetworkType: nmdata.NetworkTypeDomain,
Domains: domain.List{"example.com"},
}
rule := &nmdata.PolicyRule{
PolicyID: "policy1",
ID: "rule1",
Action: string(PolicyTrafficActionAccept),
Protocol: string(PolicyRuleProtocolALL),
}
rules := GenerateRouteFirewallRules(context.Background(), r, rule, peers, FirewallRuleDirectionIN, false)
require.Len(t, rules, 1, "dynamic route with includeIPv6=false should produce only v4 rule")
assert.Equal(t, []string{"100.64.0.1/32", "100.64.0.2/32"}, rules[0].SourceRanges)
})
}
+133
View File
@@ -0,0 +1,133 @@
package types
import (
"net"
"golang.org/x/exp/maps"
nbdns "github.com/netbirdio/netbird/dns"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
"github.com/netbirdio/netbird/shared/management/proto"
)
const (
// AllowedIPsFormat generates Wireguard AllowedIPs format (e.g. 100.64.30.1/32)
AllowedIPsFormat = "%s/32"
// AllowedIPsV6Format generates AllowedIPs format for v6 (e.g. fd12:3456:7890::1/128)
AllowedIPsV6Format = "%s/128"
)
type NetworkMap struct {
Peers []*nmdata.Peer
Network *nmdata.Network
Routes []*nmdata.Route
DNSConfig nbdns.Config
OfflinePeers []*nmdata.Peer
FirewallRules []*FirewallRule
RoutesFirewallRules []*RouteFirewallRule
ForwardingRules []*ForwardingRule
AuthorizedUsers map[string]map[string]struct{}
EnableSSH bool
// ForceRoutingPeerDNSResolution forces the peer to run/use routing-peer DNS
// resolution regardless of the account-global setting, for reverse-proxy
// domain targets.
ForceRoutingPeerDNSResolution bool
}
func (nm *NetworkMap) Merge(other *NetworkMap) {
nm.Peers = mergeUniquePeersByID(nm.Peers, other.Peers)
nm.Routes = mergeUnique(nm.Routes, other.Routes)
nm.OfflinePeers = mergeUniquePeersByID(nm.OfflinePeers, other.OfflinePeers)
nm.FirewallRules = mergeUnique(nm.FirewallRules, other.FirewallRules)
nm.RoutesFirewallRules = mergeUnique(nm.RoutesFirewallRules, other.RoutesFirewallRules)
nm.ForwardingRules = mergeUnique(nm.ForwardingRules, other.ForwardingRules)
nm.ForceRoutingPeerDNSResolution = nm.ForceRoutingPeerDNSResolution || other.ForceRoutingPeerDNSResolution
}
func mergeUniquePeersByID(peers1, peers2 []*nmdata.Peer) []*nmdata.Peer {
result := make(map[string]*nmdata.Peer)
for _, peer := range peers1 {
result[peer.ID] = peer
}
for _, peer := range peers2 {
if _, ok := result[peer.ID]; !ok {
result[peer.ID] = peer
}
}
return maps.Values(result)
}
type ForwardingRule struct {
RuleProtocol string
DestinationPorts RulePortRange
TranslatedAddress net.IP
TranslatedPorts RulePortRange
}
func (f *ForwardingRule) ToProto() *proto.ForwardingRule {
var protocol proto.RuleProtocol
switch f.RuleProtocol {
case "icmp":
protocol = proto.RuleProtocol_ICMP
case "tcp":
protocol = proto.RuleProtocol_TCP
case "udp":
protocol = proto.RuleProtocol_UDP
case "all":
protocol = proto.RuleProtocol_ALL
default:
protocol = proto.RuleProtocol_UNKNOWN
}
return &proto.ForwardingRule{
Protocol: protocol,
DestinationPort: f.DestinationPorts.ToProto(),
TranslatedAddress: ipToBytes(f.TranslatedAddress),
TranslatedPort: f.TranslatedPorts.ToProto(),
}
}
func (f *ForwardingRule) Equal(other *ForwardingRule) bool {
return f.RuleProtocol == other.RuleProtocol &&
f.DestinationPorts.Equal(&other.DestinationPorts) &&
f.TranslatedAddress.Equal(other.TranslatedAddress) &&
f.TranslatedPorts.Equal(&other.TranslatedPorts)
}
func ipToBytes(ip net.IP) []byte {
if ip4 := ip.To4(); ip4 != nil {
return ip4
}
return ip.To16()
}
type comparableObject[T any] interface {
Equal(other T) bool
}
func mergeUnique[T comparableObject[T]](arr1, arr2 []T) []T {
var result []T
for _, item := range arr1 {
if !containsEqual(result, item) {
result = append(result, item)
}
}
for _, item := range arr2 {
if !containsEqual(result, item) {
result = append(result, item)
}
}
return result
}
func containsEqual[T comparableObject[T]](slice []T, element T) bool {
for _, item := range slice {
if item.Equal(element) {
return true
}
}
return false
}
+41
View File
@@ -0,0 +1,41 @@
package types
import (
"testing"
"github.com/stretchr/testify/assert"
)
type mergeTestObject struct {
value int
}
func (t mergeTestObject) Equal(other mergeTestObject) bool {
return t.value == other.value
}
func Test_MergeUniqueArraysWithoutDuplicates(t *testing.T) {
arr1 := []mergeTestObject{{value: 1}, {value: 2}}
arr2 := []mergeTestObject{{value: 2}, {value: 3}}
result := mergeUnique(arr1, arr2)
assert.Len(t, result, 3)
assert.Contains(t, result, mergeTestObject{value: 1})
assert.Contains(t, result, mergeTestObject{value: 2})
assert.Contains(t, result, mergeTestObject{value: 3})
}
func Test_MergeUniqueHandlesEmptyArrays(t *testing.T) {
arr1 := []mergeTestObject{}
arr2 := []mergeTestObject{}
result := mergeUnique(arr1, arr2)
assert.Empty(t, result)
}
func Test_MergeUniqueHandlesOneEmptyArray(t *testing.T) {
arr1 := []mergeTestObject{{value: 1}, {value: 2}}
arr2 := []mergeTestObject{}
result := mergeUnique(arr1, arr2)
assert.Len(t, result, 2)
assert.Contains(t, result, mergeTestObject{value: 1})
assert.Contains(t, result, mergeTestObject{value: 2})
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,225 @@
package types
import (
nmdata "github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
type GroupCompact struct {
Name string
PeerIndexes []int
}
type NetworkMapComponentsCompact struct {
PeerID string
Network *nmdata.Network
AccountSettings *nmdata.AccountSettingsInfo
DNSSettings *nmdata.DNSSettings
CustomZoneDomain string
AllPeers []*nmdata.Peer
PeerIndexes []int
RouterPeerIndexes []int
Groups map[string]*GroupCompact
AllPolicies []*nmdata.Policy
PolicyIndexes []int
ResourcePoliciesMap map[string][]int
Routes []*nmdata.Route
NameServerGroups []*nmdata.NameServerGroup
AllDNSRecords []nmdata.SimpleRecord
AccountZones []nmdata.CustomZone
RoutersMap map[string]map[string]*nmdata.NetworkRouter
NetworkResources []*nmdata.NetworkResource
GroupIDToUserIDs map[string][]string
AllowedUserIDs map[string]struct{}
PostureFailedPeers map[string]map[string]struct{}
}
func (c *NetworkMapComponents) ToCompact() *NetworkMapComponentsCompact {
peerToIndex := make(map[string]int)
var allPeers []*nmdata.Peer
for id, peer := range c.Peers {
if _, exists := peerToIndex[id]; !exists {
peerToIndex[id] = len(allPeers)
allPeers = append(allPeers, peer)
}
}
for id, peer := range c.RouterPeers {
if _, exists := peerToIndex[id]; !exists {
peerToIndex[id] = len(allPeers)
allPeers = append(allPeers, peer)
}
}
peerIndexes := make([]int, 0, len(c.Peers))
for id := range c.Peers {
peerIndexes = append(peerIndexes, peerToIndex[id])
}
routerPeerIndexes := make([]int, 0, len(c.RouterPeers))
for id := range c.RouterPeers {
routerPeerIndexes = append(routerPeerIndexes, peerToIndex[id])
}
groups := make(map[string]*GroupCompact, len(c.Groups))
for id, group := range c.Groups {
peerIdxs := make([]int, 0, len(group.Peers))
for _, peerID := range group.Peers {
if idx, ok := peerToIndex[peerID]; ok {
peerIdxs = append(peerIdxs, idx)
}
}
groups[id] = &GroupCompact{
Name: group.Name,
PeerIndexes: peerIdxs,
}
}
policyToIndex := make(map[*nmdata.Policy]int)
var allPolicies []*nmdata.Policy
for _, policy := range c.Policies {
if _, exists := policyToIndex[policy]; !exists {
policyToIndex[policy] = len(allPolicies)
allPolicies = append(allPolicies, policy)
}
}
for _, policies := range c.ResourcePoliciesMap {
for _, policy := range policies {
if _, exists := policyToIndex[policy]; !exists {
policyToIndex[policy] = len(allPolicies)
allPolicies = append(allPolicies, policy)
}
}
}
policyIndexes := make([]int, len(c.Policies))
for i, policy := range c.Policies {
policyIndexes[i] = policyToIndex[policy]
}
var resourcePoliciesMap map[string][]int
if len(c.ResourcePoliciesMap) > 0 {
resourcePoliciesMap = make(map[string][]int, len(c.ResourcePoliciesMap))
for resID, policies := range c.ResourcePoliciesMap {
indexes := make([]int, len(policies))
for i, policy := range policies {
indexes[i] = policyToIndex[policy]
}
resourcePoliciesMap[resID] = indexes
}
}
return &NetworkMapComponentsCompact{
PeerID: c.PeerID,
Network: c.Network,
AccountSettings: c.AccountSettings,
DNSSettings: c.DNSSettings,
CustomZoneDomain: c.CustomZoneDomain,
AllPeers: allPeers,
PeerIndexes: peerIndexes,
RouterPeerIndexes: routerPeerIndexes,
Groups: groups,
AllPolicies: allPolicies,
PolicyIndexes: policyIndexes,
ResourcePoliciesMap: resourcePoliciesMap,
Routes: c.Routes,
NameServerGroups: c.NameServerGroups,
AllDNSRecords: c.AllDNSRecords,
AccountZones: c.AccountZones,
RoutersMap: c.RoutersMap,
NetworkResources: c.NetworkResources,
GroupIDToUserIDs: c.GroupIDToUserIDs,
AllowedUserIDs: c.AllowedUserIDs,
PostureFailedPeers: c.PostureFailedPeers,
}
}
func (c *NetworkMapComponentsCompact) ToFull() *NetworkMapComponents {
peers := make(map[string]*nmdata.Peer, len(c.PeerIndexes))
for _, idx := range c.PeerIndexes {
if idx >= 0 && idx < len(c.AllPeers) {
peer := c.AllPeers[idx]
peers[peer.ID] = peer
}
}
routerPeers := make(map[string]*nmdata.Peer, len(c.RouterPeerIndexes))
for _, idx := range c.RouterPeerIndexes {
if idx >= 0 && idx < len(c.AllPeers) {
peer := c.AllPeers[idx]
routerPeers[peer.ID] = peer
}
}
groups := make(map[string]*nmdata.Group, len(c.Groups))
for id, gc := range c.Groups {
peerIDs := make([]string, 0, len(gc.PeerIndexes))
for _, idx := range gc.PeerIndexes {
if idx >= 0 && idx < len(c.AllPeers) {
peerIDs = append(peerIDs, c.AllPeers[idx].ID)
}
}
groups[id] = &nmdata.Group{
Name: gc.Name,
Peers: peerIDs,
}
}
policies := make([]*nmdata.Policy, len(c.PolicyIndexes))
for i, idx := range c.PolicyIndexes {
if idx >= 0 && idx < len(c.AllPolicies) {
policies[i] = c.AllPolicies[idx]
}
}
var resourcePoliciesMap map[string][]*nmdata.Policy
if len(c.ResourcePoliciesMap) > 0 {
resourcePoliciesMap = make(map[string][]*nmdata.Policy, len(c.ResourcePoliciesMap))
for resID, indexes := range c.ResourcePoliciesMap {
pols := make([]*nmdata.Policy, 0, len(indexes))
for _, idx := range indexes {
if idx >= 0 && idx < len(c.AllPolicies) {
pols = append(pols, c.AllPolicies[idx])
}
}
resourcePoliciesMap[resID] = pols
}
}
return &NetworkMapComponents{
PeerID: c.PeerID,
Network: c.Network,
AccountSettings: c.AccountSettings,
DNSSettings: c.DNSSettings,
CustomZoneDomain: c.CustomZoneDomain,
Peers: peers,
RouterPeers: routerPeers,
Groups: groups,
Policies: policies,
Routes: c.Routes,
NameServerGroups: c.NameServerGroups,
AllDNSRecords: c.AllDNSRecords,
AccountZones: c.AccountZones,
ResourcePoliciesMap: resourcePoliciesMap,
RoutersMap: c.RoutersMap,
NetworkResources: c.NetworkResources,
GroupIDToUserIDs: c.GroupIDToUserIDs,
AllowedUserIDs: c.AllowedUserIDs,
PostureFailedPeers: c.PostureFailedPeers,
}
}
+70
View File
@@ -0,0 +1,70 @@
package types
import (
nbdns "github.com/netbirdio/netbird/dns"
"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
)
// This file holds the twin→real converters that survive the twin-NetworkMap
// refactor: only the DNS materialization. NetworkMap.DNSConfig stays a real
// nbdns.Config (the client DNS type), so Calculate converts the twin DNS
// components to nbdns at the output boundary. Peers/Routes/Network flow as
// twins all the way through and need no conversion.
func toRealNSGroup(n *nmdata.NameServerGroup) *nbdns.NameServerGroup {
if n == nil {
return nil
}
nameServers := make([]nbdns.NameServer, 0, len(n.NameServers))
for _, ns := range n.NameServers {
nameServers = append(nameServers, nbdns.NameServer{
IP: ns.IP,
NSType: nbdns.NameServerType(ns.NSType),
Port: ns.Port,
})
}
return &nbdns.NameServerGroup{
ID: n.ID,
Name: n.Name,
Description: n.Description,
NameServers: nameServers,
Groups: n.Groups,
Primary: n.Primary,
Domains: n.Domains,
Enabled: n.Enabled,
SearchDomainsEnabled: n.SearchDomainsEnabled,
}
}
func toRealRecords(recs []nmdata.SimpleRecord) []nbdns.SimpleRecord {
if recs == nil {
return nil
}
out := make([]nbdns.SimpleRecord, len(recs))
for i, r := range recs {
out[i] = nbdns.SimpleRecord{
Name: r.Name,
Type: r.Type,
Class: r.Class,
TTL: r.TTL,
RData: r.RData,
}
}
return out
}
func toRealZones(zones []nmdata.CustomZone) []nbdns.CustomZone {
if zones == nil {
return nil
}
out := make([]nbdns.CustomZone, len(zones))
for i, z := range zones {
out[i] = nbdns.CustomZone{
Domain: z.Domain,
Records: toRealRecords(z.Records),
SearchDomainDisabled: z.SearchDomainDisabled,
NonAuthoritative: z.NonAuthoritative,
}
}
return out
}
+139
View File
@@ -0,0 +1,139 @@
package types
import (
"errors"
"fmt"
"strconv"
"strings"
"github.com/netbirdio/netbird/shared/management/proto"
)
// PolicyTrafficActionType action type for the firewall
type PolicyTrafficActionType string
// PolicyRuleProtocolType type of traffic
type PolicyRuleProtocolType string
const (
// PolicyTrafficActionAccept indicates that the traffic is accepted
PolicyTrafficActionAccept = PolicyTrafficActionType("accept")
// PolicyTrafficActionDrop indicates that the traffic is dropped
PolicyTrafficActionDrop = PolicyTrafficActionType("drop")
)
const (
// PolicyRuleProtocolALL type of traffic
PolicyRuleProtocolALL = PolicyRuleProtocolType("all")
// PolicyRuleProtocolTCP type of traffic
PolicyRuleProtocolTCP = PolicyRuleProtocolType("tcp")
// PolicyRuleProtocolUDP type of traffic
PolicyRuleProtocolUDP = PolicyRuleProtocolType("udp")
// PolicyRuleProtocolICMP type of traffic
PolicyRuleProtocolICMP = PolicyRuleProtocolType("icmp")
// PolicyRuleProtocolNetbirdSSH type of traffic
PolicyRuleProtocolNetbirdSSH = PolicyRuleProtocolType("netbird-ssh")
)
// RulePortRange represents a range of ports for a firewall rule.
type RulePortRange struct {
Start uint16
End uint16
}
func (r *RulePortRange) ToProto() *proto.PortInfo {
return &proto.PortInfo{
PortSelection: &proto.PortInfo_Range_{
Range: &proto.PortInfo_Range{
Start: uint32(r.Start),
End: uint32(r.End),
},
},
}
}
func (r *RulePortRange) Equal(other *RulePortRange) bool {
return r.Start == other.Start && r.End == other.End
}
func ParseRuleString(rule string) (PolicyRuleProtocolType, RulePortRange, error) {
rule = strings.TrimSpace(strings.ToLower(rule))
if rule == "all" {
return PolicyRuleProtocolALL, RulePortRange{}, nil
}
if rule == "icmp" {
return PolicyRuleProtocolICMP, RulePortRange{}, nil
}
split := strings.Split(rule, "/")
if len(split) != 2 {
return "", RulePortRange{}, errors.New("invalid rule format: expected protocol/port or protocol/port-range")
}
protoStr := strings.TrimSpace(split[0])
portStr := strings.TrimSpace(split[1])
var protocol PolicyRuleProtocolType
switch protoStr {
case "tcp":
protocol = PolicyRuleProtocolTCP
case "udp":
protocol = PolicyRuleProtocolUDP
case "icmp":
return "", RulePortRange{}, errors.New("icmp does not accept ports; use 'icmp' without '/…'")
case "netbird-ssh":
return PolicyRuleProtocolNetbirdSSH, RulePortRange{Start: nativeSSHPortNumber, End: nativeSSHPortNumber}, nil
default:
return "", RulePortRange{}, fmt.Errorf("invalid protocol: %q", protoStr)
}
portRange, err := parsePortRange(portStr)
if err != nil {
return "", RulePortRange{}, err
}
return protocol, portRange, nil
}
func parsePortRange(portStr string) (RulePortRange, error) {
if strings.Contains(portStr, "-") {
rangeParts := strings.Split(portStr, "-")
if len(rangeParts) != 2 {
return RulePortRange{}, fmt.Errorf("invalid port range %q", portStr)
}
start, err := parsePort(strings.TrimSpace(rangeParts[0]))
if err != nil {
return RulePortRange{}, err
}
end, err := parsePort(strings.TrimSpace(rangeParts[1]))
if err != nil {
return RulePortRange{}, err
}
if start > end {
return RulePortRange{}, fmt.Errorf("invalid port range: start %d > end %d", start, end)
}
return RulePortRange{Start: uint16(start), End: uint16(end)}, nil
}
p, err := parsePort(portStr)
if err != nil {
return RulePortRange{}, err
}
return RulePortRange{Start: uint16(p), End: uint16(p)}, nil
}
func parsePort(portStr string) (int, error) {
if portStr == "" {
return 0, errors.New("empty port")
}
p, err := strconv.Atoi(portStr)
if err != nil {
return 0, fmt.Errorf("invalid port %q: %w", portStr, err)
}
if p < 1 || p > 65535 {
return 0, fmt.Errorf("port out of range (165535): %d", p)
}
return p, nil
}
+19
View File
@@ -0,0 +1,19 @@
package types
type ResourceType string
const (
ResourceTypePeer ResourceType = "peer"
ResourceTypeDomain ResourceType = "domain"
ResourceTypeHost ResourceType = "host"
ResourceTypeSubnet ResourceType = "subnet"
)
func (t ResourceType) Valid() bool {
switch t {
case ResourceTypePeer, ResourceTypeDomain, ResourceTypeHost, ResourceTypeSubnet:
return true
default:
return false
}
}
@@ -0,0 +1,64 @@
package types
import (
"github.com/netbirdio/netbird/route"
"github.com/netbirdio/netbird/shared/management/domain"
)
// RouteFirewallRule a firewall rule applicable for a routed network.
type RouteFirewallRule struct {
// PolicyID is the ID of the policy this rule is derived from
PolicyID string
// RouteID is the ID of the route this rule belongs to.
RouteID route.ID
// SourceRanges IP ranges of the routing peers.
SourceRanges []string
// Action of the traffic when the rule is applicable
Action string
// Destination a network prefix for the routed traffic
Destination string
// Protocol of the traffic
Protocol string
// Port of the traffic
Port uint16
// PortRange represents the range of ports for a firewall rule
PortRange RulePortRange
// Domains list of network domains for the routed traffic
Domains domain.List
// isDynamic indicates whether the rule is for DNS routing
IsDynamic bool
}
func (r *RouteFirewallRule) Equal(other *RouteFirewallRule) bool {
if r.Action != other.Action {
return false
}
if r.Destination != other.Destination {
return false
}
if r.Protocol != other.Protocol {
return false
}
if r.Port != other.Port {
return false
}
if !r.PortRange.Equal(&other.PortRange) {
return false
}
if !r.Domains.Equal(other.Domains) {
return false
}
if r.IsDynamic != other.IsDynamic {
return false
}
return true
}
+78
View File
@@ -0,0 +1,78 @@
// Package netiputil provides compact binary encoding for IP prefixes used in
// the management proto wire format.
//
// Format: [IP bytes][1 byte prefix_len]
// - IPv4: 5 bytes total (4 IP + 1 prefix_len, 0-32)
// - IPv6: 17 bytes total (16 IP + 1 prefix_len, 0-128)
//
// Address family is determined by length: 5 = v4, 17 = v6.
package netiputil
import (
"fmt"
"net/netip"
)
// EncodePrefix encodes a netip.Prefix into compact bytes.
// The address is always unmapped before encoding.
func EncodePrefix(p netip.Prefix) ([]byte, error) {
addr := p.Addr().Unmap()
bits := p.Bits()
if addr.Is4() && bits > 32 {
return nil, fmt.Errorf("invalid prefix length %d for IPv4 address %s (max 32)", bits, addr)
}
return append(addr.AsSlice(), byte(bits)), nil
}
// DecodePrefix decodes compact bytes into a netip.Prefix.
func DecodePrefix(b []byte) (netip.Prefix, error) {
switch len(b) {
case 5:
var ip4 [4]byte
copy(ip4[:], b)
bits := int(b[len(b)-1])
if bits > 32 {
return netip.Prefix{}, fmt.Errorf("invalid IPv4 prefix length %d (max 32)", bits)
}
return netip.PrefixFrom(netip.AddrFrom4(ip4), bits), nil
case 17:
var ip6 [16]byte
copy(ip6[:], b)
addr := netip.AddrFrom16(ip6).Unmap()
bits := int(b[len(b)-1])
if addr.Is4() {
if bits > 32 {
return netip.Prefix{}, fmt.Errorf("invalid prefix length %d for v4-mapped address (max 32)", bits)
}
} else if bits > 128 {
return netip.Prefix{}, fmt.Errorf("invalid IPv6 prefix length %d (max 128)", bits)
}
return netip.PrefixFrom(addr, bits), nil
default:
return netip.Prefix{}, fmt.Errorf("invalid compact prefix length %d (expected 5 or 17)", len(b))
}
}
// EncodeAddr encodes a netip.Addr into compact prefix bytes with a host prefix
// length (/32 for v4, /128 for v6). The address is always unmapped before encoding.
func EncodeAddr(a netip.Addr) []byte {
a = a.Unmap()
bits := 128
if a.Is4() {
bits = 32
}
// Host prefix lengths are always valid for the address family, so error is impossible.
b, _ := EncodePrefix(netip.PrefixFrom(a, bits))
return b
}
// DecodeAddr decodes compact prefix bytes and returns only the address,
// discarding the prefix length. Useful when the prefix length is implied
// (e.g. peer overlay IPs are always /32 or /128).
func DecodeAddr(b []byte) (netip.Addr, error) {
p, err := DecodePrefix(b)
if err != nil {
return netip.Addr{}, err
}
return p.Addr(), nil
}
+175
View File
@@ -0,0 +1,175 @@
package netiputil
import (
"net/netip"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
func TestEncodeDecodePrefix(t *testing.T) {
tests := []struct {
name string
prefix string
size int
}{
{
name: "v4 host",
prefix: "100.64.0.1/32",
size: 5,
},
{
name: "v4 network",
prefix: "10.0.0.0/8",
size: 5,
},
{
name: "v4 default",
prefix: "0.0.0.0/0",
size: 5,
},
{
name: "v6 host",
prefix: "fd00::1/128",
size: 17,
},
{
name: "v6 network",
prefix: "fd00:1234:5678::/48",
size: 17,
},
{
name: "v6 default",
prefix: "::/0",
size: 17,
},
{
name: "v4 /16 overlay",
prefix: "100.64.0.1/16",
size: 5,
},
{
name: "v6 /64 overlay",
prefix: "fd00::abcd:1/64",
size: 17,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
p := netip.MustParsePrefix(tt.prefix)
b, err := EncodePrefix(p)
require.NoError(t, err)
assert.Equal(t, tt.size, len(b), "encoded size")
decoded, err := DecodePrefix(b)
require.NoError(t, err)
assert.Equal(t, p, decoded)
})
}
}
func TestEncodePrefixUnmaps(t *testing.T) {
// v4-mapped v6 address should encode as v4
mapped := netip.MustParsePrefix("::ffff:10.1.2.3/32")
b, err := EncodePrefix(mapped)
require.NoError(t, err)
assert.Equal(t, 5, len(b), "v4-mapped should encode as 5 bytes")
decoded, err := DecodePrefix(b)
require.NoError(t, err)
assert.Equal(t, netip.MustParsePrefix("10.1.2.3/32"), decoded)
}
func TestEncodePrefixUnmapsRejectsInvalidBits(t *testing.T) {
// v4-mapped v6 with bits > 32 should return an error
mapped128 := netip.MustParsePrefix("::ffff:10.1.2.3/128")
_, err := EncodePrefix(mapped128)
require.Error(t, err)
// v4-mapped v6 with bits=96 should also return an error
mapped96 := netip.MustParsePrefix("::ffff:10.0.0.0/96")
_, err = EncodePrefix(mapped96)
require.Error(t, err)
// v4-mapped v6 with bits=32 should succeed
mapped32 := netip.MustParsePrefix("::ffff:10.1.2.3/32")
b, err := EncodePrefix(mapped32)
require.NoError(t, err)
assert.Equal(t, 5, len(b), "v4-mapped should encode as 5 bytes")
decoded, err := DecodePrefix(b)
require.NoError(t, err)
assert.Equal(t, netip.MustParsePrefix("10.1.2.3/32"), decoded)
}
func TestDecodeAddr(t *testing.T) {
v4 := netip.MustParseAddr("100.64.0.5")
b := EncodeAddr(v4)
assert.Equal(t, 5, len(b))
got, err := DecodeAddr(b)
require.NoError(t, err)
assert.Equal(t, v4, got)
v6 := netip.MustParseAddr("fd00::1")
b = EncodeAddr(v6)
assert.Equal(t, 17, len(b))
got, err = DecodeAddr(b)
require.NoError(t, err)
assert.Equal(t, v6, got)
}
func TestDecodePrefixInvalidLength(t *testing.T) {
_, err := DecodePrefix([]byte{1, 2, 3})
assert.Error(t, err)
assert.Contains(t, err.Error(), "invalid compact prefix length 3")
_, err = DecodePrefix(nil)
assert.Error(t, err)
_, err = DecodePrefix([]byte{})
assert.Error(t, err)
}
func TestDecodePrefixInvalidBits(t *testing.T) {
// v4 with bits > 32
b := []byte{10, 0, 0, 1, 33}
_, err := DecodePrefix(b)
assert.Error(t, err)
assert.Contains(t, err.Error(), "invalid IPv4 prefix length 33")
// v6 with bits > 128
b = make([]byte, 17)
b[0] = 0xfd
b[16] = 129
_, err = DecodePrefix(b)
assert.Error(t, err)
assert.Contains(t, err.Error(), "invalid IPv6 prefix length 129")
}
func TestDecodePrefixUnmapsV6Input(t *testing.T) {
addr := netip.MustParseAddr("::ffff:192.168.1.1")
// v4-mapped v6 with bits > 32 should return an error
raw := addr.As16()
bInvalid := make([]byte, 17)
copy(bInvalid, raw[:])
bInvalid[16] = 128
_, err := DecodePrefix(bInvalid)
require.Error(t, err, "v4-mapped address with /128 prefix should be rejected")
assert.Contains(t, err.Error(), "invalid prefix length")
// v4-mapped v6 with valid /32 should decode and unmap correctly
bValid := make([]byte, 17)
copy(bValid, raw[:])
bValid[16] = 32
decoded, err := DecodePrefix(bValid)
require.NoError(t, err)
assert.True(t, decoded.Addr().Is4(), "should be unmapped to v4")
assert.Equal(t, netip.MustParsePrefix("192.168.1.1/32"), decoded)
}
-4
View File
@@ -8,7 +8,3 @@ type Auth struct {
func (a *Auth) Validate(any) error {
return nil
}
func (a *Auth) ValidateHelloMsgType(any) error {
return nil
}
-10
View File
@@ -1,10 +1,8 @@
package hmac
import (
"bytes"
"crypto/hmac"
"encoding/base64"
"encoding/gob"
"fmt"
"hash"
"strconv"
@@ -18,14 +16,6 @@ type Token struct {
Signature string
}
func unmarshalToken(payload []byte) (Token, error) {
var creds Token
buffer := bytes.NewBuffer(payload)
decoder := gob.NewDecoder(buffer)
err := decoder.Decode(&creds)
return creds, err
}
// TimedHMAC generates a token with TTL and uses a pre-shared secret known to the relay server
type TimedHMAC struct {
secret string
-33
View File
@@ -1,33 +0,0 @@
package hmac
import (
"crypto/sha256"
"fmt"
"time"
log "github.com/sirupsen/logrus"
)
type TimedHMACValidator struct {
*TimedHMAC
}
func NewTimedHMACValidator(secret string, duration time.Duration) *TimedHMACValidator {
ta := NewTimedHMAC(secret, duration)
return &TimedHMACValidator{
ta,
}
}
func (a *TimedHMACValidator) Validate(credentials any) error {
b, ok := credentials.([]byte)
if !ok {
return fmt.Errorf("invalid credentials type")
}
c, err := unmarshalToken(b)
if err != nil {
log.Debugf("failed to unmarshal token: %s", err)
return err
}
return a.TimedHMAC.Validate(sha256.New, c)
}
+1 -10
View File
@@ -1,28 +1,19 @@
package auth
import (
"time"
auth "github.com/netbirdio/netbird/shared/relay/auth/hmac"
authv2 "github.com/netbirdio/netbird/shared/relay/auth/hmac/v2"
)
type TimedHMACValidator struct {
authenticatorV2 *authv2.Validator
authenticator *auth.TimedHMACValidator
}
func NewTimedHMACValidator(secret []byte, duration time.Duration) *TimedHMACValidator {
func NewTimedHMACValidator(secret []byte) *TimedHMACValidator {
return &TimedHMACValidator{
authenticatorV2: authv2.NewValidator(secret),
authenticator: auth.NewTimedHMACValidator(string(secret), duration),
}
}
func (a *TimedHMACValidator) Validate(credentials any) error {
return a.authenticatorV2.Validate(credentials)
}
func (a *TimedHMACValidator) ValidateHelloMsgType(credentials any) error {
return a.authenticator.Validate(credentials)
}
+230 -6
View File
@@ -2,15 +2,22 @@ package client
import (
"context"
"errors"
"fmt"
"net"
"net/netip"
"net/url"
"strings"
"sync"
"sync/atomic"
"time"
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/client/netevents/sweep"
auth "github.com/netbirdio/netbird/shared/relay/auth/hmac"
"github.com/netbirdio/netbird/shared/relay/client/dialer"
netErr "github.com/netbirdio/netbird/shared/relay/client/dialer/net"
"github.com/netbirdio/netbird/shared/relay/healthcheck"
"github.com/netbirdio/netbird/shared/relay/messages"
)
@@ -139,6 +146,19 @@ func (cc *connContainer) close() {
}
}
// transportConn is implemented by relay connections that know their transport.
type transportConn interface {
Protocol() string
}
// NetEvents is the OS network event view the relay consumes: availability
// gating for the reconnect guard and dial registration for the network change
// sweep.
type NetEvents interface {
NetworkWatcher
StartDial(ctx context.Context) *sweep.Dial
}
// Client is a client for the relay server. It is responsible for establishing a connection to the relay server and
// managing connections to other peers. All exported functions are safe to call concurrently. After close the connection,
// the client can be reused by calling Connect again. When the client is closed, all connections are closed too.
@@ -146,6 +166,7 @@ func (cc *connContainer) close() {
type Client struct {
log *log.Entry
connectionURL string
serverIP netip.Addr
authTokenStore *auth.TokenStore
hashedID messages.PeerID
@@ -167,16 +188,55 @@ type Client struct {
stateSubscription *PeersStateSubscription
mtu uint16
// transportFallback, when set, records datagram-too-large failures so a
// datagram-sized transport is avoided on subsequent connects. Shared via
// the manager.
transportFallback *transportFallback
// netEvents registers the relay dial for the network change sweep; the
// read loop reports the disconnect and the guard reconnects. Shared via
// the manager.
netEvents NetEvents
// datagramFallbackTriggered guards a single fallback per connection so a
// burst of oversized datagrams triggers one reconnect, not many.
datagramFallbackTriggered atomic.Bool
// transport is the negotiated relay transport of the
// current connection, guarded by mu.
transport string
}
// Transport returns the negotiated relay transport of the current connection,
// or an empty string when not connected.
func (c *Client) Transport() string {
c.mu.Lock()
defer c.mu.Unlock()
return c.transport
}
// SetTransportFallback wires the shared datagram-transport fallback tracker.
func (c *Client) SetTransportFallback(tf *transportFallback) {
c.transportFallback = tf
}
// NewClient creates a new client for the relay server. The client is not connected to the server until the Connect
// is called.
func NewClient(serverURL string, authTokenStore *auth.TokenStore, peerID string, mtu uint16) *Client {
return NewClientWithServerIP(serverURL, netip.Addr{}, authTokenStore, peerID, mtu)
}
// NewClientWithServerIP creates a new client for the relay server with a known server IP. serverIP, when valid, is
// dialed directly first; the FQDN is only attempted if the IP-based dial fails. TLS verification still uses the
// FQDN from serverURL via SNI.
func NewClientWithServerIP(serverURL string, serverIP netip.Addr, authTokenStore *auth.TokenStore, peerID string, mtu uint16) *Client {
hashedID := messages.HashID(peerID)
relayLog := log.WithFields(log.Fields{"relay": serverURL})
c := &Client{
log: relayLog,
connectionURL: serverURL,
serverIP: serverIP,
authTokenStore: authTokenStore,
hashedID: hashedID,
mtu: mtu,
@@ -304,6 +364,23 @@ func (c *Client) ServerInstanceURL() (string, error) {
return c.instanceURL.String(), nil
}
// ConnectedIP returns the IP address of the live relay-server connection,
// extracted from the underlying socket's RemoteAddr. Zero value if not
// connected or if the address is not an IP literal.
func (c *Client) ConnectedIP() netip.Addr {
c.mu.Lock()
conn := c.relayConn
c.mu.Unlock()
if conn == nil {
return netip.Addr{}
}
addr := conn.RemoteAddr()
if addr == nil {
return netip.Addr{}
}
return extractIPLiteral(addr.String())
}
// SetOnDisconnectListener sets a function that will be called when the connection to the relay server is closed.
func (c *Client) SetOnDisconnectListener(fn func(string)) {
c.listenerMutex.Lock()
@@ -330,14 +407,53 @@ func (c *Client) Close() error {
}
func (c *Client) connect(ctx context.Context) (*RelayAddr, error) {
dialers := c.getDialers()
// A sweep cancels this context, so a dial started on the old network
// aborts instead of waiting out its handshake timeout.
var dial *sweep.Dial
if c.netEvents != nil {
dial = c.netEvents.StartDial(ctx)
} else {
dial = (*sweep.Sweeper)(nil).StartDial(ctx)
}
defer dial.Release()
ctx = dial.Ctx()
rd := dialer.NewRaceDial(c.log, dialer.DefaultConnectionTimeout, c.connectionURL, dialers...)
conn, err := rd.Dial(ctx)
mode := transportModeFromEnv()
dialers := c.getDialers(mode)
var conn net.Conn
if c.serverIP.IsValid() {
var err error
conn, err = c.dialRaceDirect(ctx, mode, dialers)
if err != nil {
c.log.Infof("dial via server IP %s failed, falling back to FQDN: %v", c.serverIP, err)
conn = nil
}
}
if conn == nil {
rd := dialer.NewRaceDial(c.log, dialer.DefaultConnectionTimeout, c.connectionURL, dialers...)
if mode.sequential() {
rd.WithSequential()
}
var err error
conn, err = rd.Dial(ctx)
if err != nil {
return nil, fmt.Errorf("dial via FQDN: %w", err)
}
}
// Read the transport off the concrete connection: the sweeper's wrapper
// embeds net.Conn only, so it does not promote Protocol().
if tc, ok := conn.(transportConn); ok {
c.transport = tc.Protocol()
}
conn, err := dial.WrapConn(conn)
if err != nil {
return nil, err
return nil, fmt.Errorf("register connection: %w", err)
}
c.relayConn = conn
c.datagramFallbackTriggered.Store(false)
instanceURL, err := c.handShake(ctx)
if err != nil {
@@ -351,6 +467,55 @@ func (c *Client) connect(ctx context.Context) (*RelayAddr, error) {
return instanceURL, nil
}
// dialRaceDirect dials c.serverIP, preserving the original FQDN as the TLS ServerName for SNI.
func (c *Client) dialRaceDirect(ctx context.Context, mode TransportMode, dialers []dialer.DialeFn) (net.Conn, error) {
directURL, serverName, err := substituteHost(c.connectionURL, c.serverIP)
if err != nil {
return nil, fmt.Errorf("substitute host: %w", err)
}
c.log.Debugf("dialing via server IP %s (SNI=%s)", c.serverIP, serverName)
rd := dialer.NewRaceDial(c.log, dialer.DefaultConnectionTimeout, directURL, dialers...).
WithServerName(serverName)
if mode.sequential() {
rd.WithSequential()
}
return rd.Dial(ctx)
}
// substituteHost replaces the host portion of a rel/rels URL with ip,
// preserving the scheme and port. Returns the rewritten URL and the
// original host to use as the TLS ServerName, or empty if the original
// host is itself an IP literal (SNI requires a DNS name).
func substituteHost(serverURL string, ip netip.Addr) (string, string, error) {
u, err := url.Parse(serverURL)
if err != nil {
return "", "", fmt.Errorf("parse %q: %w", serverURL, err)
}
if u.Scheme == "" || u.Host == "" {
return "", "", fmt.Errorf("invalid relay URL %q", serverURL)
}
if !ip.IsValid() {
return "", "", errors.New("invalid server IP")
}
origHost := u.Hostname()
if _, err := netip.ParseAddr(origHost); err == nil {
origHost = ""
}
ip = ip.Unmap()
newHost := ip.String()
if ip.Is6() {
newHost = "[" + newHost + "]"
}
if port := u.Port(); port != "" {
u.Host = newHost + ":" + port
} else {
u.Host = newHost
}
return u.String(), origHost, nil
}
func (c *Client) handShake(ctx context.Context) (*RelayAddr, error) {
msg, err := messages.MarshalAuthMsg(c.hashedID, c.authTokenStore.TokenBinary())
if err != nil {
@@ -541,13 +706,53 @@ func (c *Client) writeTo(containerRef *connContainer, dstID messages.PeerID, pay
}
// the write always return with 0 length because the underling does not support the size feedback.
_, err = c.relayConn.Write(msg)
conn := c.relayConn
_, err = conn.Write(msg)
if err != nil {
c.log.Errorf("failed to write transport message: %s", err)
if errors.Is(err, netErr.ErrDatagramTooLarge) {
c.onDatagramTooLarge(conn, err)
} else {
c.log.Errorf("failed to write transport message: %s", err)
}
}
return len(payload), err
}
// onDatagramTooLarge reacts to a datagram rejected as too large for the path.
// When a non-datagram transport is available, it records a fallback for this
// server and closes the connection so the reconnect avoids datagram-sized
// transports. A single fallback is triggered per connection regardless of how
// many oversized datagrams arrive. cause carries the datagram size and budget.
func (c *Client) onDatagramTooLarge(conn net.Conn, cause error) {
// Handle one oversized datagram per connection; a burst triggers a single
// fallback (and a single log line), not many.
if !c.datagramFallbackTriggered.CompareAndSwap(false, true) {
return
}
// If the selected mode offers no non-datagram transport (e.g. pinned to a
// datagram-sized transport), reconnecting would just re-fail, so leave the
// connection up rather than loop.
if len(nonDatagramSized(c.baseDialers(transportModeFromEnv()))) == 0 {
c.log.Warnf("%s, but no non-datagram transport is available, not falling back", cause)
return
}
// Without the shared tracker a reconnect would just select the same
// transport again and re-fail, so leave the connection up rather than loop.
if c.transportFallback == nil {
c.log.Debugf("%s, but no transport fallback configured, leaving connection up", cause)
return
}
window := c.transportFallback.recordFailure(c.connectionURL)
c.log.Warnf("%s, avoiding datagram-sized transport for %s", cause, window)
if err := conn.Close(); err != nil {
c.log.Debugf("close relay connection for transport fallback: %s", err)
}
}
func (c *Client) listenForStopEvents(ctx context.Context, hc *healthcheck.Receiver, conn net.Conn, internalStopFlag *internalStopFlag) {
for {
select {
@@ -639,6 +844,7 @@ func (c *Client) close(gracefullyExit bool) error {
return nil
}
c.serviceIsRunning = false
c.transport = ""
c.muInstanceURL.Lock()
c.instanceURL = nil
@@ -716,3 +922,21 @@ func (c *Client) handlePeersWentOfflineMsg(buf []byte) {
}
c.stateSubscription.OnPeersWentOffline(peersID)
}
// extractIPLiteral returns the IP from address forms produced by the relay
// dialers (URL or host:port). Zero value if the host is not an IP.
func extractIPLiteral(s string) netip.Addr {
if u, err := url.Parse(s); err == nil && u.Host != "" {
s = u.Host
}
host, _, err := net.SplitHostPort(s)
if err != nil {
host = s
}
host = strings.Trim(host, "[]")
ip, err := netip.ParseAddr(host)
if err != nil {
return netip.Addr{}
}
return ip.Unmap()
}
+280
View File
@@ -0,0 +1,280 @@
package client
import (
"context"
"fmt"
"net"
"net/netip"
"testing"
"time"
"go.opentelemetry.io/otel"
"github.com/netbirdio/netbird/client/iface"
"github.com/netbirdio/netbird/relay/server"
"github.com/netbirdio/netbird/shared/relay/auth/allow"
)
// TestClient_ServerIPRecoversFromUnresolvableFQDN verifies that when the
// primary FQDN-based dial fails (unresolvable .invalid host), Connect
// recovers via the server IP and SNI still uses the FQDN.
func TestClient_ServerIPRecoversFromUnresolvableFQDN(t *testing.T) {
ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
defer cancel()
listenAddr, port := freeAddr(t)
srvCfg := server.Config{
Meter: otel.Meter(""),
ExposedAddress: fmt.Sprintf("rel://test-unresolvable-host.invalid:%d", port),
TLSSupport: false,
AuthValidator: &allow.Auth{},
}
srv, err := server.NewServer(srvCfg)
if err != nil {
t.Fatalf("create server: %s", err)
}
errChan := make(chan error, 1)
go func() {
if err := srv.Listen(server.ListenerConfig{Address: listenAddr}); err != nil {
errChan <- err
}
}()
t.Cleanup(func() {
if err := srv.Shutdown(context.Background()); err != nil {
t.Errorf("shutdown server: %s", err)
}
})
if err := waitForServerToStart(errChan); err != nil {
t.Fatalf("server failed to start: %s", err)
}
t.Run("no server IP, primary fails", func(t *testing.T) {
c := NewClient(srvCfg.ExposedAddress, hmacTokenStore, "alice-noip", iface.DefaultMTU)
err := c.Connect(ctx)
if err == nil {
_ = c.Close()
t.Fatalf("expected connect to fail without server IP, got nil")
}
})
t.Run("server IP recovers", func(t *testing.T) {
c := NewClientWithServerIP(srvCfg.ExposedAddress, netip.MustParseAddr("127.0.0.1"), hmacTokenStore, "alice-with-ip", iface.DefaultMTU)
if err := c.Connect(ctx); err != nil {
t.Fatalf("connect with server IP: %s", err)
}
t.Cleanup(func() { _ = c.Close() })
if !c.Ready() {
t.Fatalf("client not ready after connect")
}
if got := c.ConnectedIP(); got.String() != "127.0.0.1" {
t.Fatalf("ConnectedIP = %q, want 127.0.0.1", got)
}
})
}
// TestClient_ConnectedIPAfterFQDNDial verifies ConnectedIP returns the
// resolved IP after a successful FQDN-based dial. The underlying socket's
// RemoteAddr must be exposed through the dialer wrappers; if it returns
// the dial-time URL instead, ConnectedIP returns empty and the dial
// IP we advertise to peers is empty too.
func TestClient_ConnectedIPAfterFQDNDial(t *testing.T) {
ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
defer cancel()
listenAddr, port := freeAddr(t)
srvCfg := server.Config{
Meter: otel.Meter(""),
ExposedAddress: fmt.Sprintf("rel://localhost:%d", port),
TLSSupport: false,
AuthValidator: &allow.Auth{},
}
srv, err := server.NewServer(srvCfg)
if err != nil {
t.Fatalf("create server: %s", err)
}
errChan := make(chan error, 1)
go func() {
if err := srv.Listen(server.ListenerConfig{Address: listenAddr}); err != nil {
errChan <- err
}
}()
t.Cleanup(func() { _ = srv.Shutdown(context.Background()) })
if err := waitForServerToStart(errChan); err != nil {
t.Fatalf("server failed to start: %s", err)
}
c := NewClient(srvCfg.ExposedAddress, hmacTokenStore, "alice-fqdn", iface.DefaultMTU)
if err := c.Connect(ctx); err != nil {
t.Fatalf("connect: %s", err)
}
t.Cleanup(func() { _ = c.Close() })
got := c.ConnectedIP().String()
if got != "127.0.0.1" && got != "::1" {
t.Fatalf("ConnectedIP after FQDN dial = %q, want 127.0.0.1 or ::1", got)
}
}
func TestSubstituteHost(t *testing.T) {
tests := []struct {
name string
serverURL string
ip string
wantURL string
wantServerName string
wantErr bool
}{
{
name: "rels with port",
serverURL: "rels://relay.netbird.io:443",
ip: "10.0.0.5",
wantURL: "rels://10.0.0.5:443",
wantServerName: "relay.netbird.io",
},
{
name: "rel with port",
serverURL: "rel://relay.example.com:80",
ip: "192.0.2.1",
wantURL: "rel://192.0.2.1:80",
wantServerName: "relay.example.com",
},
{
name: "ipv6 server IP bracketed",
serverURL: "rels://relay.example.com:443",
ip: "2001:db8::1",
wantURL: "rels://[2001:db8::1]:443",
wantServerName: "relay.example.com",
},
{
name: "no port",
serverURL: "rels://relay.example.com",
ip: "10.0.0.5",
wantURL: "rels://10.0.0.5",
wantServerName: "relay.example.com",
},
{
name: "ipv6 server with port returns empty SNI",
serverURL: "rels://[2001:db8::5]:443",
ip: "10.0.0.5",
wantURL: "rels://10.0.0.5:443",
wantServerName: "",
},
{
name: "ipv4 server with port returns empty SNI",
serverURL: "rels://10.0.0.5:443",
ip: "10.0.0.6",
wantURL: "rels://10.0.0.6:443",
wantServerName: "",
},
{
name: "ipv6 server IP no port",
serverURL: "rels://relay.example.com",
ip: "2001:db8::1",
wantURL: "rels://[2001:db8::1]",
wantServerName: "relay.example.com",
},
{
name: "missing scheme",
serverURL: "relay.example.com:443",
ip: "10.0.0.5",
wantErr: true,
},
{
name: "empty",
serverURL: "",
ip: "10.0.0.5",
wantErr: true,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
var ip netip.Addr
if tt.ip != "" {
ip = netip.MustParseAddr(tt.ip)
}
gotURL, gotName, err := substituteHost(tt.serverURL, ip)
if tt.wantErr {
if err == nil {
t.Fatalf("expected error, got nil")
}
return
}
if err != nil {
t.Fatalf("unexpected error: %s", err)
}
if gotURL != tt.wantURL {
t.Errorf("URL = %q, want %q", gotURL, tt.wantURL)
}
if gotName != tt.wantServerName {
t.Errorf("ServerName = %q, want %q", gotName, tt.wantServerName)
}
})
}
}
func TestClient_ConnectedIPEmptyWhenNotConnected(t *testing.T) {
c := NewClient("rel://example.invalid:80", hmacTokenStore, "x", iface.DefaultMTU)
if got := c.ConnectedIP(); got.IsValid() {
t.Fatalf("ConnectedIP on disconnected client = %q, want zero", got)
}
}
// staticAddr is a net.Addr that returns a fixed string. Used to verify
// ConnectedIP parses RemoteAddr correctly.
type staticAddr struct{ s string }
func (a staticAddr) Network() string { return "tcp" }
func (a staticAddr) String() string { return a.s }
type stubConn struct {
net.Conn
remote net.Addr
}
func (s stubConn) RemoteAddr() net.Addr { return s.remote }
func TestClient_ConnectedIPParsesRemoteAddr(t *testing.T) {
tests := []struct {
name string
s string
want string
}{
{"hostport ipv4", "127.0.0.1:50301", "127.0.0.1"},
{"hostport ipv6 bracketed", "[::1]:50301", "::1"},
{"url with ipv4", "rel://127.0.0.1:50301", "127.0.0.1"},
{"url with ipv6", "rels://[2001:db8::1]:443", "2001:db8::1"},
{"fqdn url returns empty", "rel://relay.example.com:50301", ""},
{"fqdn hostport returns empty", "relay.example.com:50301", ""},
{"plain ipv4 no port", "10.0.0.1", "10.0.0.1"},
{"empty", "", ""},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
c := &Client{relayConn: stubConn{remote: staticAddr{s: tt.s}}}
got := c.ConnectedIP()
var gotStr string
if got.IsValid() {
gotStr = got.String()
}
if gotStr != tt.want {
t.Errorf("ConnectedIP(%q) = %q, want %q", tt.s, gotStr, tt.want)
}
})
}
}
// freeAddr returns a 127.0.0.1 address with an OS-assigned port. The
// listener is closed before returning, so the port is briefly free for
// the caller to bind. Avoids hardcoded ports that can collide.
func freeAddr(t *testing.T) (string, int) {
t.Helper()
l, err := net.Listen("tcp", "127.0.0.1:0")
if err != nil {
t.Fatalf("get free port: %s", err)
}
addr := l.Addr().(*net.TCPAddr)
_ = l.Close()
return addr.String(), addr.Port
}
+18
View File
@@ -0,0 +1,18 @@
package dialer
// DatagramSized is implemented by dialers whose connections carry each write in
// a single datagram, so a write can be rejected when it exceeds the path's
// datagram budget (e.g. QUIC). Transports without this capability (e.g.
// WebSocket over TCP) impose no per-write size limit, so the relay client can
// fall back to them when a datagram-sized transport rejects a write as too
// large. The capability is advertised per dialer rather than hardcoded, so a
// new transport only needs to declare whether it is datagram-sized.
type DatagramSized interface {
DatagramSized()
}
// IsDatagramSized reports whether d produces datagram-sized connections.
func IsDatagramSized(d DialeFn) bool {
_, ok := d.(DatagramSized)
return ok
}
+5
View File
@@ -4,4 +4,9 @@ import "errors"
var (
ErrClosedByServer = errors.New("closed by server")
// ErrDatagramTooLarge is returned when a transport message exceeds the
// QUIC datagram size the path to the relay can carry. The relay client
// treats it as a signal to fall back to a non-datagram transport.
ErrDatagramTooLarge = errors.New("datagram frame too large")
)
+19 -6
View File
@@ -8,7 +8,6 @@ import (
"time"
"github.com/quic-go/quic-go"
log "github.com/sirupsen/logrus"
netErr "github.com/netbirdio/netbird/shared/relay/client/dialer/net"
)
@@ -52,15 +51,17 @@ func (c *Conn) Read(b []byte) (n int, err error) {
}
func (c *Conn) Write(b []byte) (int, error) {
err := c.session.SendDatagram(b)
if err != nil {
err = c.remoteCloseErrHandling(err)
log.Errorf("failed to write to QUIC stream: %v", err)
return 0, err
if err := c.session.SendDatagram(b); err != nil {
return 0, c.writeErrHandling(err, len(b))
}
return len(b), nil
}
// Protocol returns the transport name for this connection.
func (c *Conn) Protocol() string {
return Network
}
func (c *Conn) RemoteAddr() net.Addr {
return c.session.RemoteAddr()
}
@@ -95,3 +96,15 @@ func (c *Conn) remoteCloseErrHandling(err error) error {
}
return err
}
// writeErrHandling normalizes SendDatagram errors. A datagram that exceeds the
// path's QUIC packet budget is mapped to ErrDatagramTooLarge (annotated with the
// datagram size and path budget) so the relay client can fall back to a
// non-datagram transport.
func (c *Conn) writeErrHandling(err error, size int) error {
var tooLarge *quic.DatagramTooLargeError
if errors.As(err, &tooLarge) {
return fmt.Errorf("%w: %d byte datagram over path budget %d", netErr.ErrDatagramTooLarge, size, tooLarge.MaxDatagramPayloadSize)
}
return c.remoteCloseErrHandling(err)
}
+20 -12
View File
@@ -23,7 +23,13 @@ func (d Dialer) Protocol() string {
return Network
}
func (d Dialer) Dial(ctx context.Context, address string) (net.Conn, error) {
// DatagramSized marks QUIC as a datagram-sized transport: relay traffic is
// carried in QUIC DATAGRAM frames, which must fit a single packet.
func (d Dialer) DatagramSized() {
// Intentional marker method; presence is the capability signal.
}
func (d Dialer) Dial(ctx context.Context, address, serverName string) (net.Conn, error) {
quicURL, err := prepareURL(address)
if err != nil {
return nil, err
@@ -32,11 +38,14 @@ func (d Dialer) Dial(ctx context.Context, address string) (net.Conn, error) {
// Get the base TLS config
tlsClientConfig := quictls.ClientQUICTLSConfig()
// Set ServerName to hostname if not an IP address
host, _, splitErr := net.SplitHostPort(quicURL)
if splitErr == nil && net.ParseIP(host) == nil {
// It's a hostname, not an IP - modify directly
tlsClientConfig.ServerName = host
switch {
case serverName != "" && net.ParseIP(serverName) == nil:
tlsClientConfig.ServerName = serverName
default:
host, _, splitErr := net.SplitHostPort(quicURL)
if splitErr == nil && net.ParseIP(host) == nil {
tlsClientConfig.ServerName = host
}
}
quicConfig := &quic.Config{
@@ -44,18 +53,17 @@ func (d Dialer) Dial(ctx context.Context, address string) (net.Conn, error) {
MaxIdleTimeout: 4 * time.Minute,
EnableDatagrams: true,
InitialPacketSize: nbRelay.QUICInitialPacketSize,
Tracer: connectionTracer(quicURL),
}
udpConn, err := nbnet.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
udpConn, err := nbnet.ListenUDP("udp", &net.UDPAddr{Port: 0})
if err != nil {
log.Errorf("failed to listen on UDP: %s", err)
return nil, err
return nil, fmt.Errorf("listen udp: %w", err)
}
udpAddr, err := net.ResolveUDPAddr("udp", quicURL)
if err != nil {
log.Errorf("failed to resolve UDP address: %s", err)
return nil, err
return nil, fmt.Errorf("resolve %s: %w", quicURL, err)
}
session, err := quic.Dial(ctx, udpConn, udpAddr, tlsClientConfig, quicConfig)
@@ -63,7 +71,7 @@ func (d Dialer) Dial(ctx context.Context, address string) (net.Conn, error) {
if errors.Is(err, context.Canceled) {
return nil, err
}
log.Errorf("failed to dial to Relay server via QUIC '%s': %s", quicURL, err)
log.Debugf("failed to dial to Relay server via QUIC '%s': %s", quicURL, err)
return nil, err
}
@@ -0,0 +1,145 @@
package quic
import (
"testing"
"github.com/quic-go/quic-go/qlog"
"github.com/quic-go/quic-go/qlogwriter"
log "github.com/sirupsen/logrus"
"github.com/sirupsen/logrus/hooks/test"
)
func TestCloseReason(t *testing.T) {
transportErr := qlog.TransportErrorCode(0x2) // CONNECTION_REFUSED
appErr := qlog.ApplicationErrorCode(42)
tests := []struct {
name string
event qlog.ConnectionClosed
want string
}{
{
// A close carrying nothing but an initiator still reads sensibly.
name: "initiator only",
event: qlog.ConnectionClosed{Initiator: qlog.InitiatorLocal},
want: "closed by local",
},
{
name: "transport error with trigger",
event: qlog.ConnectionClosed{
Initiator: qlog.InitiatorRemote,
ConnectionError: &transportErr,
Trigger: qlog.ConnectionCloseTriggerIdleTimeout,
},
want: "closed by remote, transport error: CONNECTION_REFUSED, trigger: idle_timeout",
},
{
name: "application error with reason",
event: qlog.ConnectionClosed{
Initiator: qlog.InitiatorLocal,
ApplicationError: &appErr,
Reason: "bye",
},
want: "closed by local, application error: 42, reason: bye",
},
{
// Transport and application errors are mutually exclusive in
// practice; if both are set the transport code wins.
name: "transport error takes precedence over application error",
event: qlog.ConnectionClosed{
Initiator: qlog.InitiatorLocal,
ConnectionError: &transportErr,
ApplicationError: &appErr,
},
want: "closed by local, transport error: CONNECTION_REFUSED",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
if got := closeReason(tt.event); got != tt.want {
t.Errorf("closeReason() = %q, want %q", got, tt.want)
}
})
}
}
func TestLogSinkRecordEvent(t *testing.T) {
tests := []struct {
name string
event qlogwriter.Event
wantLevel log.Level
wantMsg string
}{
{
name: "settled MTU is logged at info",
event: qlog.MTUUpdated{Value: 1400, Done: true},
wantLevel: log.InfoLevel,
wantMsg: "QUIC path MTU settled at 1400",
},
{
// Probing fires repeatedly during discovery, so it stays at debug.
name: "MTU probe is logged at debug",
event: qlog.MTUUpdated{Value: 1300, Done: false},
wantLevel: log.DebugLevel,
wantMsg: "QUIC path MTU probing at 1300",
},
{
name: "connection closed is logged at debug",
event: qlog.ConnectionClosed{Initiator: qlog.InitiatorRemote},
wantLevel: log.DebugLevel,
wantMsg: "QUIC connection closed: closed by remote",
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
logger, hook := test.NewNullLogger()
logger.SetLevel(log.DebugLevel)
recorder := logSink{log: logger.WithField("relay", "relay.example.com:443")}
recorder.RecordEvent(tt.event)
entries := hook.AllEntries()
if len(entries) != 1 {
t.Fatalf("got %d log entries, want 1", len(entries))
}
if entries[0].Level != tt.wantLevel {
t.Errorf("level = %v, want %v", entries[0].Level, tt.wantLevel)
}
if entries[0].Message != tt.wantMsg {
t.Errorf("message = %q, want %q", entries[0].Message, tt.wantMsg)
}
if relay := entries[0].Data["relay"]; relay != "relay.example.com:443" {
t.Errorf("relay field = %v, want relay.example.com:443", relay)
}
})
}
}
// Events the relay client does not care about must not produce log lines.
func TestLogSinkIgnoresUnhandledEvents(t *testing.T) {
logger, hook := test.NewNullLogger()
logger.SetLevel(log.DebugLevel)
recorder := logSink{log: logger.WithField("relay", "relay.example.com:443")}
recorder.RecordEvent(qlog.PacketLost{})
if entries := hook.AllEntries(); len(entries) != 0 {
t.Errorf("got %d log entries, want 0", len(entries))
}
}
func TestLogSinkSupportsSchemas(t *testing.T) {
trace := logSink{log: log.WithField("relay", "relay.example.com:443")}
if !trace.SupportsSchemas(qlog.EventSchema) {
t.Errorf("SupportsSchemas(%q) = false, want true", qlog.EventSchema)
}
if trace.SupportsSchemas("urn:ietf:params:qlog:events:http3-12") {
t.Error("SupportsSchemas() = true for an unrelated schema, want false")
}
if trace.AddProducer() == nil {
t.Error("AddProducer() = nil, want a recorder")
}
}
+70
View File
@@ -0,0 +1,70 @@
package quic
import (
"context"
"fmt"
"strings"
"github.com/quic-go/quic-go"
"github.com/quic-go/quic-go/qlog"
"github.com/quic-go/quic-go/qlogwriter"
log "github.com/sirupsen/logrus"
)
// logSink implements both qlogwriter.Trace and qlogwriter.Recorder, forwarding
// the few qlog events the relay client cares about to logrus instead of
// writing a qlog file. It holds no mutable state and logrus entries are safe
// to share, so one value can serve every producer on the connection.
type logSink struct {
log *log.Entry
}
func (s logSink) AddProducer() qlogwriter.Recorder { return s }
func (s logSink) SupportsSchemas(schema string) bool { return schema == qlog.EventSchema }
func (s logSink) RecordEvent(event qlogwriter.Event) {
switch e := event.(type) {
case qlog.MTUUpdated:
if e.Done {
s.log.Infof("QUIC path MTU settled at %d", e.Value)
return
}
s.log.Debugf("QUIC path MTU probing at %d", e.Value)
case qlog.ConnectionClosed:
s.log.Debugf("QUIC connection closed: %s", closeReason(e))
}
}
func (s logSink) Close() error { return nil }
// connectionTracer returns a QUIC tracer that logs the DPLPMTUD result and the
// reason a relay connection closed, so the path MTU settled on and teardown
// cause are visible in logs. Lines carry the relay address as a structured
// field, matching the rest of the relay client logging.
func connectionTracer(addr string) func(context.Context, bool, quic.ConnectionID) qlogwriter.Trace {
relayLog := log.WithField("relay", addr)
return func(context.Context, bool, quic.ConnectionID) qlogwriter.Trace {
return logSink{log: relayLog}
}
}
// closeReason renders a ConnectionClosed event as a single line. The event
// carries the error as separate initiator, code, trigger and reason fields,
// any of which may be unset.
func closeReason(e qlog.ConnectionClosed) string {
parts := []string{fmt.Sprintf("closed by %s", e.Initiator)}
switch {
case e.ConnectionError != nil:
parts = append(parts, fmt.Sprintf("transport error: %s", *e.ConnectionError))
case e.ApplicationError != nil:
parts = append(parts, fmt.Sprintf("application error: %d", *e.ApplicationError))
}
if e.Trigger != "" {
parts = append(parts, fmt.Sprintf("trigger: %s", e.Trigger))
}
if e.Reason != "" {
parts = append(parts, fmt.Sprintf("reason: %s", e.Reason))
}
return strings.Join(parts, ", ")
}
+87 -5
View File
@@ -3,6 +3,7 @@ package dialer
import (
"context"
"errors"
"fmt"
"net"
"time"
@@ -14,7 +15,9 @@ const (
)
type DialeFn interface {
Dial(ctx context.Context, address string) (net.Conn, error)
// Dial connects to address. serverName, when non-empty, overrides the TLS
// ServerName used for SNI/cert validation. Empty means derive from address.
Dial(ctx context.Context, address, serverName string) (net.Conn, error)
Protocol() string
}
@@ -27,8 +30,10 @@ type dialResult struct {
type RaceDial struct {
log *log.Entry
serverURL string
serverName string
dialerFns []DialeFn
connectionTimeout time.Duration
sequential bool
}
func NewRaceDial(log *log.Entry, connectionTimeout time.Duration, serverURL string, dialerFns ...DialeFn) *RaceDial {
@@ -40,9 +45,34 @@ func NewRaceDial(log *log.Entry, connectionTimeout time.Duration, serverURL stri
}
}
// WithServerName sets a TLS SNI/cert validation override. Used when serverURL
// contains an IP literal but the cert is issued for a different hostname.
//
// Mutates the receiver and is not safe for concurrent reconfiguration; a
// RaceDial is intended to be constructed per dial and discarded.
func (r *RaceDial) WithServerName(serverName string) *RaceDial {
r.serverName = serverName
return r
}
// WithSequential makes Dial try the dialers in order, falling back to the next
// only when one fails to connect, instead of racing them concurrently.
//
// Mutates the receiver and is not safe for concurrent reconfiguration; a
// RaceDial is intended to be constructed per dial and discarded.
func (r *RaceDial) WithSequential() *RaceDial {
r.sequential = true
return r
}
func (r *RaceDial) Dial(ctx context.Context) (net.Conn, error) {
if r.sequential {
return r.dialSequential(ctx)
}
connChan := make(chan dialResult, len(r.dialerFns))
winnerConn := make(chan net.Conn, 1)
errChan := make(chan error, 1)
abortCtx, abort := context.WithCancel(ctx)
defer abort()
@@ -50,26 +80,53 @@ func (r *RaceDial) Dial(ctx context.Context) (net.Conn, error) {
go r.dial(dfn, abortCtx, connChan)
}
go r.processResults(connChan, winnerConn, abort)
go r.processResults(connChan, winnerConn, errChan, abort)
conn, ok := <-winnerConn
if !ok {
return nil, errors.New("failed to dial to Relay server on any protocol")
return nil, <-errChan
}
return conn, nil
}
// dialSequential tries each dialer in order, returning the first connection and
// falling back to the next on failure.
func (r *RaceDial) dialSequential(ctx context.Context) (net.Conn, error) {
var errs []error
for _, dfn := range r.dialerFns {
if err := ctx.Err(); err != nil {
return nil, err
}
attemptCtx, cancel := context.WithTimeout(ctx, r.connectionTimeout)
r.log.Infof("dialing Relay server via %s", dfn.Protocol())
conn, err := dfn.Dial(attemptCtx, r.serverURL, r.serverName)
cancel()
if err != nil {
if errors.Is(err, context.Canceled) {
return nil, err
}
r.log.Errorf("failed to dial via %s: %s", dfn.Protocol(), err)
errs = append(errs, fmt.Errorf("%s: %w", dfn.Protocol(), err))
continue
}
r.log.Infof("successfully dialed via: %s", dfn.Protocol())
return conn, nil
}
return nil, dialErr(errs)
}
func (r *RaceDial) dial(dfn DialeFn, abortCtx context.Context, connChan chan dialResult) {
ctx, cancel := context.WithTimeout(abortCtx, r.connectionTimeout)
defer cancel()
r.log.Infof("dialing Relay server via %s", dfn.Protocol())
conn, err := dfn.Dial(ctx, r.serverURL)
conn, err := dfn.Dial(ctx, r.serverURL, r.serverName)
connChan <- dialResult{Conn: conn, Protocol: dfn.Protocol(), Err: err}
}
func (r *RaceDial) processResults(connChan chan dialResult, winnerConn chan net.Conn, abort context.CancelFunc) {
func (r *RaceDial) processResults(connChan chan dialResult, winnerConn chan net.Conn, errChan chan error, abort context.CancelFunc) {
var hasWinner bool
errsByProtocol := make(map[string]error)
for i := 0; i < len(r.dialerFns); i++ {
dr := <-connChan
if dr.Err != nil {
@@ -77,6 +134,7 @@ func (r *RaceDial) processResults(connChan chan dialResult, winnerConn chan net.
r.log.Infof("connection attempt aborted via: %s", dr.Protocol)
} else {
r.log.Errorf("failed to dial via %s: %s", dr.Protocol, dr.Err)
errsByProtocol[dr.Protocol] = fmt.Errorf("%s: %w", dr.Protocol, dr.Err)
}
continue
}
@@ -94,5 +152,29 @@ func (r *RaceDial) processResults(connChan chan dialResult, winnerConn chan net.
hasWinner = true
winnerConn <- dr.Conn
}
if !hasWinner {
errChan <- dialErr(r.orderedErrs(errsByProtocol))
}
close(winnerConn)
}
// orderedErrs returns the per-protocol errors in dialer order, so the combined
// error is stable regardless of which attempt failed first.
func (r *RaceDial) orderedErrs(byProtocol map[string]error) []error {
errs := make([]error, 0, len(byProtocol))
for _, dfn := range r.dialerFns {
if err, ok := byProtocol[dfn.Protocol()]; ok {
errs = append(errs, err)
}
}
return errs
}
// dialErr combines per-dialer failures, preserving the underlying reasons
// (e.g. "connection refused") rather than a generic message.
func dialErr(errs []error) error {
if len(errs) == 0 {
return errors.New("no relay transport available")
}
return errors.Join(errs...)
}
+64 -1
View File
@@ -28,7 +28,7 @@ type MockDialer struct {
protocolStr string
}
func (m *MockDialer) Dial(ctx context.Context, address string) (net.Conn, error) {
func (m *MockDialer) Dial(ctx context.Context, address, _ string) (net.Conn, error) {
return m.dialFunc(ctx, address)
}
@@ -250,3 +250,66 @@ func TestRaceDialFirstSuccessfulDialerWins(t *testing.T) {
}
}
}
func TestRaceDialSequentialFallback(t *testing.T) {
logger := logrus.NewEntry(logrus.New())
serverURL := "test.server.com"
var firstDialed, secondDialed bool
preferred := &MockDialer{
protocolStr: "quic",
dialFunc: func(ctx context.Context, address string) (net.Conn, error) {
firstDialed = true
return nil, errors.New("quic unreachable")
},
}
fallbackConn := &MockConn{remoteAddr: &MockAddr{network: "ws"}}
fallback := &MockDialer{
protocolStr: "ws",
dialFunc: func(ctx context.Context, address string) (net.Conn, error) {
secondDialed = true
return fallbackConn, nil
},
}
rd := NewRaceDial(logger, DefaultConnectionTimeout, serverURL, preferred, fallback).WithSequential()
conn, err := rd.Dial(context.Background())
if err != nil {
t.Fatalf("expected fallback to succeed, got %v", err)
}
if conn != fallbackConn {
t.Errorf("expected fallback connection, got %v", conn)
}
if !firstDialed || !secondDialed {
t.Errorf("expected both dialers attempted in order, first=%v second=%v", firstDialed, secondDialed)
}
}
func TestRaceDialSequentialPreferredWins(t *testing.T) {
logger := logrus.NewEntry(logrus.New())
serverURL := "test.server.com"
preferredConn := &MockConn{remoteAddr: &MockAddr{network: "quic"}}
preferred := &MockDialer{
protocolStr: "quic",
dialFunc: func(ctx context.Context, address string) (net.Conn, error) {
return preferredConn, nil
},
}
fallback := &MockDialer{
protocolStr: "ws",
dialFunc: func(ctx context.Context, address string) (net.Conn, error) {
t.Errorf("fallback dialer must not be tried when preferred succeeds")
return nil, errors.New("should not happen")
},
}
rd := NewRaceDial(logger, DefaultConnectionTimeout, serverURL, preferred, fallback).WithSequential()
conn, err := rd.Dial(context.Background())
if err != nil {
t.Fatalf("expected preferred to succeed, got %v", err)
}
if conn != preferredConn {
t.Errorf("expected preferred connection, got %v", conn)
}
}
@@ -0,0 +1,9 @@
//go:build !js
package ws
// closeConn closes the underlying WebSocket immediately, skipping the close
// handshake.
func (c *Conn) closeConn() error {
return c.Conn.CloseNow()
}
+25
View File
@@ -0,0 +1,25 @@
//go:build js
package ws
import (
"github.com/coder/websocket"
log "github.com/sirupsen/logrus"
)
// closeConn closes the browser WebSocket without blocking the caller.
//
// The browser close API only accepts codes 1000 and 3000-4999, so CloseNow's
// 1001 (going away) throws an InvalidAccessError. Close with a valid code
// waits for the browser close event before returning, which can park the
// calling goroutine (the relay teardown path holds its mutexes while closing)
// until the close handshake finishes. Run the close in the background and
// report success; a teardown close error is not actionable.
func (c *Conn) closeConn() error {
go func() {
if err := c.Conn.Close(websocket.StatusNormalClosure, ""); err != nil {
log.Debugf("failed to close relay websocket: %v", err)
}
}()
return nil
}
+19 -4
View File
@@ -12,17 +12,32 @@ import (
type Conn struct {
ctx context.Context
*websocket.Conn
remoteAddr WebsocketAddr
remoteAddr net.Addr
}
func NewConn(wsConn *websocket.Conn, serverAddress string) net.Conn {
// NewConn builds a relay ws.Conn. underlying is the raw TCP/TLS conn captured
// from the http transport's DialContext; when set, RemoteAddr returns its
// peer address (an IP literal). When nil (e.g. wasm), RemoteAddr falls back
// to the dial-time URL.
func NewConn(wsConn *websocket.Conn, serverAddress string, underlying net.Conn) net.Conn {
var addr net.Addr = WebsocketAddr{serverAddress}
if underlying != nil {
if ra := underlying.RemoteAddr(); ra != nil {
addr = ra
}
}
return &Conn{
ctx: context.Background(),
Conn: wsConn,
remoteAddr: WebsocketAddr{serverAddress},
remoteAddr: addr,
}
}
// Protocol returns the transport name for this connection.
func (c *Conn) Protocol() string {
return Network
}
func (c *Conn) Read(b []byte) (n int, err error) {
t, ioReader, err := c.Conn.Reader(c.ctx)
if err != nil {
@@ -62,5 +77,5 @@ func (c *Conn) SetDeadline(t time.Time) error {
}
func (c *Conn) Close() error {
return c.Conn.CloseNow()
return c.closeConn()
}
@@ -2,10 +2,14 @@
package ws
import "github.com/coder/websocket"
import (
"net"
func createDialOptions() *websocket.DialOptions {
"github.com/coder/websocket"
)
func createDialOptions(serverName string, underlyingOut *net.Conn) *websocket.DialOptions {
return &websocket.DialOptions{
HTTPClient: httpClientNbDialer(),
HTTPClient: httpClientNbDialer(serverName, underlyingOut),
}
}
+7 -3
View File
@@ -2,9 +2,13 @@
package ws
import "github.com/coder/websocket"
import (
"net"
func createDialOptions() *websocket.DialOptions {
// WASM version doesn't support HTTPClient
"github.com/coder/websocket"
)
func createDialOptions(_ string, _ *net.Conn) *websocket.DialOptions {
// WASM version doesn't support HTTPClient or custom TLS config.
return &websocket.DialOptions{}
}
+42 -20
View File
@@ -9,7 +9,6 @@ import (
"net"
"net/http"
"net/url"
"strings"
"github.com/coder/websocket"
log "github.com/sirupsen/logrus"
@@ -23,48 +22,66 @@ type Dialer struct {
}
func (d Dialer) Protocol() string {
return "WS"
return Network
}
func (d Dialer) Dial(ctx context.Context, address string) (net.Conn, error) {
func (d Dialer) Dial(ctx context.Context, address, serverName string) (net.Conn, error) {
wsURL, err := prepareURL(address)
if err != nil {
return nil, err
}
opts := createDialOptions()
var underlying net.Conn
opts := createDialOptions(serverName, &underlying)
parsedURL, err := url.Parse(wsURL)
if err != nil {
return nil, err
}
parsedURL.Path = relay.WebSocketURLPath
wsConn, resp, err := websocket.Dial(ctx, parsedURL.String(), opts)
wsConn, resp, err := websocket.Dial(ctx, wsURL, opts)
if err != nil {
if errors.Is(err, context.Canceled) {
return nil, err
}
log.Errorf("failed to dial to Relay server '%s': %s", wsURL, err)
// websocket.Dial wraps the cause in verbose layers; surface the
// underlying network error when present.
var opErr *net.OpError
if errors.As(err, &opErr) {
return nil, opErr
}
return nil, err
}
if resp.Body != nil {
_ = resp.Body.Close()
}
conn := NewConn(wsConn, address)
conn := NewConn(wsConn, address, underlying)
return conn, nil
}
// prepareURL rewrites a rel://host[:port] or rels://host[:port] address into a
// ws://host[:port]/relay or wss://host[:port]/relay URL, preserving any
// non-standard port from the input.
func prepareURL(address string) (string, error) {
if !strings.HasPrefix(address, "rel:") && !strings.HasPrefix(address, "rels:") {
return "", fmt.Errorf("unsupported scheme: %s", address)
parsed, err := url.Parse(address)
if err != nil {
return "", fmt.Errorf("parse relay address %q: %w", address, err)
}
return strings.Replace(address, "rel", "ws", 1), nil
switch parsed.Scheme {
case "rel":
parsed.Scheme = "ws"
case "rels":
parsed.Scheme = "wss"
default:
return "", fmt.Errorf("unsupported scheme: %s", parsed.Scheme)
}
if parsed.Host == "" {
return "", fmt.Errorf("missing host in relay address %q", address)
}
parsed.Path = relay.WebSocketURLPath
return parsed.String(), nil
}
func httpClientNbDialer() *http.Client {
// httpClientNbDialer builds the http client used by the websocket library.
// underlyingOut, when non-nil, is populated with the raw conn from the
// transport's DialContext so the caller can read its RemoteAddr.
func httpClientNbDialer(serverName string, underlyingOut *net.Conn) *http.Client {
customDialer := nbnet.NewDialer()
certPool, err := x509.SystemCertPool()
@@ -75,10 +92,15 @@ func httpClientNbDialer() *http.Client {
customTransport := &http.Transport{
DialContext: func(ctx context.Context, network, addr string) (net.Conn, error) {
return customDialer.DialContext(ctx, network, addr)
c, err := customDialer.DialContext(ctx, network, addr)
if err == nil && underlyingOut != nil {
*underlyingOut = c
}
return c, err
},
TLSClientConfig: &tls.Config{
RootCAs: certPool,
RootCAs: certPool,
ServerName: serverName,
},
}
+76
View File
@@ -0,0 +1,76 @@
package ws
import (
"testing"
)
func TestPrepareURL(t *testing.T) {
tests := []struct {
name string
input string
want string
wantErr bool
}{
{
name: "rel scheme with non-standard port",
input: "rel://test-domain-2:45678",
want: "ws://test-domain-2:45678/relay",
},
{
name: "rels scheme with non-standard port",
input: "rels://test-domain-2:45678",
want: "wss://test-domain-2:45678/relay",
},
{
name: "rel scheme without port",
input: "rel://test-domain-2",
want: "ws://test-domain-2/relay",
},
{
name: "rels scheme without port",
input: "rels://test-domain-2",
want: "wss://test-domain-2/relay",
},
{
name: "rel scheme with IP and port",
input: "rel://1.2.3.4:45678",
want: "ws://1.2.3.4:45678/relay",
},
{
name: "rel scheme with hostname starting with rel",
input: "rel://relay.example.com:45678",
want: "ws://relay.example.com:45678/relay",
},
{
name: "rel scheme with IPv6 and port",
input: "rel://[2001:db8::1]:45678",
want: "ws://[2001:db8::1]:45678/relay",
},
{
name: "rels scheme with IPv6 loopback and port",
input: "rels://[::1]:45678",
want: "wss://[::1]:45678/relay",
},
{
name: "unsupported scheme",
input: "http://test-domain-2:45678",
wantErr: true,
},
{
name: "no scheme",
input: "test-domain-2:45678",
wantErr: true,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
got, err := prepareURL(tt.input)
if (err != nil) != tt.wantErr {
t.Fatalf("prepareURL(%q) err = %v, wantErr %v", tt.input, err, tt.wantErr)
}
if got != tt.want {
t.Errorf("prepareURL(%q) = %q, want %q", tt.input, got, tt.want)
}
})
}
}

Some files were not shown because too many files have changed in this diff Show More