[management, reverse proxy] Add reverse proxy feature (#5291)

* implement reverse proxy


---------

Co-authored-by: Alisdair MacLeod <git@alisdairmacleod.co.uk>
Co-authored-by: mlsmaycon <mlsmaycon@gmail.com>
Co-authored-by: Eduard Gert <kontakt@eduardgert.de>
Co-authored-by: Viktor Liu <viktor@netbird.io>
Co-authored-by: Diego Noguês <diego.sure@gmail.com>
Co-authored-by: Diego Noguês <49420+diegocn@users.noreply.github.com>
Co-authored-by: Bethuel Mmbaga <bethuelmbaga12@gmail.com>
Co-authored-by: Zoltan Papp <zoltan.pmail@gmail.com>
Co-authored-by: Ashley Mensah <ashleyamo982@gmail.com>

management/internals/modules/reverseproxy/domain/domain.go

@@ -0,0 +1,17 @@
+package domain
+
+type Type string
+
+const (
+	TypeFree   Type = "free"
+	TypeCustom Type = "custom"
+)
+
+type Domain struct {
+	ID            string `gorm:"unique;primaryKey;autoIncrement"`
+	Domain        string `gorm:"unique"` // Domain records must be unique, this avoids domain reuse across accounts.
+	AccountID     string `gorm:"index"`
+	TargetCluster string // The proxy cluster this domain should be validated against
+	Type          Type   `gorm:"-"`
+	Validated     bool
+}

management/internals/modules/reverseproxy/domain/interface.go

@@ -0,0 +1,12 @@
+package domain
+
+import (
+	"context"
+)
+
+type Manager interface {
+	GetDomains(ctx context.Context, accountID, userID string) ([]*Domain, error)
+	CreateDomain(ctx context.Context, accountID, userID, domainName, targetCluster string) (*Domain, error)
+	DeleteDomain(ctx context.Context, accountID, userID, domainID string) error
+	ValidateDomain(ctx context.Context, accountID, userID, domainID string)
+}

management/internals/modules/reverseproxy/domain/manager/api.go

@@ -0,0 +1,136 @@
+package manager
+
+import (
+	"encoding/json"
+	"net/http"
+
+	"github.com/gorilla/mux"
+
+	"github.com/netbirdio/netbird/management/internals/modules/reverseproxy/domain"
+	nbcontext "github.com/netbirdio/netbird/management/server/context"
+	"github.com/netbirdio/netbird/shared/management/http/api"
+	"github.com/netbirdio/netbird/shared/management/http/util"
+	"github.com/netbirdio/netbird/shared/management/status"
+)
+
+type handler struct {
+	manager Manager
+}
+
+func RegisterEndpoints(router *mux.Router, manager Manager) {
+	h := &handler{
+		manager: manager,
+	}
+
+	router.HandleFunc("/domains", h.getAllDomains).Methods("GET", "OPTIONS")
+	router.HandleFunc("/domains", h.createCustomDomain).Methods("POST", "OPTIONS")
+	router.HandleFunc("/domains/{domainId}", h.deleteCustomDomain).Methods("DELETE", "OPTIONS")
+	router.HandleFunc("/domains/{domainId}/validate", h.triggerCustomDomainValidation).Methods("GET", "OPTIONS")
+}
+
+func domainTypeToApi(t domain.Type) api.ReverseProxyDomainType {
+	switch t {
+	case domain.TypeCustom:
+		return api.ReverseProxyDomainTypeCustom
+	case domain.TypeFree:
+		return api.ReverseProxyDomainTypeFree
+	}
+	// By default return as a "free" domain as that is more restrictive.
+	// TODO: is this correct?
+	return api.ReverseProxyDomainTypeFree
+}
+
+func domainToApi(d *domain.Domain) api.ReverseProxyDomain {
+	resp := api.ReverseProxyDomain{
+		Domain:    d.Domain,
+		Id:        d.ID,
+		Type:      domainTypeToApi(d.Type),
+		Validated: d.Validated,
+	}
+	if d.TargetCluster != "" {
+		resp.TargetCluster = &d.TargetCluster
+	}
+	return resp
+}
+
+func (h *handler) getAllDomains(w http.ResponseWriter, r *http.Request) {
+	userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
+	if err != nil {
+		util.WriteError(r.Context(), err, w)
+		return
+	}
+
+	domains, err := h.manager.GetDomains(r.Context(), userAuth.AccountId, userAuth.UserId)
+	if err != nil {
+		util.WriteError(r.Context(), err, w)
+		return
+	}
+
+	ret := make([]api.ReverseProxyDomain, 0)
+	for _, d := range domains {
+		ret = append(ret, domainToApi(d))
+	}
+
+	util.WriteJSONObject(r.Context(), w, ret)
+}
+
+func (h *handler) createCustomDomain(w http.ResponseWriter, r *http.Request) {
+	userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
+	if err != nil {
+		util.WriteError(r.Context(), err, w)
+		return
+	}
+
+	var req api.PostApiReverseProxiesDomainsJSONRequestBody
+	if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
+		util.WriteErrorResponse("couldn't parse JSON request", http.StatusBadRequest, w)
+		return
+	}
+
+	domain, err := h.manager.CreateDomain(r.Context(), userAuth.AccountId, userAuth.UserId, req.Domain, req.TargetCluster)
+	if err != nil {
+		util.WriteError(r.Context(), err, w)
+		return
+	}
+
+	util.WriteJSONObject(r.Context(), w, domainToApi(domain))
+}
+
+func (h *handler) deleteCustomDomain(w http.ResponseWriter, r *http.Request) {
+	userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
+	if err != nil {
+		util.WriteError(r.Context(), err, w)
+		return
+	}
+
+	domainID := mux.Vars(r)["domainId"]
+	if domainID == "" {
+		util.WriteError(r.Context(), status.Errorf(status.InvalidArgument, "domain ID is required"), w)
+		return
+	}
+
+	if err := h.manager.DeleteDomain(r.Context(), userAuth.AccountId, userAuth.UserId, domainID); err != nil {
+		util.WriteError(r.Context(), err, w)
+		return
+	}
+
+	w.WriteHeader(http.StatusNoContent)
+}
+
+func (h *handler) triggerCustomDomainValidation(w http.ResponseWriter, r *http.Request) {
+	userAuth, err := nbcontext.GetUserAuthFromContext(r.Context())
+	if err != nil {
+		util.WriteError(r.Context(), err, w)
+		return
+	}
+
+	domainID := mux.Vars(r)["domainId"]
+	if domainID == "" {
+		util.WriteError(r.Context(), status.Errorf(status.InvalidArgument, "domain ID is required"), w)
+		return
+	}
+
+	go h.manager.ValidateDomain(r.Context(), userAuth.AccountId, userAuth.UserId, domainID)
+
+	w.WriteHeader(http.StatusAccepted)
+}

management/internals/modules/reverseproxy/domain/manager/manager.go

@@ -0,0 +1,279 @@
+package manager
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"strings"
+
+	log "github.com/sirupsen/logrus"
+
+	"github.com/netbirdio/netbird/management/internals/modules/reverseproxy/domain"
+	"github.com/netbirdio/netbird/management/server/permissions"
+	"github.com/netbirdio/netbird/management/server/permissions/modules"
+	"github.com/netbirdio/netbird/management/server/permissions/operations"
+	"github.com/netbirdio/netbird/management/server/types"
+	"github.com/netbirdio/netbird/shared/management/status"
+)
+
+type store interface {
+	GetAccount(ctx context.Context, accountID string) (*types.Account, error)
+
+	GetCustomDomain(ctx context.Context, accountID string, domainID string) (*domain.Domain, error)
+	ListFreeDomains(ctx context.Context, accountID string) ([]string, error)
+	ListCustomDomains(ctx context.Context, accountID string) ([]*domain.Domain, error)
+	CreateCustomDomain(ctx context.Context, accountID string, domainName string, targetCluster string, validated bool) (*domain.Domain, error)
+	UpdateCustomDomain(ctx context.Context, accountID string, d *domain.Domain) (*domain.Domain, error)
+	DeleteCustomDomain(ctx context.Context, accountID string, domainID string) error
+}
+
+type proxyURLProvider interface {
+	GetConnectedProxyURLs() []string
+}
+
+type Manager struct {
+	store              store
+	validator          domain.Validator
+	proxyURLProvider   proxyURLProvider
+	permissionsManager permissions.Manager
+}
+
+func NewManager(store store, proxyURLProvider proxyURLProvider, permissionsManager permissions.Manager) Manager {
+	return Manager{
+		store:            store,
+		proxyURLProvider: proxyURLProvider,
+		validator: domain.Validator{
+			Resolver: net.DefaultResolver,
+		},
+		permissionsManager: permissionsManager,
+	}
+}
+
+func (m Manager) GetDomains(ctx context.Context, accountID, userID string) ([]*domain.Domain, error) {
+	ok, err := m.permissionsManager.ValidateUserPermissions(ctx, accountID, userID, modules.Services, operations.Read)
+	if err != nil {
+		return nil, status.NewPermissionValidationError(err)
+	}
+	if !ok {
+		return nil, status.NewPermissionDeniedError()
+	}
+
+	domains, err := m.store.ListCustomDomains(ctx, accountID)
+	if err != nil {
+		return nil, fmt.Errorf("list custom domains: %w", err)
+	}
+
+	var ret []*domain.Domain
+
+	// Add connected proxy clusters as free domains.
+	// The cluster address itself is the free domain base (e.g., "eu.proxy.netbird.io").
+	allowList := m.proxyURLAllowList()
+	log.WithFields(log.Fields{
+		"accountID":      accountID,
+		"proxyAllowList": allowList,
+	}).Debug("getting domains with proxy allow list")
+
+	for _, cluster := range allowList {
+		ret = append(ret, &domain.Domain{
+			Domain:    cluster,
+			AccountID: accountID,
+			Type:      domain.TypeFree,
+			Validated: true,
+		})
+	}
+
+	// Add custom domains.
+	for _, d := range domains {
+		ret = append(ret, &domain.Domain{
+			ID:            d.ID,
+			Domain:        d.Domain,
+			AccountID:     accountID,
+			TargetCluster: d.TargetCluster,
+			Type:          domain.TypeCustom,
+			Validated:     d.Validated,
+		})
+	}
+
+	return ret, nil
+}
+
+func (m Manager) CreateDomain(ctx context.Context, accountID, userID, domainName, targetCluster string) (*domain.Domain, error) {
+	ok, err := m.permissionsManager.ValidateUserPermissions(ctx, accountID, userID, modules.Services, operations.Create)
+	if err != nil {
+		return nil, status.NewPermissionValidationError(err)
+	}
+	if !ok {
+		return nil, status.NewPermissionDeniedError()
+	}
+
+	// Verify the target cluster is in the available clusters
+	allowList := m.proxyURLAllowList()
+	clusterValid := false
+	for _, cluster := range allowList {
+		if cluster == targetCluster {
+			clusterValid = true
+			break
+		}
+	}
+	if !clusterValid {
+		return nil, fmt.Errorf("target cluster %s is not available", targetCluster)
+	}
+
+	// Attempt an initial validation against the specified cluster only
+	var validated bool
+	if m.validator.IsValid(ctx, domainName, []string{targetCluster}) {
+		validated = true
+	}
+
+	d, err := m.store.CreateCustomDomain(ctx, accountID, domainName, targetCluster, validated)
+	if err != nil {
+		return d, fmt.Errorf("create domain in store: %w", err)
+	}
+	return d, nil
+}
+
+func (m Manager) DeleteDomain(ctx context.Context, accountID, userID, domainID string) error {
+	ok, err := m.permissionsManager.ValidateUserPermissions(ctx, accountID, userID, modules.Services, operations.Delete)
+	if err != nil {
+		return status.NewPermissionValidationError(err)
+	}
+	if !ok {
+		return status.NewPermissionDeniedError()
+	}
+
+	if err := m.store.DeleteCustomDomain(ctx, accountID, domainID); err != nil {
+		// TODO: check for "no records" type error. Because that is a success condition.
+		return fmt.Errorf("delete domain from store: %w", err)
+	}
+	return nil
+}
+
+func (m Manager) ValidateDomain(ctx context.Context, accountID, userID, domainID string) {
+	ok, err := m.permissionsManager.ValidateUserPermissions(ctx, accountID, userID, modules.Services, operations.Create)
+	if err != nil {
+		log.WithFields(log.Fields{
+			"accountID": accountID,
+			"domainID":  domainID,
+		}).WithError(err).Error("validate domain")
+		return
+	}
+	if !ok {
+		log.WithFields(log.Fields{
+			"accountID": accountID,
+			"domainID":  domainID,
+		}).WithError(err).Error("validate domain")
+	}
+
+	log.WithFields(log.Fields{
+		"accountID": accountID,
+		"domainID":  domainID,
+	}).Info("starting domain validation")
+
+	d, err := m.store.GetCustomDomain(context.Background(), accountID, domainID)
+	if err != nil {
+		log.WithFields(log.Fields{
+			"accountID": accountID,
+			"domainID":  domainID,
+		}).WithError(err).Error("get custom domain from store")
+		return
+	}
+
+	// Validate only against the domain's target cluster
+	targetCluster := d.TargetCluster
+	if targetCluster == "" {
+		log.WithFields(log.Fields{
+			"accountID": accountID,
+			"domainID":  domainID,
+			"domain":    d.Domain,
+		}).Warn("domain has no target cluster set, skipping validation")
+		return
+	}
+
+	log.WithFields(log.Fields{
+		"accountID":     accountID,
+		"domainID":      domainID,
+		"domain":        d.Domain,
+		"targetCluster": targetCluster,
+	}).Info("validating domain against target cluster")
+
+	if m.validator.IsValid(context.Background(), d.Domain, []string{targetCluster}) {
+		log.WithFields(log.Fields{
+			"accountID": accountID,
+			"domainID":  domainID,
+			"domain":    d.Domain,
+		}).Info("domain validated successfully")
+		d.Validated = true
+		if _, err := m.store.UpdateCustomDomain(context.Background(), accountID, d); err != nil {
+			log.WithFields(log.Fields{
+				"accountID": accountID,
+				"domainID":  domainID,
+				"domain":    d.Domain,
+			}).WithError(err).Error("update custom domain in store")
+			return
+		}
+	} else {
+		log.WithFields(log.Fields{
+			"accountID":     accountID,
+			"domainID":      domainID,
+			"domain":        d.Domain,
+			"targetCluster": targetCluster,
+		}).Warn("domain validation failed - CNAME does not match target cluster")
+	}
+}
+
+// proxyURLAllowList retrieves a list of currently connected proxies and
+// their URLs
+func (m Manager) proxyURLAllowList() []string {
+	var reverseProxyAddresses []string
+	if m.proxyURLProvider != nil {
+		reverseProxyAddresses = m.proxyURLProvider.GetConnectedProxyURLs()
+	}
+	return reverseProxyAddresses
+}
+
+// DeriveClusterFromDomain determines the proxy cluster for a given domain.
+// For free domains (those ending with a known cluster suffix), the cluster is extracted from the domain.
+// For custom domains, the cluster is determined by checking the registered custom domain's target cluster.
+func (m Manager) DeriveClusterFromDomain(ctx context.Context, accountID, domain string) (string, error) {
+	allowList := m.proxyURLAllowList()
+	if len(allowList) == 0 {
+		return "", fmt.Errorf("no proxy clusters available")
+	}
+
+	if cluster, ok := ExtractClusterFromFreeDomain(domain, allowList); ok {
+		return cluster, nil
+	}
+
+	customDomains, err := m.store.ListCustomDomains(ctx, accountID)
+	if err != nil {
+		return "", fmt.Errorf("list custom domains: %w", err)
+	}
+
+	targetCluster, valid := extractClusterFromCustomDomains(domain, customDomains)
+	if valid {
+		return targetCluster, nil
+	}
+
+	return "", fmt.Errorf("domain %s does not match any available proxy cluster", domain)
+}
+
+func extractClusterFromCustomDomains(domain string, customDomains []*domain.Domain) (string, bool) {
+	for _, customDomain := range customDomains {
+		if strings.HasSuffix(domain, "."+customDomain.Domain) {
+			return customDomain.TargetCluster, true
+		}
+	}
+	return "", false
+}
+
+// ExtractClusterFromFreeDomain extracts the cluster address from a free domain.
+// Free domains have the format: <name>.<nonce>.<cluster> (e.g., myapp.abc123.eu.proxy.netbird.io)
+// It matches the domain suffix against available clusters and returns the matching cluster.
+func ExtractClusterFromFreeDomain(domain string, availableClusters []string) (string, bool) {
+	for _, cluster := range availableClusters {
+		if strings.HasSuffix(domain, "."+cluster) {
+			return cluster, true
+		}
+	}
+	return "", false
+}

management/internals/modules/reverseproxy/domain/validator.go

@@ -0,0 +1,88 @@
+package domain
+
+import (
+	"context"
+	"net"
+	"strings"
+
+	log "github.com/sirupsen/logrus"
+)
+
+type resolver interface {
+	LookupCNAME(context.Context, string) (string, error)
+}
+
+type Validator struct {
+	Resolver resolver
+}
+
+// NewValidator initializes a validator with a specific DNS Resolver.
+// If a Validator is used without specifying a Resolver, then it will
+// use the net.DefaultResolver.
+func NewValidator(resolver resolver) *Validator {
+	return &Validator{
+		Resolver: resolver,
+	}
+}
+
+// IsValid looks up the CNAME record for the passed domain with a prefix
+// and compares it against the acceptable domains.
+// If the returned CNAME matches any accepted domain, it will return true,
+// otherwise, including in the event of a DNS error, it will return false.
+// The comparison is very simple, so wildcards will not match if included
+// in the acceptable domain list.
+func (v *Validator) IsValid(ctx context.Context, domain string, accept []string) bool {
+	_, valid := v.ValidateWithCluster(ctx, domain, accept)
+	return valid
+}
+
+// ValidateWithCluster validates a custom domain and returns the matched cluster address.
+// Returns the cluster address and true if valid, or empty string and false if invalid.
+func (v *Validator) ValidateWithCluster(ctx context.Context, domain string, accept []string) (string, bool) {
+	if v.Resolver == nil {
+		v.Resolver = net.DefaultResolver
+	}
+
+	lookupDomain := "validation." + domain
+	log.WithFields(log.Fields{
+		"domain":       domain,
+		"lookupDomain": lookupDomain,
+		"acceptList":   accept,
+	}).Debug("looking up CNAME for domain validation")
+
+	cname, err := v.Resolver.LookupCNAME(ctx, lookupDomain)
+	if err != nil {
+		log.WithFields(log.Fields{
+			"domain":       domain,
+			"lookupDomain": lookupDomain,
+		}).WithError(err).Warn("CNAME lookup failed for domain validation")
+		return "", false
+	}
+
+	nakedCNAME := strings.TrimSuffix(cname, ".")
+	log.WithFields(log.Fields{
+		"domain":     domain,
+		"cname":      cname,
+		"nakedCNAME": nakedCNAME,
+		"acceptList": accept,
+	}).Debug("CNAME lookup result for domain validation")
+
+	for _, acceptDomain := range accept {
+		normalizedAccept := strings.TrimSuffix(acceptDomain, ".")
+		if nakedCNAME == normalizedAccept {
+			log.WithFields(log.Fields{
+				"domain":  domain,
+				"cname":   nakedCNAME,
+				"cluster": acceptDomain,
+			}).Info("domain CNAME matched cluster")
+			return acceptDomain, true
+		}
+	}
+
+	log.WithFields(log.Fields{
+		"domain":     domain,
+		"cname":      nakedCNAME,
+		"acceptList": accept,
+	}).Warn("domain CNAME does not match any accepted cluster")
+	return "", false
+}

management/internals/modules/reverseproxy/domain/validator_test.go

@@ -0,0 +1,56 @@
+package domain_test
+
+import (
+	"context"
+	"testing"
+
+	"github.com/netbirdio/netbird/management/internals/modules/reverseproxy/domain"
+)
+
+type resolver struct {
+	CNAME string
+}
+
+func (r resolver) LookupCNAME(_ context.Context, _ string) (string, error) {
+	return r.CNAME, nil
+}
+
+func TestIsValid(t *testing.T) {
+	tests := map[string]struct {
+		resolver interface {
+			LookupCNAME(context.Context, string) (string, error)
+		}
+		domain string
+		accept []string
+		expect bool
+	}{
+		"match": {
+			resolver: resolver{"bar.example.com."}, // Including trailing "." in response.
+			domain:   "foo.example.com",
+			accept:   []string{"bar.example.com"},
+			expect:   true,
+		},
+		"no match": {
+			resolver: resolver{"invalid"},
+			domain:   "foo.example.com",
+			accept:   []string{"bar.example.com"},
+			expect:   false,
+		},
+		"accept trailing dot": {
+			resolver: resolver{"bar.example.com."},
+			domain:   "foo.example.com",
+			accept:   []string{"bar.example.com."}, // Including trailing "." in accept.
+			expect:   true,
+		},
+	}
+
+	for name, test := range tests {
+		t.Run(name, func(t *testing.T) {
+			validator := domain.NewValidator(test.resolver)
+			actual := validator.IsValid(t.Context(), test.domain, test.accept)
+			if test.expect != actual {
+				t.Errorf("Incorrect return value:\nexpect: %v\nactual: %v", test.expect, actual)
+			}
+		})
+	}
+}