[management] Prevent deleting custom domains used by services (#7515)

Deleting a custom domain released its name while services still pointed
at it, leaving them on a namespace the account no longer held.

Deletion now refuses with 412 when a service in the same account uses the
domain or a subdomain, including disabled ones. Service writes revalidate
authorization inside their transaction and hold a shared lock on the
matching registrations, so a delete racing a create cannot strand either.
The dependency lookup is account-scoped: registrations are unique by name,
so another account can hold team.example.com under example.com and its
services are authorized by its own registration.

management/internals/modules/reverseproxy/service/manager/domain_validation_test.go

@@ -125,3 +125,53 @@ func TestUpdateService_RefusesMoveToUnvalidatedDomain(t *testing.T) {
 	require.NoError(t, err)
 	assert.Equal(t, "app.proven.example.com", stored.Domain, "the service must keep its original domain")
 }
+
+func TestCreateService_DomainDeletedBeforeWrite(t *testing.T) {
+	ctx := context.Background()
+	mgr, testStore := setupIntegrationTest(t)
+	withRealDomainManager(t, mgr, testStore)
+
+	d, err := testStore.CreateCustomDomain(ctx, testAccountID, "proven.example.com", validationTestCluster, true)
+	require.NoError(t, err)
+	svc := newTestService("app.proven.example.com")
+	require.NoError(t, mgr.initializeServiceForCreate(ctx, testAccountID, svc))
+
+	// Delete after the initial authorization check, before the service transaction starts.
+	require.NoError(t, testStore.DeleteCustomDomain(ctx, testAccountID, d.ID))
+	err = mgr.persistNewService(ctx, testAccountID, svc)
+	require.Error(t, err, "an earlier validation result must not authorize a deleted registration")
+	sErr, ok := status.FromError(err)
+	require.True(t, ok, "the caller must receive a typed precondition error")
+	assert.Equal(t, status.PreconditionFailed, sErr.Type(), "the service must require current domain authorization")
+	services, err := testStore.GetAccountServices(ctx, store.LockingStrengthNone, testAccountID)
+	require.NoError(t, err)
+	assert.Empty(t, services, "the failed write must not leave a service")
+}
+
+func TestUpdateService_DomainDeletedBeforeWrite(t *testing.T) {
+	ctx := context.Background()
+	mgr, testStore := setupIntegrationTest(t)
+	withRealDomainManager(t, mgr, testStore)
+	_, err := testStore.CreateCustomDomain(ctx, testAccountID, "original.example.com", validationTestCluster, true)
+	require.NoError(t, err)
+	d, err := testStore.CreateCustomDomain(ctx, testAccountID, "destination.example.com", validationTestCluster, true)
+	require.NoError(t, err)
+	svc, err := mgr.CreateService(ctx, testAccountID, testUserID, newTestService("app.original.example.com"))
+	require.NoError(t, err)
+	moved := svc.Copy()
+	moved.Domain = "app.destination.example.com"
+	cluster, err := mgr.resolveEffectiveCluster(ctx, testAccountID, moved)
+	require.NoError(t, err)
+
+	require.NoError(t, testStore.DeleteCustomDomain(ctx, testAccountID, d.ID))
+	err = testStore.ExecuteInTransaction(ctx, func(tx store.Store) error {
+		return mgr.executeServiceUpdate(ctx, tx, testAccountID, moved, &serviceUpdateInfo{}, nil, cluster)
+	})
+	require.Error(t, err, "a domain deleted after cluster resolution must reject the update")
+	sErr, ok := status.FromError(err)
+	require.True(t, ok, "the caller must receive a typed precondition error")
+	assert.Equal(t, status.PreconditionFailed, sErr.Type(), "the move must require current domain authorization")
+	stored, err := testStore.GetServiceByID(ctx, store.LockingStrengthNone, testAccountID, svc.ID)
+	require.NoError(t, err)
+	assert.Equal(t, svc.Domain, stored.Domain, "the service must retain its authorized domain")
+}

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

@@ -74,6 +74,7 @@ const unknownHostPlaceholder = "unknown"
 // ClusterDeriver derives the proxy cluster from a domain.
 type ClusterDeriver interface {
 	DeriveClusterFromDomain(ctx context.Context, accountID, domain string) (string, error)
+	ValidateServiceDomain(ctx context.Context, tx store.Store, accountID, domain, cluster string) error
 	GetClusterDomains() []string
 }
 
@@ -332,6 +333,9 @@ func (m *Manager) persistNewService(ctx context.Context, accountID string, svc *
 	}
 
 	return m.store.ExecuteInTransaction(ctx, func(transaction store.Store) error {
+		if err := m.validateServiceDomain(ctx, transaction, accountID, svc, svc.ProxyCluster); err != nil {
+			return err
+		}
 		if svc.Domain != "" {
 			if err := m.checkDomainAvailable(ctx, transaction, svc.Domain, ""); err != nil {
 				return err
@@ -461,6 +465,9 @@ func (m *Manager) persistNewEphemeralService(ctx context.Context, accountID, pee
 	}
 
 	return m.store.ExecuteInTransaction(ctx, func(transaction store.Store) error {
+		if err := m.validateServiceDomain(ctx, transaction, accountID, svc, svc.ProxyCluster); err != nil {
+			return err
+		}
 		if err := m.validateEphemeralPreconditions(ctx, transaction, accountID, peerID, svc); err != nil {
 			return err
 		}
@@ -622,6 +629,9 @@ func (m *Manager) resolveEffectiveCluster(ctx context.Context, accountID string,
 }
 
 func (m *Manager) executeServiceUpdate(ctx context.Context, transaction store.Store, accountID string, service *service.Service, updateInfo *serviceUpdateInfo, customPorts *bool, effectiveCluster string) error {
+	if err := m.validateServiceDomain(ctx, transaction, accountID, service, effectiveCluster); err != nil {
+		return err
+	}
 	existingService, err := transaction.GetServiceByID(ctx, store.LockingStrengthUpdate, accountID, service.ID)
 	if err != nil {
 		return err
@@ -677,6 +687,13 @@ func (m *Manager) executeServiceUpdate(ctx context.Context, transaction store.St
 	return nil
 }
 
+func (m *Manager) validateServiceDomain(ctx context.Context, tx store.Store, accountID string, svc *service.Service, cluster string) error {
+	if m.clusterDeriver == nil {
+		return nil
+	}
+	return m.clusterDeriver.ValidateServiceDomain(ctx, tx, accountID, svc.Domain, cluster)
+}
+
 // validateL4PortDiffOnClusterDiff checks if custom L4 ports are configured and validates port changes across clusters.
 // It ensures no port changes if custom ports are unsupported for a given cluster and protocol mode.
 // Returns an error if validation fails, otherwise returns nil.

management/internals/modules/reverseproxy/service/manager/manager_test.go

@@ -655,6 +655,10 @@ func (d *testClusterDeriver) GetClusterDomains() []string {
 	return d.domains
 }
 
+func (d *testClusterDeriver) ValidateServiceDomain(context.Context, store.Store, string, string, string) error {
+	return nil
+}
+
 const (
 	testAccountID = "test-account"
 	testPeerID    = "test-peer-1"