feat: gate GPU Service on Kubernetes cluster availability

GPU Service today only schedules on Kubernetes clusters; Docker / cloud
clusters can't host the CRDs. Without any awareness of that, Org
members whose Org has no K8s cluster (and no cluster_access grant on
one) saw a menu they couldn't use and a form that bottomed out with
backend errors.

Two changes lock the UX down:

- Boot probes the caller's cluster list once and stashes
  hasKubernetesCluster in initialState. A new canSeeGpuService
  predicate gates the menu — admins and Org owners always see it
  (they can add the cluster); everyone else only sees it when a
  reachable K8s cluster actually exists.

- The instances page filters its own cluster list to Kubernetes
  before deciding what to show. With nothing reachable we render the
  Deployments-style 'No clusters available. Add a Kubernetes cluster
  to get started.' empty state and hide the create-instance CTA;
  admins and Org owners additionally get the 'Add cluster' button
  that jumps to cluster management.

src/access.extensions.ts

@@ -4,6 +4,7 @@
 export type AccessPredicates = {
   canSeeAdmin: boolean;
   canSeeOrgAdmin: boolean;
+  canSeeGpuService: boolean;
   canManageCurrentOrg: boolean;
   canSeeUser: boolean;
   canDelete: boolean;

src/access.ts

@@ -1,6 +1,9 @@
 import { applyAccessExtensions } from './access.extensions';
 
-export default (initialState: { currentUser?: Global.UserInfo }) => {
+export default (initialState: {
+  currentUser?: Global.UserInfo;
+  hasKubernetesCluster?: boolean;
+}) => {
   const isPlatformAdmin = !!(
     initialState &&
     initialState.currentUser &&
@@ -11,6 +14,12 @@ export default (initialState: { currentUser?: Global.UserInfo }) => {
     initialState.currentUser &&
     !initialState.currentUser.is_admin
   );
+  // GPU Service is Kubernetes-only. We only gate visibility down when
+  // the probe in `getInitialState` came back with a definitive answer;
+  // `undefined` (probe failed / not yet ready) collapses to the
+  // role-based default so a transient network blip can't lock anyone
+  // out of the menu.
+  const hasKubernetesCluster = initialState?.hasKubernetesCluster;
 
   // Predicate roles, top-down by strictness:
   //   * `canSeeAdmin` — strictly platform admin (`users.is_admin`).
@@ -18,6 +27,11 @@ export default (initialState: { currentUser?: Global.UserInfo }) => {
   //   * `canSeeOrgAdmin` — admin-style menus that work cross-org
   //     (Dashboard, Resources, Models, Cluster Management). Defaults
   //     to platform admin; extensions widen to include org admins.
+  //   * `canSeeGpuService` — GPU Service menu. Anyone allowed to
+  //     manage clusters (admins, Org owners) sees it; non-admins fall
+  //     through to "show only if a Kubernetes cluster is actually
+  //     reachable" so Org members without scheduling access don't see
+  //     a dead-end menu item.
   //   * `canManageCurrentOrg` — pages that only make sense inside a
   //     specific org context (member / group management). Defaults to
   //     `false`; extensions widen when both an org is selected AND
@@ -27,6 +41,7 @@ export default (initialState: { currentUser?: Global.UserInfo }) => {
   return applyAccessExtensions({
     canSeeAdmin: isPlatformAdmin,
     canSeeOrgAdmin: isPlatformAdmin,
+    canSeeGpuService: isPlatformAdmin || hasKubernetesCluster !== false,
     canManageCurrentOrg: false,
     canSeeUser,
     canDelete: true,

src/app.tsx

@@ -3,6 +3,8 @@ import { GPUStackVersionAtom, UpdateCheckAtom } from '@/atoms/user';
 import { setAtomStorage } from '@/atoms/utils';
 import { DEFAULT_ENTER_PAGE, GPUSTACK_API_BASE_URL } from '@/config/settings';
 import { COLOR_PRIMARY } from '@/config/theme/constants';
+import { queryClusterList } from '@/pages/cluster-management/apis';
+import { ProviderValueMap } from '@/pages/cluster-management/config';
 import { enterprisePluginReady } from '@/plugins/enterprise-ready';
 import { GPUStackPluginManager } from '@/plugins/manager';
 import { requestConfig } from '@/request-config';
@@ -34,11 +36,28 @@ const checkDefaultPage = async (userInfo: any) => {
   }
 };
 
+// Probes the caller's cluster list once so access predicates can gate
+// GPU Service (Kubernetes-only). Cheap (one list request) and never
+// blocks login — any failure just falls back to `undefined`, which
+// the predicate treats as "unknown / don't restrict beyond role".
+const probeHasKubernetesCluster = async (): Promise<boolean | undefined> => {
+  try {
+    const res = await queryClusterList({ page: -1 });
+    return (res?.items ?? []).some(
+      (c) => c?.provider === ProviderValueMap.Kubernetes
+    );
+  } catch (error) {
+    console.error('probeHasKubernetesCluster error', error);
+    return undefined;
+  }
+};
+
 // runtime configuration
 export async function getInitialState(): Promise<{
   fetchUserInfo: () => Promise<Global.UserInfo>;
   currentUser?: Global.UserInfo;
   pluginData?: Record<string, any>;
+  hasKubernetesCluster?: boolean;
 }> {
   const { location } = history;
 
@@ -122,12 +141,16 @@ export async function getInitialState(): Promise<{
   getAppVersionInfo();
 
   if (![DEFAULT_ENTER_PAGE.login].includes(location.pathname)) {
-    const userInfo = await fetchUserInfo();
+    const [userInfo, hasKubernetesCluster] = await Promise.all([
+      fetchUserInfo(),
+      probeHasKubernetesCluster()
+    ]);
     checkDefaultPage(userInfo);
     return {
       fetchUserInfo,
       currentUser: userInfo,
-      pluginData
+      pluginData,
+      hasKubernetesCluster
     };
   }
   return {

src/config/global.d.ts

@@ -92,6 +92,12 @@ declare namespace Global {
   interface InitialStateType {
     fetchUserInfo: () => Promise<UserInfo>;
     currentUser?: UserInfo;
+    // Captured at app boot so access predicates can gate GPU Service —
+    // the feature is Kubernetes-only, and Org members without a K8s
+    // cluster they can schedule on shouldn't see the menu. Refreshed
+    // by full page reload (e.g. OrgSwitcher) which re-runs
+    // getInitialState.
+    hasKubernetesCluster?: boolean;
   }
 
   type SearchParams = Pagination & { search?: string; [key: string]: any };

src/locales/en-US/no-result.ts

@@ -36,6 +36,8 @@ export default {
   'noresult.catalog.nofound': 'No matching models found.',
   'noresult.resources.cluster':
     'No clusters available. Add a cluster to get started.',
+  'noresult.resources.k8sCluster':
+    'No clusters available. Add a Kubernetes cluster to get started.',
   'noresult.resources.worker':
     'No workers available. Add a worker to get started.',
   'noresult.resources.gotocluster': 'Create Your First Cluster',

src/locales/ja-JP/no-result.ts

@@ -36,6 +36,8 @@ export default {
   'noresult.catalog.nofound': 'No matching models found.',
   'noresult.resources.cluster':
     'No clusters available. Add a cluster to get started.',
+  'noresult.resources.k8sCluster':
+    'No clusters available. Add a Kubernetes cluster to get started.',
   'noresult.resources.worker':
     'No workers available. Add a worker to get started.',
   'noresult.resources.gotocluster': 'Create Your First Cluster',

src/locales/ru-RU/no-result.ts

@@ -37,6 +37,8 @@ export default {
   'noresult.catalog.nofound': 'Подходящие модели не найдены.',
   'noresult.resources.cluster':
     'No clusters available. Add a cluster to get started.',
+  'noresult.resources.k8sCluster':
+    'No clusters available. Add a Kubernetes cluster to get started.',
   'noresult.resources.worker':
     'No workers available. Add a worker to get started.',
   'noresult.resources.gotocluster': 'Create Your First Cluster',
@@ -67,8 +69,7 @@ export default {
   'noresult.gpuservice.storage.subTitle': 'Хранилища ещё не добавлены.',
   'noresult.gpuservice.storage.nofound': 'Подходящие хранилища не найдены.',
   'noresult.gpuservice.storageType.title': 'Нет типов хранилищ',
-  'noresult.gpuservice.storageType.subTitle':
-    'Типы хранилищ ещё не добавлены.',
+  'noresult.gpuservice.storageType.subTitle': 'Типы хранилищ ещё не добавлены.',
   'noresult.gpuservice.storageType.nofound':
     'Подходящие типы хранилищ не найдены.',
   'noresult.gpuservice.sshkey.title': 'Нет открытых ключей SSH',

src/locales/tr-TR/no-result.ts

@@ -38,6 +38,8 @@ export default {
   'noresult.catalog.nofound': 'Eşleşen model bulunamadı.',
   'noresult.resources.cluster':
     'Kullanılabilir küme yok. Başlamak için bir küme ekleyin.',
+  'noresult.resources.k8sCluster':
+    'Kullanılabilir küme yok. Başlamak için bir Kubernetes kümesi ekleyin.',
   'noresult.resources.worker':
     'Kullanılabilir işçi düğüm yok. Başlamak için bir işçi düğüm ekleyin.',
   'noresult.resources.gotocluster': 'İlk Kümenizi Oluşturun',

src/locales/zh-CN/no-result.ts

@@ -35,6 +35,8 @@ export default {
   'noresult.catalog.subTitle': '尚未配置任何模型。',
   'noresult.catalog.nofound': '未找到匹配的模型',
   'noresult.resources.cluster': '暂无可用集群，请添加集群以开始使用。',
+  'noresult.resources.k8sCluster':
+    '暂无可用集群，请添加 Kubernetes 集群以开始使用。',
   'noresult.resources.worker': '暂无可用节点，请添加节点以开始使用。',
   'noresult.resources.gotocluster': '创建您的第一个集群',
   'noresult.resources.gotoworker': '添加节点',

src/pages/gpu-service/instances/index.tsx

@@ -1,13 +1,14 @@
 import { PageAction } from '@/config';
 import { PaginationKey, TABLE_SORT_DIRECTIONS } from '@/config/settings';
 import useTableFetch from '@/hooks/use-table-fetch';
+import { ProviderValueMap } from '@/pages/cluster-management/config';
 import { useQueryClusterList } from '@/pages/cluster-management/services/use-query-cluster-list';
 import { DeleteModal, FilterBar, IconFont, NoResult } from '@gpustack/core-ui';
-import { useIntl } from '@umijs/max';
+import { useAccess, useIntl, useNavigate } from '@umijs/max';
 import { useMemoizedFn } from 'ahooks';
 import { ConfigProvider, message, Modal, Table } from 'antd';
 import _ from 'lodash';
-import { useEffect } from 'react';
+import { useEffect, useMemo } from 'react';
 import PageBox from '../../_components/page-box';
 import {
   deleteGPUServiceInstance,
@@ -27,6 +28,8 @@ import useUpdateInstance from './services/use-update-instance';
 
 const GPUService: React.FC = () => {
   const intl = useIntl();
+  const navigate = useNavigate();
+  const access = useAccess();
   const [, modalContextHolder] = Modal.useModal();
 
   const {
@@ -71,13 +74,24 @@ const GPUService: React.FC = () => {
   const {
     fetchClusterList,
     cancelRequest: cancelClusterRequest,
-    clusterList
+    clusterList,
+    loading: clusterLoading
   } = useQueryClusterList();
 
   useEffect(() => {
     fetchClusterList({ page: -1 });
   }, []);
 
+  // GPU Service today is Kubernetes-only — Docker / cloud clusters
+  // can't host the CRDs. Filter so the page reflects scheduling
+  // reality even when the caller owns non-K8s clusters.
+  const k8sClusterList = useMemo(
+    () =>
+      clusterList.filter((c) => c.provider === ProviderValueMap.Kubernetes),
+    [clusterList]
+  );
+  const hasK8sCluster = k8sClusterList.length > 0;
+
   const handleModalOk = async (data: FormData) => {
     try {
       if (openInstanceModalStatus.realAction === PageAction.CREATE) {
@@ -129,6 +143,35 @@ const GPUService: React.FC = () => {
 
   const renderEmpty = (type?: string) => {
     if (type !== 'Table') return;
+    // No K8s cluster the caller can schedule on — replace the "no
+    // instances" empty state with a cluster-bootstrap prompt. The
+    // "Add cluster" CTA is reserved for callers who can actually
+    // create one (platform admin / Org owner); members see the
+    // explanation without a misleading button.
+    if (!clusterLoading && !hasK8sCluster) {
+      return (
+        <NoResult
+          loading={dataSource.loading || clusterLoading}
+          loadend={dataSource.loadend}
+          dataSource={[]}
+          image={<IconFont type="icon-cloud-outlined" />}
+          title={intl.formatMessage({
+            id: 'noresult.gpuservice.instance.title'
+          })}
+          subTitle={intl.formatMessage({
+            id: 'noresult.resources.k8sCluster'
+          })}
+          {...(access.canSeeOrgAdmin
+            ? {
+                buttonText: intl.formatMessage({
+                  id: 'noresult.resources.gotocluster'
+                }),
+                onClick: () => navigate('/cluster-management/clusters/list')
+              }
+            : {})}
+        />
+      );
+    }
     return (
       <NoResult
         loading={dataSource.loading}
@@ -168,7 +211,9 @@ const GPUService: React.FC = () => {
           handleInputChange={handleNameChange}
           rowSelection={rowSelection}
           widths={{ input: 300 }}
-          handleClickPrimary={openCreateInstanceModal}
+          handleClickPrimary={
+            hasK8sCluster ? openCreateInstanceModal : undefined
+          }
           buttonText={intl.formatMessage({ id: 'gpuservice.instance.add' })}
           handleDeleteByBatch={handleDeleteBatch}
         />