feat(kiloclaw): support Northflank tier resize

apps/web/src/app/admin/components/KiloclawInstances/KiloclawInstanceDetail.tsx

@@ -1421,6 +1421,7 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
 
   const provider = data?.workerStatus?.provider ?? null;
   const isFlyProvider = provider === 'fly';
+  const isNorthflankProvider = provider === 'northflank';
   const runtimeId = data?.workerStatus?.runtimeId ?? null;
   const storageId = data?.workerStatus?.storageId ?? null;
   const flyMachineId = data?.workerStatus?.flyMachineId ?? null;
@@ -1838,7 +1839,10 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
 
   // Poll status during resize phases
   const resizePolling =
-    resizePhase === 'stopping' || resizePhase === 'starting' || resizePhase === 'waiting';
+    resizePhase === 'stopping' ||
+    resizePhase === 'starting' ||
+    resizePhase === 'waiting' ||
+    (isNorthflankProvider && resizePhase === 'resizing');
   useQuery({
     queryKey: ['machine-resize-poll', userId, instanceId, resizePolling],
     queryFn: async () => {
@@ -1865,6 +1869,19 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
     if (!data || !userId) return;
 
     try {
+      if (isNorthflankProvider) {
+        setResizePhase('resizing');
+        await resizeMachineMutation({
+          userId,
+          instanceId: data.id,
+          instanceType: selectedInstanceType,
+        });
+        invalidateMachineQueries();
+        setResizePhase('done');
+        toast.success('Northflank resize completed');
+        return;
+      }
+
       // Step 1: Stop if running — retry up to 3 times since Fly can be slow
       if (currentStatus !== 'stopped') {
         setResizePhase('stopping');
@@ -3295,7 +3312,10 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
                 <div className="space-y-1">
                   <p className="text-sm font-medium">
                     {resizePhase === 'stopping' && 'Stopping machine...'}
-                    {resizePhase === 'resizing' && 'Updating machine size...'}
+                    {resizePhase === 'resizing' &&
+                      (isNorthflankProvider
+                        ? 'Resizing Northflank deployment...'
+                        : 'Updating machine size...')}
                     {resizePhase === 'starting' && 'Starting machine with new size...'}
                     {resizePhase === 'waiting' && 'Waiting for machine to be ready...'}
                   </p>
@@ -3342,9 +3362,11 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
               <div className="flex items-center gap-3 rounded border border-green-600/30 bg-green-600/5 p-4">
                 <CheckCircle2 className="h-5 w-5 shrink-0 text-green-600" />
                 <div>
-                  <p className="text-sm font-medium text-green-600">Machine resize complete</p>
+                  <p className="text-sm font-medium text-green-600">Runtime resize complete</p>
                   <p className="text-muted-foreground text-xs">
-                    Machine is running with the new size.
+                    {isNorthflankProvider
+                      ? 'Northflank completed the deployment rollout.'
+                      : 'Machine is running with the new size.'}
                   </p>
                 </div>
                 <Button
@@ -3398,11 +3420,12 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
             <DialogHeader>
               <DialogTitle className="flex items-center gap-2 text-orange-500">
                 <AlertTriangle className="h-5 w-5" />
-                Resize Machine
+                Resize runtime
               </DialogTitle>
               <DialogDescription className="pt-3">
-                This will stop the machine, update its CPU/memory and storage spec, and restart it.
-                The user will be disconnected during the restart.
+                {isNorthflankProvider
+                  ? 'Northflank will resize this instance by rolling the deployment onto the target compute plan. The instance may restart during the rollout.'
+                  : 'This will stop the machine, update its CPU/memory and storage spec, and restart it. The user will be disconnected during the restart.'}
                 <span className="text-foreground mt-2 block font-medium">
                   User: {data?.user_email ?? data?.user_id}
                 </span>
@@ -3444,6 +3467,14 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
                   })}
                 </select>
               </div>
+              {isNorthflankProvider && (
+                <Alert className="border-muted-foreground/30 bg-muted/30">
+                  <AlertDescription className="text-muted-foreground">
+                    Northflank applies the compute change through a deployment rollout. The worker
+                    waits for Northflank to report completion before saving the new tier.
+                  </AlertDescription>
+                </Alert>
+              )}
               {data?.workerStatus?.provider === 'fly' &&
                 getTier(selectedInstanceType).volumeSizeGb >
                   (data?.workerStatus?.volumeSizeGb ?? 10) && (
@@ -3456,6 +3487,18 @@ export function KiloclawInstanceDetail({ instanceId }: { instanceId: string }) {
                     </AlertDescription>
                   </Alert>
                 )}
+              {isNorthflankProvider &&
+                getTier(selectedInstanceType).volumeSizeGb >
+                  (data?.workerStatus?.volumeSizeGb ?? 10) && (
+                  <Alert className="border-orange-500/30 bg-orange-500/10">
+                    <AlertTriangle className="h-4 w-4 text-orange-500" />
+                    <AlertDescription className="text-orange-700 dark:text-orange-300">
+                      Northflank volume will grow from {data?.workerStatus?.volumeSizeGb ?? 10} GB
+                      to {getTier(selectedInstanceType).volumeSizeGb} GB. Volumes can grow but
+                      cannot be shrunk.
+                    </AlertDescription>
+                  </Alert>
+                )}
               {data?.workerStatus?.provider === 'docker-local' &&
                 getTier(selectedInstanceType).volumeSizeGb !==
                   (data?.workerStatus?.volumeSizeGb ?? 10) && (

services/kiloclaw/src/durable-objects/kiloclaw-instance.test.ts

@@ -9225,21 +9225,84 @@ describe('resizeMachine', () => {
     expect(storage._store.get('instanceType')).toBe('perf-4-8');
   });
 
-  it('rejects Northflank resize before persisting tier changes', async () => {
+  it('persists Northflank resize after volume and deployment plan updates are accepted', async () => {
     const { instance, storage } = createInstance();
-    await seedProvisioned(storage, {
+    await seedNorthflankInstance(storage, {
       provider: 'northflank',
-      providerState: { provider: 'northflank' },
-      status: 'stopped',
+      providerState: northflankProviderState(),
+      status: 'running',
+      instanceType: 'perf-1-3',
+      machineSize: { cpus: 1, memory_mb: 3072, cpu_kind: 'performance' },
+      volumeSizeGb: 10,
+    });
+    vi.mocked(fetch).mockImplementation(async input => {
+      const url = fetchInputUrl(input);
+      if (url.endsWith('/volumes/volume-1')) {
+        return new Response(JSON.stringify({ data: {} }), {
+          status: 200,
+          headers: { 'content-type': 'application/json' },
+        });
+      }
+      if (url.endsWith('/services/deployment/service-1')) {
+        return new Response(JSON.stringify({ data: { id: 'service-1', name: 'kc-ki-test' } }), {
+          status: 200,
+          headers: { 'content-type': 'application/json' },
+        });
+      }
+      throw new Error(`Unhandled Northflank API request: ${url}`);
+    });
+
+    const result = await instance.resizeMachine('perf-4-8');
+
+    expect(result.newTier).toBe('perf-4-8');
+    expect(storage._store.get('instanceType')).toBe('perf-4-8');
+    expect(storage._store.get('machineSize')).toEqual({
+      cpus: 4,
+      memory_mb: 8192,
+      cpu_kind: 'performance',
+    });
+    expect(storage._store.get('volumeSizeGb')).toBe(20);
+    expect(storage._store.get('providerState')).toEqual(
+      expect.objectContaining({ ingressHost: 'kc-ki-test.code.run' })
+    );
+  });
+
+  it('leaves Northflank tier state unchanged when provider resize fails', async () => {
+    const { instance, storage } = createInstance();
+    await seedNorthflankInstance(storage, {
+      status: 'running',
       instanceType: 'perf-1-3',
       machineSize: { cpus: 1, memory_mb: 3072, cpu_kind: 'performance' },
       volumeSizeGb: 10,
     });
+    vi.mocked(fetch).mockImplementation(async input => {
+      const url = fetchInputUrl(input);
+      if (url.endsWith('/volumes/volume-1')) {
+        return new Response(JSON.stringify({ data: {} }), {
+          status: 200,
+          headers: { 'content-type': 'application/json' },
+        });
+      }
+      if (url.endsWith('/services/deployment/service-1')) {
+        return new Response(JSON.stringify({ error: 'deployment patch failed' }), {
+          status: 500,
+          headers: { 'content-type': 'application/json' },
+        });
+      }
+      throw new Error(`Unhandled Northflank API request: ${url}`);
+    });
 
     await expect(instance.resizeMachine('perf-4-8')).rejects.toThrow(
-      'Instance tier resize is not yet supported on Northflank instances'
+      'Northflank API patchDeploymentService failed (500)'
     );
+
     expect(storage._store.get('instanceType')).toBe('perf-1-3');
+    expect(storage._store.get('machineSize')).toEqual({
+      cpus: 1,
+      memory_mb: 3072,
+      cpu_kind: 'performance',
+    });
+    expect(storage._store.get('volumeSizeGb')).toBe(10);
   });
 
   it('clears any active admin size override and reports it in the response', async () => {

services/kiloclaw/src/durable-objects/kiloclaw-instance/index.ts

@@ -2962,9 +2962,11 @@ export class KiloClawInstance extends DurableObject<KiloClawEnv> {
       cannotPrefix: 'resize',
       beforePhrase: 'resizing machine tier',
       notSupportedSubject: 'Instance tier resize',
-      // Tier resize calls fly.extendVolume on storage growth; Fly volume
-      // extends require the machine to be stopped.
-      requireStopped: true,
+      // Fly tier resize calls fly.extendVolume on storage growth, which
+      // requires the machine to be stopped. Northflank uses deployment
+      // rollout semantics and does not require a stopped instance.
+      requireStopped: this.s.provider !== 'northflank',
+      allowNorthflank: true,
     });
 
     const targetTier = getTier(targetTierKey);
@@ -3004,6 +3006,18 @@ export class KiloClawInstance extends DurableObject<KiloClawEnv> {
       await this.persist({ volumeSizeGb: targetTier.volumeSizeGb });
     }
 
+    if (this.s.provider === 'northflank') {
+      const result = await this.provider().resizeRuntime?.({
+        env: this.env,
+        state: this.s,
+        targetTier: targetTier.key,
+      });
+      if (!result) {
+        throw new Error('Provider northflank does not support tier resize');
+      }
+      await this.persistProviderResult(result);
+    }
+
     // Capture and clear any active admin override before applying the tier
     // change. The customer is now paying for the new tier; carrying a
     // pre-existing override would either silently downgrade them (override
@@ -3091,6 +3105,7 @@ export class KiloClawInstance extends DurableObject<KiloClawEnv> {
     beforePhrase?: string;
     notSupportedSubject: string;
     requireStopped: boolean;
+    allowNorthflank?: boolean;
   }): void {
     if (!this.s.userId) {
       throw new Error('Instance is not provisioned');
@@ -3115,7 +3130,7 @@ export class KiloClawInstance extends DurableObject<KiloClawEnv> {
         { status: 409 }
       );
     }
-    if (this.s.provider === 'northflank') {
+    if (this.s.provider === 'northflank' && args.allowNorthflank !== true) {
       throw new Error(`${args.notSupportedSubject} is not yet supported on Northflank instances`);
     }
   }

services/kiloclaw/src/northflank/client.test.ts

@@ -13,6 +13,7 @@ import {
   isNorthflankNotFound,
   listServices,
   putProjectSecret,
+  updateVolume,
 } from './client';
 import { getNorthflankConfig } from './config';
 import type { NorthflankClientConfig } from './client';
@@ -154,6 +155,37 @@ describe('Northflank Worker fetch client', () => {
     expect(url).toBe('https://api.northflank.com/v1/projects/project-1/volumes/kc-ki-test');
   });
 
+  it('updates volume storage size through the documented empty response endpoint', async () => {
+    const fetchMock = mockFetchSequence([[200, { data: {} }]]);
+
+    await expect(
+      updateVolume(config, 'project-1', 'volume-1', { storageSizeMb: 20480 })
+    ).resolves.toBeUndefined();
+
+    const [url, init] = firstFetchCall(fetchMock);
+    expect(url).toBe('https://api.northflank.com/v1/projects/project-1/volumes/volume-1');
+    const requestInit = expectRequestInit(init);
+    expect(requestInit.method).toBe('POST');
+    expect(JSON.parse(expectStringBody(requestInit.body))).toEqual({
+      spec: {
+        storageSize: 20480,
+      },
+    });
+  });
+
+  it('updates volumes through team-scoped routes when a team ID is configured', async () => {
+    const fetchMock = mockFetchSequence([[200, { data: {} }]]);
+
+    await updateVolume({ ...config, teamId: 'team-1' }, 'project-1', 'volume-1', {
+      storageSizeMb: 40960,
+    });
+
+    const [url] = firstFetchCall(fetchMock);
+    expect(url).toBe(
+      'https://api.northflank.com/v1/teams/team-1/projects/project-1/volumes/volume-1'
+    );
+  });
+
   it('finds services by deterministic name with a direct GET', async () => {
     const fetchMock = mockFetchSequence([
       [200, { data: { id: 'kc-ki-test', name: 'kc-ki-test' } }],

services/kiloclaw/src/northflank/client.ts

@@ -55,6 +55,7 @@ const NorthflankSecretDetailsSchema = z.object({ id: z.string(), name: z.string(
 const ProjectResponseSchema = z.object({ data: NorthflankProjectSchema });
 const VolumeResponseSchema = z.object({ data: NorthflankVolumeSchema });
 const VolumeListResponseSchema = z.object({ data: z.array(NorthflankVolumeSchema) });
+const EmptyDataResponseSchema = z.object({ data: z.object({}).passthrough() });
 const ServiceResponseSchema = z.object({ data: NorthflankServiceSchema });
 const ServiceListResponseSchema = z.object({
   data: z.object({ services: z.array(NorthflankServiceSchema) }).passthrough(),
@@ -419,6 +420,26 @@ export async function getVolume(
   return response.data;
 }
 
+export async function updateVolume(
+  config: NorthflankClientConfig,
+  projectId: string,
+  volumeId: string,
+  input: { storageSizeMb: number }
+): Promise<void> {
+  await requestJson(
+    config,
+    `/projects/${encodeURIComponent(projectId)}/volumes/${encodeURIComponent(volumeId)}`,
+    jsonInit('POST', {
+      spec: {
+        storageSize: input.storageSizeMb,
+      },
+    }),
+    EmptyDataResponseSchema,
+    [200],
+    'updateVolume'
+  );
+}
+
 export async function deleteVolume(
   config: NorthflankClientConfig,
   projectId: string,

services/kiloclaw/src/northflank/config.test.ts

@@ -19,8 +19,8 @@ describe('getNorthflankConfig', () => {
       deploymentPlan: 'nf-compute-200',
       deploymentPlans: {
         'perf-1-3': 'nf-compute-200',
-        'perf-4-8': 'nf-compute-200',
-        'perf-4-16': 'nf-compute-200',
+        'perf-4-8': 'nf-compute-400',
+        'perf-4-16': 'nf-compute-400-16',
       },
       storageClassName: 'nf-multi-rw',
       storageAccessMode: 'ReadWriteMany',
@@ -54,8 +54,8 @@ describe('getNorthflankConfig', () => {
       deploymentPlan: 'nf-compute-200',
       deploymentPlans: {
         'perf-1-3': 'nf-compute-200',
-        'perf-4-8': 'nf-compute-200',
-        'perf-4-16': 'nf-compute-200',
+        'perf-4-8': 'nf-compute-400',
+        'perf-4-16': 'nf-compute-400-16',
       },
       storageClassName: 'nf-ssd-rwo',
       storageAccessMode: 'ReadWriteOnce',
@@ -79,6 +79,21 @@ describe('getNorthflankConfig', () => {
       'NF_VOLUME_SIZE_MB must be a positive integer'
     );
   });
+
+  it('accepts explicit per-tier deployment plan overrides', () => {
+    expect(
+      getNorthflankConfig({
+        ...baseEnv,
+        NF_DEPLOYMENT_PLAN_PERF_1_3: 'custom-1-3',
+        NF_DEPLOYMENT_PLAN_PERF_4_8: 'custom-4-8',
+        NF_DEPLOYMENT_PLAN_PERF_4_16: 'custom-4-16',
+      } as never).deploymentPlans
+    ).toEqual({
+      'perf-1-3': 'custom-1-3',
+      'perf-4-8': 'custom-4-8',
+      'perf-4-16': 'custom-4-16',
+    });
+  });
 });
 
 describe('northflankClientConfig', () => {