[bug]: Instance wrapper shows "didn't start up correctly" on transient /api/instances/ failure

Is there an existing issue for this?

• I have searched the existing issues

Current behavior

When the API is briefly unavailable (restart, 502 Bad Gateway, short network blip), a request to GET /api/instances/ fails.

The web app then shows the full-page maintenance empty state:

"Looks like Plane didn't start up correctly!
Some services might have failed to start..."

Even if instance data was already loaded successfully earlier in the session.

Expected:
If instance info is already in the client store, keep the app usable. Only show the maintenance screen when there is no instance data (true cold start / API completely down and never loaded).

Actual:
In apps/web/core/lib/wrappers/instance-wrapper.tsx, any SWR error on INSTANCE_INFORMATION immediately renders <MaintenanceView />, regardless of whether instance is already available.

Suggested fix:
Change:
if (instanceSWRError) return <MaintenanceView />;
to:
if (instanceSWRError && !instance) return <MaintenanceView />;

Steps to reproduce

1. Start local Plane stack (API + web).
2. Open the web app and confirm it loads normally (instance info fetched successfully).
3. Restart the API container / briefly stop API so /api/instances/ returns 502 or network error for a few seconds.
4. Keep the browser tab open (or trigger a re-fetch / reload of instance info via SWR).
5. Observe the UI switches to the maintenance page ("Plane didn't start up correctly!") instead of staying on the already-loaded app.

Optional logs:

• Network tab: GET /api/instances/ → 502 / failed
• After API is healthy again, refresh still may be needed depending on client state

Environment

Deploy preview

Browser

Google Chrome

Variant

Local

Version

1.4.2