Fix console rough edges for cloud Windows
Found while running Task 10344's live oracles on dev (operator login, GMind org).
1. Add node → Gmind's cloud: the OS picker offers Windows Server even when the org's plan lacks Windows nodes; the refusal ("Windows nodes are not on this organization's plan.") appears only after Create, at step 4. Disable or annotate the option on step 2 instead.
2. Add node progress says "About 2:56 left" for a Windows launch; m8i.large took 8 min and m7i.metal-24xl took 30 min to come online. Base the estimate on OS and metal.
3. A healthy Windows boot on m7i.metal-24xl showed "Stalled" at 15 min, then came online at 30 min with no action. Give metal (and Windows) a longer stall bound, or say "slow boot" rather than "Stalled".
4. Products → Instance types → Browse says "No operator cloud connection — add one in Machines > Connections" while the sidebar org filter is set to a tenant; clearing the filter shows the connections. Ignore the filter on this operator page, or name it as the cause.
5. Customers → Overrides → Features: ticking a switch looks done, but nothing saves until Save → Apply; navigating away hits a bare browser "leave page?" prompt. Show the unsaved state next to the switch.
Out of scope: MicroVM runtime, Windows MicroVM support (settled by Task 10344), any API behaviour beyond what the items name.
Done means:
- Each item fixed or explicitly declined with a reason in the PR body.
- A screenshot per fixed item on dev at desktop width.
- `./dc check quick` green; PR merged through the merge lane; Jaah Deploy green.
Outcome
Fixed all 5 cloud-Windows rough edges: off-plan OS disabled on step 2, OS/metal boot ETA, longer stall bound, Instance types ignore org filter, Unsaved chip on Overrides switches. Quick gate green; PR #7056 merged (000bf39); its deploy superseded by a newer one. Visual check done on slot stack, all 5 render well. Memory: used mem-25288.
Artifacts
Activity
- Created
- Run
- Runner started
- Runner ended (done)