mirror of
https://github.com/supabase/supabase.git
synced 2026-10-12 04:45:08 +03:00
Ali Waseem
6fe0ad442b
## Summary Fixes [FE-3095](https://linear.app/supabase/issue/FE-3095/compute-size-hover-card-shows-badge-and-cpumemory-from-out-of-sync). The compute size hover card on the project home dashboard was sourcing its badge and its CPU/memory rows from two different cached responses, which can disagree: | Field shown | Previous source | |---|---| | Badge ("XLARGE") | `project.infra_compute_size` (project-detail query) | | Cores / memory | `selected_addons[compute_instance].variant.meta` (project-addons query) | A customer reported seeing an **XLARGE** badge next to **2-core ARM (Shared) / 1 GB** — the micro-tier specs — and asked whether their upgrade had actually been applied. The upgrade was applied; only the rendered card was contradictory. ## Fix Source both the badge and the CPU/memory rows from the same logical fact: look up the variant in `available_addons` whose identifier matches `ci_${infra_compute_size}` and read its `meta`. `available_addons` is essentially a static catalog of variant specs, so once it's loaded the card cannot show specs that disagree with the badge. This also collapses the special-cased `INSTANCE_MICRO_SPECS` fallback into the existing `getAvailableComputeOptions` helper (which already provides micro/nano fallbacks). The nano UX text ("Shared / Up to 0.5 GB") is preserved by switching that JSX branch to key on `computeSize === 'nano'`. ## Out of scope - `useProjectAddonUpdateMutation` does not invalidate `projectKeys.detail`. That's hygiene worth doing later, but project-detail has a 30s `staleTime` and the resize already drives 5s polling via the `RESIZING` status path, so the badge refreshes naturally and this fix doesn't depend on it. ## Test plan - [ ] Hover the compute badge on a project at each compute size (nano, micro, small, ..., 16xlarge) and confirm CPU and memory rows match the badge. - [ ] Resize a project from micro → large; on completion, confirm the hover card shows large specs (no transient micro values). - [ ] Open the dashboard for a free-tier project on micro that has no `compute_instance` entry in `selected_addons` and confirm the card still shows micro specs (i.e. `getAvailableComputeOptions` micro fallback is engaged). - [ ] Confirm the "Unlock more compute" CTA still appears for non-highest sizes and disappears at the highest size.