How is Promise.allSettled different from Promise.all, and when would you use it?
`Promise.all` rejects as soon as any input rejects — short-circuits. `Promise.allSettled` always resolves once every input has settled, with an array of `{status, value|reason}` objects. Use `allSettled` for independent best-effort work (dashboard cards, batch operations, analytics) where partial success is meaningful.
Difference at a glance
Promise.all([ok, fail, ok]);
// Rejects with fail's reason as soon as fail rejects.
// Other ok promises still run but their results are lost.
Promise.allSettled([ok, fail, ok]);
// Always resolves once all settle.
// Returns [{status:"fulfilled", value}, {status:"rejected", reason}, {status:"fulfilled", value}]When to use allSettled
- Dashboard with N independent cards — one failure shouldn't blank the others.
- Batch save — return per-item success/failure to show the user partial results.
- Cache warmup — best-effort, don't fail if a few items don't load.
- Analytics fan-out — fire to several backends, log failures, don't propagate.
- Multi-source fetch (CDN mirrors) — record which succeeded.
When to stay with all
- Atomic compound operations — if any leg fails, the whole thing is invalid.
- Page load that needs all data to render correctly — fail-fast is right.
Use shape
const results = await Promise.allSettled(cardLoaders);
const cards = results.map((r, i) =>
r.status === "fulfilled"
? { ok: true, data: r.value }
: { ok: false, error: r.reason, retryFn: cardLoaders[i] },
);Render each card with its own ok/error UI; show retry where it failed.
Related primitives
| Method | Resolves when | Rejects when |
|---|---|---|
Promise.all | All fulfill | Any rejects |
Promise.allSettled | All settle | Never |
Promise.race | First settles (fulfills or rejects) | First rejects |
Promise.any | First fulfills | All reject (with AggregateError) |
Pre-allSettled workaround
Before allSettled was standard:
Promise.all(promises.map((p) => p.then(
(v) => ({ status: "fulfilled", value: v }),
(r) => ({ status: "rejected", reason: r }),
)));Same shape; allSettled is just the standard form.
Interview framing
"Promise.all short-circuits on the first rejection — fine when failure means the whole operation is invalid. Promise.allSettled always resolves with per-promise outcomes — what you want for independent best-effort work like a dashboard with several cards, a batch operation reporting partial success, or analytics fan-out. The shape is { status, value | reason } so you can render success vs failure per item. The pre-2020 polyfill was .map wrapping each promise's outcome."
Follow-up questions
- •What does Promise.any do?
- •When would you use Promise.race?
- •How would you implement allSettled in pre-ES2020 code?
Common mistakes
- •Using Promise.all and losing partial results.
- •Forgetting to inspect status before reading value/reason.
Performance considerations
- •No perf difference. Choice is about correctness / error semantics.
Edge cases
- •Empty input — both resolve immediately ([] result).
- •Thenables (not real Promises) — still work via Promise.resolve coercion.
Real-world examples
- •Dashboard widget loading, batch upload UIs, multi-API aggregation.