> ## Documentation Index
> Fetch the complete documentation index at: https://docs.letmepost.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# platform_unavailable

> Upstream is down, throttled, or returning 5xx. Transient — retry later.

The upstream platform is unreachable, throttled, or returning 5xx. Treat as transient. The post is **not** retried automatically by `POST /v1/posts` — that's the caller's job, with idempotency.

For scheduled posts the worker retries internally per the queue policy.

## What triggers it

* The platform is down (e.g., Meta Graph 5xx outage).
* The platform throttled us at the network layer (rare; per-app rate limit usually surfaces as `platform_rejected`).
* DNS / TLS error reaching the platform.
* The platform took longer than our request timeout to respond.

## Response shape

```json platform_unavailable.json theme={"system"}
{
  "error": {
    "code": "platform_unavailable",
    "message": "Bluesky API returned 503 Service Unavailable.",
    "platform": "bluesky",
    "platformResponse": "<raw upstream body or null>",
    "remediation": "Retry with the same Idempotency-Key after a short backoff.",
    "docUrl": "https://docs.letmepost.dev/errors/platform_unavailable",
    "requestId": "req_..."
  }
}
```

## How to handle

Retry with **the same `Idempotency-Key`**. Our replay cache holds for 24 hours; sending the same request again within that window is safe — if the upstream succeeded the first time despite the 5xx, you'll get the cached `201` back.

Standard exponential backoff is fine: 1s, 2s, 4s, 8s, …, up to your tolerance.

## Related

* [Idempotency](/idempotency) — why retries are safe.
* [`rate_limited`](/errors/rate_limited) — letmepost's own rate limit.
* [`platform_rejected`](/errors/platform_rejected) — upstream said no for content reasons.