> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.pivotal.app/llms.txt.
> For full documentation content, see https://docs.pivotal.app/llms-full.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.pivotal.app/_mcp/server.

# Retries and backoff

Pivotal retries failed deliveries for 24 hours with exponential backoff. After the final retry, the endpoint is marked unhealthy and we email the workspace owner. The full delivery log lives in the dashboard with one-click replay.

WHAT COUNTS AS SUCCESS

* Any 2xx response inside the 10-second window.

WHAT COUNTS AS FAILURE

* Non-2xx response (4xx and 5xx alike).
* Connection refused, DNS failure, TLS handshake failure.
* Timeout: no response inside 10 seconds.

RETRY SCHEDULE

| Attempt | Delay after previous                |
| ------- | ----------------------------------- |
| 1       | immediate                           |
| 2       | 5 seconds                           |
| 3       | 5 minutes                           |
| 4       | 30 minutes                          |
| 5       | 2 hours                             |
| 6       | 5 hours                             |
| 7       | 10 hours                            |
| 8       | 24 hours from first attempt — final |

WHEN DELIVERY GIVES UP

After attempt 8, Pivotal:

1. Marks the delivery as `failed` in the dashboard.
2. Increments the endpoint's failure counter.
3. Disables the endpoint if it has crossed 100 consecutive failures.
4. Emails the workspace owner with a link to the delivery log.

You can re-fire any past delivery from the dashboard — payload and headers are identical to the original. The Svix SDK treats replays as new deliveries with the same `event.id`, so your idempotent handler keeps you safe.