Attempt

An attempt is one HTTP request within a delivery's retry chain. Each attempt records what was sent, what was received, and how long it took. Multiple attempts roll up to a single delivery.

If a delivery is "everything that happened trying to reach destination X for event Y," an attempt is "this specific request at this specific time."

What every attempt records

For each attempt, Hooksbase persists:

• Timestamp — when the attempt started
• Resolved target — the destination snapshot selected when the delivery was created
• Request headers — what was sent, including per-attempt signing headers and persisted custom headers
• Response status — 2xx, 4xx, 5xx, timeout
• Response body — truncated to a configurable limit
• Latency — request duration in milliseconds
• Error — if the attempt failed at the network layer (DNS, TLS, connection refused)

This per-attempt detail is what makes debugging tractable. When a delivery fails after eight attempts, you can see which attempts were 503 (transient), which were 422 (your handler rejected), and which timed out (downstream slowness).

Why distinguishing attempt from delivery matters

A common reporting mistake is counting attempts when you should be counting deliveries (or vice versa). "Our success rate is 99%" — at the attempt level or the delivery level? Eight failed attempts followed by a successful one is a successful delivery (one), not nine events (eight failed attempts + one success).

Hooksbase reports both — attempt-level for debugging, delivery-level for SLOs and billing. For the observability framework: Observability for webhook delivery.