Webhook delivery failure during high-volume repository events

Answer: Yes, a standardized retry policy applies to all webhook event types

Bitbucket Cloud retries any webhook delivery that receives a non‑2xx HTTP response or encounters a network timeout, regardless of the event type (push, pull request, etc.).

Confirmed facts

• Maximum of 5 delivery attempts per event.
• Backoff intervals follow exponential delay: 1 min, 2 min, 4 min, 8 min, 16 min.
• After the fifth unsuccessful attempt the event is dropped and logged as a failure.
• The policy is uniform across all event types; there is no separate retry logic for pushes vs. pull‑request events.

Likely explanation for high‑volume failures

During bursts of repository activity the webhook service may hit platform‑level rate limits or the endpoint may experience increased latency, causing more non‑2xx responses. Because the retry schedule is fixed, a persistently unhealthy endpoint will exhaust its five attempts and the event will be lost.

Steps to improve reliability for this case

1. Inspect the webhook logs in Bitbucket Cloud (Settings → Webhooks → Recent deliveries) to see the exact HTTP status codes and timestamps of failed attempts.
2. Verify that your endpoint returns a 2xx response within the platform’s timeout (default 10 seconds). If you see 504 or timeout errors, increase the endpoint’s capacity or reduce processing time.
3. If the logs show frequent 429 (Too Many Requests) or 503 responses, consider implementing a queue or rate‑limiting layer in front of your endpoint to smooth spikes.
4. Optionally, enable a secondary webhook URL as a fallback so that a temporary outage on the primary endpoint does not cause event loss.

To tailor the recommendation further, please confirm the predominant HTTP status code (e.g., 504 Gateway Timeout, 500 Internal Server Error, or 429 Too Many Requests) observed in the failed deliveries.