Retries and logs
Retry policy
Section titled “Retry policy”If your application responds with any code other than 2xx, or does not respond within the timeout, Rapid considers the attempt a failure and sends the webhook again.
There are 4 attempts in total: the initial one plus 3 more, each waiting longer than the previous one.
- Right away 1st attempt, as soon as the event is queued
- +5 min 2nd attempt, 5 minutes after the previous failure (5 minutes from the start)
- +15 min 3rd attempt (20 minutes from the start)
- +30 min 4th and last attempt (50 minutes from the start)
If the 4th also fails, the delivery is closed and the alert stays available in the dashboard for manual action (see Manual redelivery).
If the first attempt never goes out because of a failure on Rapid’s side, the alert is sent automatically later, usually within 20 minutes, with the same alert_id. This applies to alerts received in the last 2 days, while the webhook is active; alerts that arrived while the webhook was turned off are not sent later.
Timeout
Section titled “Timeout”Each attempt has a 30-second timeout. If your application does not respond within that time, the attempt is considered a failure.
What counts as success
Section titled “What counts as success”Any 2xx code: 200, 201, 202, 204 and the rest of the range. The response body is optional.
Any other code (including 4xx and 5xx) triggers a retry.
Redirects
Section titled “Redirects”Register the final URL. Rapid handles redirects like this:
| Your URL’s response | What happens |
|---|---|
301, 302, 303 | Failure. These codes turn the POST into a GET and drop the body, so following them would deliver nothing. The attempt goes into the retry flow and the log shows where your URL redirects to. |
307, 308 to the same domain | Followed, with the POST, body and headers intact. Up to 3 redirects in a row. |
307, 308 to another domain | Failure. The signed payload does not leave the domain you registered. |
The common cases are the server appending / to the end of the path and switching domains, such as from example.com to www.example.com (which counts as another domain). In the delivery log, the reason starts with [redirecionamento], along with the destination address.
Address that is not public
Section titled “Address that is not public”If the registered address does not resolve to a public internet address (it points to a local or private network, or the domain does not exist), the delivery fails without any request and goes into the normal retry flow. In the delivery log, the reason starts with [destino]. Fix the URL or the domain’s DNS; see Use a public internet address.
If the reason starts with [https], the registered URL does not use HTTPS and Rapid does not deliver in plain text: register an https:// URL.
If the reason starts with [assinatura], the failure is on Rapid’s side: the delivery does not go out unsigned, and it is retried automatically. There is nothing to do in your system.
Blocked companies
Section titled “Blocked companies”If the company is blocked for non-payment at Rapid, the webhook flow keeps working normally: alerts are still received from the providers, stored and delivered to the configured URL. The dashboard remains available. What the block cuts off is the credential-based API (read and write: GET/PATCH /chargeback-alert/alerts, transactions, merchants, legacy aliases), which starts responding 403 COMPANY_BLOCKED, and the defense of disputes, including those already in progress.
To stop receiving new alerts, the enrollment has to be deactivated directly at the provider (an action carried out by the Rapid team).
Delivery logs
Section titled “Delivery logs”Every attempt is recorded internally with:
- Event delivered (
chargeback_alert.received,dispute.fraud.revoke_access,webhook.test) - Date and time of the attempt
- Destination URL
- HTTP response code
- Response body (truncated at 1000 characters; a response larger than 64 KB is not stored)
- Attempt number
You see these records in the Rapid dashboard in two places: in Settings › Webhook, under Delivery history, which filters by result (only failures, for example) and by period; and in Activity, Webhooks tab, which also filters by alert and is available to the Owner, Admin, Developer and Auditor roles. When the failure never reached your server, the dashboard shows an explanation of the reason; the original text, which starts with [https], [destino], [redirecionamento] or [assinatura], appears on hover.
Manual redelivery
Section titled “Manual redelivery”When every automatic attempt fails, the alert stays in the dashboard. To send it again, use the Retry button in the Delivery history, under Settings › Webhook. The redelivered event arrives as a new webhook with the same alert_id, which is why idempotency is essential.
The button does not appear on a failed attempt when the same alert was delivered in a later attempt. If the redelivery cannot go out at that moment, the dashboard tells you, and you can try again shortly.
Resend the failures of a period
Section titled “Resend the failures of a period”After an outage of your endpoint, use Resend failures in the Delivery history, with the webhook turned on. Choose the period and Rapid sends again, at once, the alerts whose delivery failed in that period and that have not reached your system yet (an alert that was delivered at least once is left out; for it, use the row’s Retry).
- Up to 500 alerts per request, oldest first. If the period has more, the dashboard tells you how many are left: request again in a few minutes.
- An alert that still has an automatic attempt ahead is not sent twice, even if you request twice in a row.
- Only the webhook goes out: the alert email is not sent again.
To check from your system what was left unanswered, query the alerts pending in the period (status=pending with date_from and date_to).
Redelivery from the dashboard applies to alerts. The dispute.fraud.revoke_access event has no manual redelivery.
Test event
Section titled “Test event”The Send test event button, under Settings › Webhook, sends your URL a webhook.test with the same headers and the same signature as real deliveries. It is sent once, with no retry, and the dashboard shows right away the code your URL responded with. The attempt stays in the Delivery history like any other, without Retry, and it does not count toward the alert delivery failures. You can send up to 5 tests per minute.