Skip to content

Guides

A ticket, start to finish

A customer reports: “Webhooks stopped retrying after the first failure.” Support uses Driver to explain the behavior, then gives engineering the evidence needed to reproduce it.

The ticket IDs, files, records, and outputs below are illustrative. They show the shape of a useful investigation, rather than a report from a real customer or repository.

Start with the support ticket

Jira and the relevant codebase must be connected and accessible to the investigation agent. In this example, the support ticket is SUP-2143. Engineering has mapped the customer’s release to the accessible payments codebase branch release/4.2.

With Automatic triage enabled, Driver investigates a new ticket and posts the automated report on Jira. Open Continue investigation in Driver and send a question to start your private continuation. You can also open Investigations and paste the support ticket URL to start directly.

Paste the full Jira URL in an actual request. Include the relevant log excerpt or delivery record with its source, environment, and timestamp, as support does here.

Review the finding and the next question

An illustrative result:

Finding or open questionSupporting sourceWhat the source establishes
The retry predicate excludes HTTP 429.payments, release/4.2, src/webhooks/retry.ts:18-24The predicate accepts HTTP 500 through 599; HTTP 429 returns false.
The reported delivery reached that condition.Supplied production delivery record, October 14 at 09:12 UTC, event evt_781The delivery received HTTP 429 and recorded retryable=false.
Existing tests leave HTTP 429 uncovered.tests/webhooks/retry.test.ts:31-46Tests cover HTTP 500 and 503, with no HTTP 429 case.
Should HTTP 429 honor the destination’s Retry-After header?No policy supplied yet.The owner must confirm the intended behavior before engineering changes it.

Open the cited predicate and check the branch. Support asks the service owner about the intended policy, then supplies the answer in the private conversation: “HTTP 429 should retry and honor Retry-After. Please include that requirement in the handoff.”

The code explains the decision, while the supplied record ties the reported delivery to it. The missing policy is now an explicit engineering requirement.

Prepare the customer reply

When Driver has finished responding and the ticket report is available, select Investigation report > Draft > Customer reply. Review the text and use Copy reply to put it in the customer conversation.

An illustrative reviewed reply:

Hand off to engineering

Select Investigation report > Draft > Engineering handoff, review the draft, and use Copy handoff. Support creates engineering issue ENG-482, links SUP-2143, and includes the new policy confirmation from the private follow-up.

An illustrative handoff excerpt:

The engineer starts a new investigation with the ENG-482 URL and SUP-2143 URL, or pastes this handoff. The original automated report remains the common starting snapshot; it does not contain support’s later private conversation.

Verify and report the fix

Engineering implements the change and references ENG-482 in the pull request. QA follows Verify a fix against the reported failure to check the HTTP 429 reproduction and the existing HTTP 500 behavior. Engineering records the tested revision and rollout state on ENG-482.

Support uses that Jira update to tell the customer what was verified and when the fix reaches their environment. Give feedback on the Driver investigation and close your investigation when your work is complete. Jira resolution follows your team’s ticket process.