Retry a failed Make run without blindly creating duplicates
Inspect both the failed execution and the destination before retrying.
A timeout does not prove the destination rejected the write. Make retries incomplete executions with their original module settings, so distinguish temporary failures from configuration errors.
Work through the setup
- Read the failed module. Open the scenario Incomplete executions tab, choose Details and inspect the module warning. Record the error, business record ID and last confirmed successful action.
- Inspect the destination. Search for that same business ID where the scenario writes data. If the record already exists, decide whether the recovery should update it instead of creating it again.
- Choose the right recovery. Make describes retries for temporary connection/rate-limit failures and manual resolution when the blueprint or module settings need changes. Do not assume editing the current scenario changes a stored retry.
- Check the result once. After the reviewed recovery, confirm exactly one intended destination record and inspect downstream actions. Keep a note linking the source ID to the destination ID for another future retry.
A worked example
Order A-104 was written to a sheet, but the next module timed out. Restarting the whole scenario from the original order may append a second row. Looking up A-104 first helps you distinguish “missing write” from “write happened, response lost”. This is a design example, not a guarantee of exactly-once execution.
What to watch out for
- A timeout is an unknown outcome, not proof nothing happened.
- Skipping an error can discard work while marking a run successful.
- Retries consume resources and can repeat external side effects.
Identify the completed side effect
Before replaying a run, inspect whether the destination already contains the order, payment reference or row the scenario was meant to create. A source timeout does not prove the destination rejected the request. Keep the source event identifier in your reconciliation notes. Where the app supports a documented idempotency mechanism, use that mechanism rather than a homemade name match. A successful retry counter is not a substitute for checking the destination once.
Sources and evidence
- Manage incomplete executions - Help Center (read November 7, 2026)
- Scenario settings - Help Center (read November 7, 2026)
- Fix errors and warnings - Help Center (read November 7, 2026)
Common checks
Will a retry use my newly edited module settings?
The official guide says stored retries run with the settings from when the error happened. Configuration changes can require manual resolution.
Published November 7, 2026. Sources last read November 7, 2026. Send a correction.