Handling Failures
Steps that call other services can fail for reasons that you cannot control. Examples are a timeout, a rate limit or a service that is down for a short time. NocoDB can automatically try such a step again before it stops. When a run fails, NocoDB shows you exactly what occurred.
Retry a step automatically
Action steps have an On failure section in their settings. This section is collapsed by default. Open it to select what occurs when the step fails.
| Setting | What it does |
|---|---|
| Retry if this step fails | Turns on automatic retries for the step. |
| Number of retries | Sets how many more times to try (up to 10). |
| Wait before retrying | Sets how long to wait before the next try. |
| Wait longer each time | Doubles the wait after each try, up to 1 hour. |
| Retry on | Temporary problems only: timeouts, rate limits and server errors. Recommended. Any error: all failures, also failures that will occur again in the same way. |
While a step waits for the next try, the run shows Retrying at… and the time of the next try. If the service asks NocoDB to wait (for example, after a rate limit), NocoDB waits for that time, up to 1 hour.
Which steps retry by default
Steps that only read data retry automatically on temporary problems. Examples are Find record, List records and AI steps. You do not have to configure anything.
Steps that send or change data do not retry unless you turn on retries. Examples are sending an email, creating or updating a record and calling an API. A retry can do the action again, for example send the same email two times.
When a run fails
When a step still fails after its retries, NocoDB marks the run as Failed. Open the run in the execution logs to see:
- Which step failed and its error
- Each attempt, with its error and what the step returned
- The inputs and outputs of the steps that ran before it
From there, you can retry the run. After you correct the problem, continue from the failed step, or run the whole workflow again.
If a run is interrupted
A run can be interrupted on our side, for example during an update. If the current step is safe to repeat, NocoDB automatically continues the run. If not, NocoDB marks the run as Stopped unexpectedly. Nothing is lost. Retry the run to continue from where it was.
If a schedule can't be read
If the schedule of a Scheduled workflow is not valid, NocoDB turns the workflow off. NocoDB does not skip runs silently. The toolbar marks the workflow as Turned off and shows the reason. A failed run with the explanation shows in the run history. Thus all users who subscribe to the failures of the workflow get a notification. To turn the workflow on again, correct the schedule and publish the workflow.
Availability
- Available on the NocoDB Cloud Plus plan and higher, and on licensed on-premise deployments.
Last updated on