integrator.io resolves and retries certain categories of errors on its own, without any action from you. This article explains the four automatic behaviors — auto-resolving duplicate errors, auto-retrying intermittent errors, auto-detecting a restored connection, and auto-recovering from rate-limit errors — and what each one depends on so you can tell when to expect them to kick in.
Auto-resolving duplicates and auto-retrying intermittent errors both depend on Auto-resolve errors with matching trace key being enabled for the flow. Auto-detecting a restored connection and auto-recovering from rate limits are connection-level settings, enabled independently.
Auto-resolve depends on the trace key
Auto-resolve identifies duplicate error records by matching their trace key — the unique identifier for a record. If your trace key isn't set or overridden correctly, the platform can't reliably recognize that two errors refer to the same record, so it's important to confirm your trace key setup before relying on auto-resolve. For instructions, see Set a custom trace key to uniquely identify a record.
Auto-resolving duplicate errors
When a flow runs again and a record with a previously-errored trace key either succeeds or is ignored, the previous error for that trace key moves from the Open errors tab to the Resolved errors tab automatically — you don't need to review or dismiss it yourself. The following scenarios illustrate the behavior:
| Scenario | Without auto-resolve | With auto-resolve |
|---|---|---|
| The flow runs again and a new error occurs for a record with the same trace key. | Both the new and the previous error appear in Open errors. | Only the new error appears in Open errors; the previous one moves to Resolved errors, since it's no longer the relevant blocker. |
| The flow runs again and the record with the same trace key now processes successfully. | The previous error stays in Open errors. | No error appears in Open errors; the previous one moves to Resolved errors, since no further action is needed. |
| The flow runs again and the record with the same trace key is now ignored. | The previous error stays in Open errors. | No error appears in Open errors; the previous one moves to Resolved errors, for the same reason. |
Manually resolved errors are labeled as user resolved; automatically resolved errors are labeled as auto resolved, so you can tell the difference in the Resolved errors tab.
Auto-retrying intermittent errors
When integrator.io classifies an error as Intermittent and Auto-resolve is enabled for the flow, the platform automatically retries it on a fixed schedule, up to four attempts:
| Attempt | Time elapsed since the error | Action |
|---|---|---|
| Initial | 00:00 | Error occurs; auto-retry mode starts. |
| First | 00:30 | First retry attempt. |
| Second | 01:30 | Second retry attempt. |
| Third | 03:30 | Third retry attempt. |
| Final | 07:30 | Final retry attempt. |
Intermittent errors are more common than they might seem — they represent cases where a third-party API didn't succeed on the first attempt but recovers without any change to the original request.
Auto-detecting a restored connection
When Auto-resolve is enabled and integrator.io receives an error classified as Connection because an endpoint is offline, it pauses the flow rather than continuing to fail against an unreachable endpoint. It then checks periodically to see whether the connection has been restored, using an exponential backoff: it checks after 1 hour, then 2, 4, 8, 16, and 32 hours. As soon as the connection comes back online, integrator.io detects it automatically and resumes the flow. Any errors that occurred before or after the pause remain available for you to retry once the flow is running again.
Auto-recovering from rate-limit errors
A rate limit — sometimes called a governance limit, particularly by applications like NetSuite and Salesforce — caps how many API calls an endpoint accepts within a given period. When a connection exceeds that cap, it starts receiving rate-limit errors.
The Auto-recover rate limit errors connection setting retries these errors automatically, lowering concurrency to 1 and retrying with a delay that doubles after every attempt — 1 minute, 2 minutes, 4 minutes, and so on up to 1,024 minutes — until the connection's target concurrency level is reached again. This setting is enabled by default for new connections; for existing connections, you need to enable it manually:
- Go to Resources > Connections.
- Select the connection you want to edit.
- Under the Advanced section, select the Auto-recover rate limit errors checkbox, if it isn't already selected.
- Enter a Target concurrency level based on your endpoint's API policy.
- Select Save & close.
If you're using a Celigo prebuilt connection to an external application, such as NetSuite or Salesforce, that connection already accounts for the rate and concurrency limits the endpoint imposes.