How to Configure Integration Retry

Integration Retry enables automatic retry logic for failed integrations. When an integration run fails, the system can automatically retry the execution based on a configurable schedule, which reduces the need for manual intervention in cases of transient failures.

Prerequisites

• Admin or staff permissions on the Integration Gateway instance.

• An existing integration with an Integration Config record.

Configure Retry Settings

1. Navigate to Build.

2. Right-click the integration you want to configure and select Configure.

3. In the Configure Integration dialog, expand the Retry Settings panel.

4. Select a Retry Mode:

   • None — No retries (default).

   • Fixed Interval — Retries at a constant delay between each attempt.

   • Exponential Backoff — Retries with increasing delays. Each subsequent retry waits longer than the previous one.

5. Set the Initial Retry Delay (seconds) — the base wait time before the first retry. For Fixed Interval, this is the delay between every attempt. For Exponential Backoff, this value doubles with each subsequent attempt.

6. Set the Retry Delay Jitter (seconds) — a random value between 0 and this number is added to each delay. Jitter prevents multiple failing integrations from retrying at the same time.

7. Set the Retry Count — the maximum number of retry attempts before the integration is marked as failed.

8. Click Save.

Field Reference

Field	Type	Default	Range
Retry Mode	Select	None	None, Fixed Interval, Exponential Backoff
Initial Retry Delay (seconds)	Integer	5	1–600
Retry Delay Jitter (seconds)	Integer	2	0–120
Retry Count	Integer	3	1–10

How nCino Calculates Retry Delay

Fixed Interval:

delay = retry_delay_seconds + random(0, retry_delay_jitter)

Exponential Backoff:

delay = (retry_delay_seconds × 2^attempt) + random(0, retry_delay_jitter)

Where attempt starts at 0 for the first retry.

Example: With a 5-second base delay, 2-second jitter, and Exponential Backoff:

Attempt	Base Delay	Possible Total (with jitter)
1st retry	5s (5 × 2⁰)	5–7s
2nd retry	10s (5 × 2¹)	10–12s
3rd retry	20s (5 × 2²)	20–22s

Viewing Retry History

Each retry attempt creates its own entry in the Run History. Retry runs are labeled with retry_for_run_id:<original_id>, which links them back to the original failed run.

The retry_attempt variable is available in integration hooks, which allows hook logic to behave differently based on the current attempt number.

Limits

The maximum effective retry delay cannot exceed 5 minutes (300 seconds). For Exponential Backoff mode, the system validates that the longest possible delay (on the final retry attempt plus jitter) stays within this limit.