A connector can fail briefly because its service is busy or limiting requests. A retry policy repeats that action after a delay. The retry count means additional attempts after the first request, so a count of two allows at most three requests.
Open the action's Settings and choose Fixed interval or Exponential interval. Set the count and an ISO duration such as PT10S. Fixed repeats the same interval; exponential increases it after each failure. This workspace records simulated intervals immediately, keeping every attempt visible in Run details. It retries 408, 429 and 5xx responses. Authentication and validation failures go straight to recovery.
A retried write can duplicate work if the destination already accepted the first request. Establish the write identity and destination behavior before enabling retries. A successful retry permits the next action but never erases the earlier failed attempt.
Retry count is the number of additional attempts after the first request. Fixed intervals repeat the same delay; exponential intervals grow. The local runner records the waits without delaying the browser. A successful retry leaves the failed attempt visible and permits the next action.