Retries

Automatic retry and backoff when a step fails.

Why retries#

Most action failures are temporary. Retrying with backoff turns a flaky dependency into a non-event, and keeps your alerts meaningful — you only hear about failures that persist.

Per-node config#

Every node exposes retry settings in its inspector — the max attempt count and the delay between attempts. Defaults are sensible; raise them for slow or rate-limited APIs, lower them for steps that must not repeat.

Backoff#

Retries wait between attempts (backoff) so you don’t hammer a struggling API. Each attempt is logged on the run, so you can see how many it took to succeed.

When retries run out#

If a step exhausts its attempts, the run is marked failed and your alerts fire. Fix the cause, then replay the run with its original input.