What it does
Error handling decides what happens when a step fails — and makes sure the person who has to fix it knows what went wrong. An automation you cannot trust is an automation you end up turning off.
How it works
- The run stops and says why. The failed step is recorded with its error, so the run history points at the exact node rather than at the workflow as a whole.
- Run it again. From the run history, run a finished execution again. It repeats the run from the start, with the workflow version and trigger data it began with, as a new run — the failed one stays on record.
- Safe automatic retries. The engine retries a step automatically only when it is both retryable and idempotent — a sent message or a created order is never repeated for you. Retries wait 30 seconds and back off from there, up to an hour, so a struggling service is not hammered.
- Error branches and continue-on-error. A workflow can route a failure to its own steps, or let the other branches finish. Either way, a run with a failed step is still reported as failed.
- No caps to run into. No plan limits how many workflows, steps or runs you have, so nothing stops a run because of a count.
In practice
An HTTP request to your CRM fails because a key expired. The run shows the step that failed and the response it got. Update the connection, then run the execution again from its history: it replays the same event through the same version of the workflow.
Details
Error reporting, running an execution again, automatic retries and On error branches are on every plan, Free included.
