Watching a run

A run is one execution of a workflow. Whatever started it (the Run button, a schedule, a webhook, mail arriving or an assistant), it is recorded the same way.

Live, step by step

While a run is going, the canvas shows it: each step lights up as it starts and changes colour as it succeeds or fails. You do not have to refresh or wait for the end. A step typically takes around half a second; a run over fifty items takes a few minutes, most of it waiting on the services in between.

What each step records

Open a step in a run to see:

• Input. The values its fields had once the variables were filled in.
• Output. What it returned, which is what the steps after it read.
• Log. What it did, line by line, including every retry and why.
• Status and time. Whether it succeeded, failed or was skipped, and how long it took.

In a loop, each turn is recorded on its own, so you can see which item failed.

What it cost

Every run records what it cost, on the run itself rather than only on an invoice at the end of the month. See Credits and billing.

Stopping a run

A run that is going somewhere it should not can be cancelled while it runs. Steps that already finished stay finished; what they sent has been sent.

Run history

Past runs are listed per workflow, with who or what started them, how long they took, how they ended and what they cost. How long history is kept depends on your plan. Run logs are yours, and yours to delete.

When something went wrong

1. Find the failed step: it is marked on the canvas and in the run.
2. Read its log and output. A failed HTTP call has the status code and the body the service sent back.
3. Fix the step, its fields or the account it acts as.
4. Run it again on the same data.

If it is a failure you expect to happen again, give the step an error route.