Testing and runs
Try out a flow without sending anything and see what real runs have done
Last updated on
In the editor you work with two tabs to check that a flow does what you want: Test run for a trial and Runs for what really happened. You open them on the right in the panel. The fourth tab, History, shows earlier versions of the flow, see Earlier versions.
A test run
Choose the Test run tab. A test goes through the flow with sample data and shows which steps it would take.
Nothing real happens
A test run sends nothing, creates nothing and changes nothing. Delays are skipped, so you see the end result straight away. Actions only show what they would do, with the text Not carried out in a test.
Sample data
At the top is Test with a real record. For each record of the trigger, such as Occupation, choose a record via Choose a record (search by name or number). Tillor loads the record as it is now, with the records it refers to, into the sample data. The test still sends, creates and changes nothing.
Under Sample event data is the data the test works with, in JSON form. If the automation has already run for real, the test starts with the data of the last real run. Otherwise Tillor fills in a sample that has the shape of the event: a text becomes example, a number 1 and a date the current moment.
Change the values to try other cases, for example a different terrain type or an invoice with a different status. If the text is not valid JSON, it says The sample data is not valid JSON. and you cannot start the test.
If the flow contains an API request, there is also a Sample API response field. Paste the response you expect from the request there. With it the test can go through the steps after the request, and Tillor offers the fields of the response as variables. See Using an API response.
What happens at these steps
For some steps the outcome depends on the outside world: Send webhook, API request, actions with a Skipped or Error output, and logic such as Open now?, Look up a record, Wait for an event and Ask for a decision. Under What happens at these steps? you choose per step which output the test carries on through. That way you also test the path for when things go wrong.
The result
Click Run test. You get to see the steps in order, with for each step:
- the filled-in values, for example the title of a task as it would be created
- for a condition, per line what was compared and whether it matched. Green is passed, red is not, with the value that was actually there (was …)
- which output the flow carried on through (Continues on: Yes)
- for a delay, until when it would wait
After running, the button is called Run again. The test plays the steps one by one (Step 2 of 5…). With Show all you skip that playback. On the canvas the path the test took is coloured, and the steps that did not get their turn are dimmed. Click a step name in the result to jump to that step. At the bottom it says Finished after 5 steps.
If the filter under Only when does not match your sample data, you see The trigger's filter does not match this data, so the flow would not start. Then check which condition is red. With Clear you remove the result from the canvas again.
What a test does not show
- Whether an invoice can really be sent, whether an endpoint is reachable, or whether an API path exists
- What records look like after a delay. The test works with your sample data and reads nothing again
- Errors that only occur with real data, such as an empty field
So always take a look at the first real runs on the Runs tab.
Real runs
The Runs tab shows the last 20 real runs of this automation, the newest at the top. The list refreshes itself about every 15 seconds. As long as nothing has started yet, it says No runs yet. Enable the automation and wait for its event.
Each run shows the start time, the number of steps or (when waiting) when it carries on, and a status.
| Status | What it means |
|---|---|
| Running | The run is going at this moment |
| Waiting | The run is in a delay, waiting for an event or for an approval, and carries on later |
| Completed | The flow has been run through to the end |
| Failed | A step could not be carried out. The message is shown as Failed: … |
| Canceled | The run stopped because the automation was paused or deleted while it was waiting |
Click a run to open it. You then see:
- all the steps that were taken, with the time and what the step did
- the filled-in values, for example the title the task really got
- for a condition, which rules matched
- for a failed step, the error message, which may be in English
On the canvas the path this run took is coloured. Steps that did not get their turn are dimmed. Click a step name to jump to that step.
Below the steps is a button such as 3 payloads. It shows the data the run started with and what each action produced, such as the created task id or the response of an API request. That way you can check which values a variable had.
At the bottom of an opened run is also a Trace: a code of letters and digits. If something is wrong with a run, pass that code on to support. That way we can find exactly this run.
Events that did not pass the filter under Only when, or for which the automation already ran with Run once per record, do not start a run and are therefore not in the list.
In the overview of automations, each automation shows the time of its last run, with a small bar per recent run next to it.
What happens next?
- After a test run nothing has changed in Tillor. Only when you turn on Active does the automation really run.
- Real runs stay on the Runs tab. It shows the last 20.
- A failed run results in an email to whoever made the automation, at most one per hour.
Problems?
A run that will not start
See Frequently asked questions for a checklist.