It would be very helpful if automations have some sort of output on automation status once they run.
If there was a way we could send an email on failure, time-out, success, etc…
With this we could identify if any automation running in the background is having any issues that need to be addressed.
Hi @erikheckler are you asking for notifications for Automations errors, beyond the listing in the Run History? This is a request on our radar. Over the next few months we’ll start working on a centralized error log. Our first target is App Trigger errors, but the vision is that other errors including Automations would follow in the future.
As a workaround, some customers use tulip tables to store statuses and monitor those statuses with a helper Automation, to notify via email/Slack or another integration when an automation run has failed.
@OlgaStroilova Yes. Ideally we’d want to have a way that we can see at a glance any automation that has a run failure.
Then if possible a way to either re-run that specific run ID (useful in the case of a timeout). Or a way to see the underlying data that triggered the run, so we can have an idea of where to go to fix the error in the first place.
My current work-around is to store the unique ID of the table row I’m updating to a dummy table. Then the automation only looks to see if that table has a new ID in it. Once it sees a new ID it goes through the automation steps. Once it’s complete, it’s final step is to delete the ID off the table.
In this case, if the table is has no rows, then I know all automations successfully completed. And if it has a value, I can simply copy the ID, delete the row, then re-add the ID to re-run the automation.
Thank you for the detail. Yes, in the longer term centralized run history is on the longer term radar, after we get Functions out this year.
We’re also working on Automations Tracing now, that would show more detailed run history including the data that caused the failure.
This should hopefully help you and your team navigate failures individually.
We’ll also look into easy reruns/retries - this would require storing the input data and allowing a retry. Thank you for the suggestion.