Automation debugging and understanding behavior

Ask the Community
1

Hello Community,

I have an automation setup and have a few doubts regarding failure handling and debugging.

Scenario

My automation takes in an object set, applies a filter, and then the filtered objects trigger the automation. The automation then runs a function to update another object set.

For example:

  • Object Set A contains a large list of drivers across multiple regions along with their license numbers.

  • Since the dataset is very large, I created multiple automations, each filtering drivers for a smaller region.

  • These filtered objects trigger a function that updates Object Set B, where driver details and license numbers (as foreign key) need to be populated.

Doubt 1: Behavior when automation fails

Suppose the automation is triggered by 100 objects.

  • The function processes the first 70 objects successfully.

  • At the 71st object, there is an issue (for example, a validation condition causes the loop to get stuck and the function eventually times out).

In this situation, I want to understand:

  1. Are the first 70 successful updates committed/written back, or is the entire transaction rolled back?

  2. Does the automation continue processing objects 72 onward, or does execution stop completely once the failure occurs?

  3. Is the behavior transactional or partial-success based?

Doubt 2: Reasons for automation failure

I am aware of some common failure scenarios such as:

  • Function timeout

  • Automation being triggered by more than 10,000 objects

  • Attempting to write/update more than 100,000 objects

However, I am unable to clearly identify the exact reason for failures in some cases.

For example:

  • My automation runs hourly.

  • I received an email saying:

“Please verify the automation’s configuration if failures are unexpected. There was 1 failure for effects of type action.”

But when I checked the automation page:

  • The run status showed as “Succeeded”

  • No expected updates were present in the output dataset but how do I access the error logs

So, I would like to understand:

  • How can we properly debug such cases?

  • Where can we see detailed failure logs/error traces?

  • Is there a way to identify which specific object caused the issue?

  • Can partial failures occur even when the automation status shows as succeeded?

Would appreciate any guidance or best practices for debugging and designing resilient automations.