Maestro Error while calling the 2nd agent

Help Maestro
1

Hi All,

I have a BPMN maestro flow where I am running using debug and while running in debug mode, the flow error out in the 2nd agent (different than 1st agent) with the below error -

Error: Incident at element Activity_kKaf3G: Failure in the Orchestrator Job

But this 2nd agent running without any error while I debug it individually.

Can anyone please help on this ?

2

Hi @abhilashmohanty86

Seems the second activity could not start the job with the same context as when you debug it individually.

Please verify the following

  • Orchestrator folder/environment used in BPMN matches the one you used for debugging.

  • Open the failed job in Orchestrator and check the detailed logs, it normally points to the missing argument or permission issue.

  • Check the BPMN Activity configuration and confirm the correct process version

Once the arguments and folder permissions line up, BPMN flow runs without this error.

Regards,
Christopher

3

@abhilashmohanty86

Can you try recreating the issue with an new project? if yes, it’s a configuration issue somewhere, else it could be the Maestro project error and better to use new project.

4

@abhilashmohanty86

Are you calling any jobs as part of the second agent?

what speicifc error are you getting in jobs page?

cheers

5

@abhilashmohanty86 It sounds like the 2nd agent itself is probably fine, since it works when you debug it separately. So the issue may be more about how Maestro is invoking that agent from the BPMN flow rather than the agent logic alone.

Can you check below:

Input mapping between Agent 1 → Agent 2
Sometimes the 2nd agent runs fine alone because you pass test data manually, but fails in the Maestro flow because one required input is null, empty, or in a different format

Please share complete logs

6

Thanks you all for your inputs. :slightly_smiling_face:
Though in some point of time these input will definitely help me.

I found the issue to be on the structured output data validation that the agent was prompted to do. It was trying to output string but the agent was returning NULL based on the input supplied which was correct. So I made the structured output as optional in the prompt. It worked fine after that.