DataDogStepFunctions with Multiple add_state_machines Invocations Fails Due to LogGroup Collision

[mjdean94]

Expected Behavior

A CDK synth should still work if add_state_machines is called multiple times.

Actual Behavior

CDK synth fails with "There is already a Construct with name 'LogGroup' in DatadogStepFunctions." Removing the existing logging configuration on the statemachine seems to fix this.

Steps to Reproduce the Problem

1. Define two statemachines with log groups.

self.my_statemachine_one = steps.StateMachine(self, "StatemachineOne", logs=steps.LogOptions(destination=self.statemachine_one_log_group, level=steps.LogLevel.ALL)
self.my_statemachine_two = steps.StateMachine(self, "StatemachineTwo", logs=steps.LogOptions(destination=self.statemachine_two_log_group, level=steps.LogLevel.ALL)

1. Define DatadogStepFunctions construct.

self.datadog_step_functions = DataDogStepFunctions(self, "DDStep", forwarder_arn=dd_forwarder_arn) 

1. Call add_state_machines twice.

self.datadog_step_functions.add_state_machines([self.my_statemachine_one])
self.datadog_step_functions.add_state_machines([self.my_statemachine_two])

1. Run a CDK synth. Build fails.

Specifications

• datadog-cdk-constructs-v2>=4.2.0
• python>=3.13

[mjdean94]

My guess here is that it has to do with the scopes being different at different points in the code.

1. If you use an existing log group, it sets the scope of the construct to this: https://github.com/DataDog/datadog-cdk-constructs/blob/main/src/datadog-step-functions.ts#L150

2. If you don't have logging configured, the scope is the stateMachine: https://github.com/DataDog/datadog-cdk-constructs/blob/main/src/datadog-step-functions.ts#L107