improve warning message on Capturing ColdStart metrics

[amirkaws]

Expected Behaviour

When metric decorator is added on top of the lambda handler to capture Cold Start metric, we should not get any warning message for cold or warm lambda invocation.

Current Behaviour

When metric decorator is added on top of the lambda handler to capture Cold Start metric, we get a warning message. No metrics to publish, skipping.

Code snippet

@metrics.log_metrics(capture_cold_start_metric=True)
def lambda_handler(event: dict, context: LambdaContext):
    ...

Possible Solution

No response

Steps to Reproduce

• Add @metrics.log_metrics(capture_cold_start_metric=True) to the lambda handler
• Ensure no more metrics is added in the lambda
• Run Lambda
• UserWarning: No metrics to publish, skipping appears in function log

AWS Lambda Powertools for Python version

latest

AWS Lambda function runtime

3.9

Packaging format used

Lambda Layers

Debugging logs

No response

[boring-cyborg]

Thanks for opening your first issue here! We'll come back to you as soon as we can.
In the meantime, check out the #python channel on our AWS Lambda Powertools Discord: Invite link

[rubenfonseca]

Hi @amirkaws thank you so much for opening this issue! We did a check on the code, and as we understand it, the feature works correctly. We think the confusion is that capture_cold_start creates a separate EMF Blob because it uses a different Metric dimensions (to avoid data loss). It happens regardless if there are application metrics or not, as long as it's a cold start.

The "empty metrics" warning is expected if there are no metrics being added to be flushed. Cold start metrics will always be flushed if there is a cold start:

Does this help explain the behavior you're seeing?

[amirkaws]

Hi @rubenfonseca, thanks for your prompts response. From the end user prospective, if capturing the ColdStart is the only metric I am interested in, Should I get warning, which is the case here?

[rubenfonseca]

The reason we add the warning and print what seems to be an empty metric warning is that it could be useful to detect bugs on the customer code (we already had cases with conditional metrics). In those cases, there could be useful metric metadata which would be lost otherwise.

Having said that, you have a point that the current DX could be misleading or create confusion. Would it make sense to change the warning message to be more explicit? "No application metrics to publish, skipping".

[amirkaws]

Sure no problem, I just wanted to ensure that's expected results, customer capturing cold start metrics and we are warning them metrics to publish. As you mentioned may be more specific message would help. thanks

[rubenfonseca]

@amirkaws we were discussing this internally and come up with even a better one:

No application metrics to publish. The cold-start metric may be published if enabled. If application metrics should never be empty, consider using "raise_on_empty_metrics".

WDYT?