TypeError: trace_id must be an integer occurs when multiple SQS messages to process

[annisall]

I have a lambda that is triggered by SQS. When there are multiple SQS messages to be processed, I can see that there's TypeError: trace_id must be an integer in the logs and no metrics is sent to DataDog.

Expected Behavior

Metrics are sent to DataDog without any errors.

Actual Behavior

TypeError mentioned above is raised and no metrics are sent to datadog.

Steps to Reproduce the Problem

1. Create a SQS queue and lambda that it triggers
2. Add multiple messages to SQS queue
3. Check the logs in Cloudwatch and search for the error

Specifications

• Datadog Lambda Layer version: dd_lambda_layer:datadog-python38_3.33.0
• Python version: 3.8.0

I tested with 3.33.0, 3.32.0 and 2.30.0. Both 3.33.0 and 3.32.0 had this problem, but with 2.30.0 the error didn't occur and data was sent to DataDog.

Stacktrace

File "/opt/python/datadog_lambda/wrapper.py", line 161, in _before
set_correlation_ids()
File "/opt/python/datadog_lambda/tracing.py", line 297, in set_correlation_ids
span = tracer.trace("dummy.span")
File "/opt/python/ddtrace/tracer.py", line 626, in trace
return self.start_span(
File "/opt/python/ddtrace/tracer.py", line 427, in start_span
span = Span(
File "/opt/python/ddtrace/span.py", line 114, in __init__
raise TypeError("trace_id must be an integer")
TypeError: trace_id must be an integer

[DylanLovesCoffee]

Hey @annisall, thanks for raising the issue. I was able to reproduce this with v3.33.0 when I have DD_TRACE_ENABLED=false. We'll keep digging into this.

[DylanLovesCoffee]

👋 @annisall I just released a new version of the Layer and serverless plugin which has the fix for this. Thanks again for reporting!