default dd-trace django Configuration = Results in Duplicate Spans for DB Queries

[colinbjohnson]

Summary of problem

When using a default install of the dd-trace module and running django requests to a backend database will be duplicated.

Which version of dd-trace-py are you using?

ddtrace==2.13.0

Which version of pip are you using?

pip 24.2

Which libraries and their versions are you using?

asgiref==3.8.1
bytecode==0.15.1
ddtrace==2.13.0
Deprecated==1.2.14
Django==5.1.1
envier==0.5.2
importlib_metadata==8.4.0
opentelemetry-api==1.27.0
protobuf==5.28.2
psycopg2-binary==2.9.9
sqlparse==0.5.1
typing_extensions==4.12.2
wrapt==1.16.0
xmltodict==0.13.0
zipp==3.20.2

How can we reproduce your problem?

1. Checkout the repository and directory here: https://github.com/DataDog/trace-examples/tree/master/python/django/django-tutorial
2. Run the command DD_API_KEY=... docker compose up --build -d
3. Login to DataDog and from the left-hand navigation select APM -> Traces -> Explorer

What is the result that you get?

We get nearly duplicate spans - one span is under the service defaultdb and one span is under the service postgres. A screenshot is provided below:

What is the result that you expected?

We expect, by default, to have only one span provided.

Other Notes:

1. We are aware that we can use a configuration line config.django['instrument_databases'] = False to remove the duplicate spans.
2. We do feel that the default behavior of a tracing library would be to somehow avoid the duplication of spans as this creates additional data that is not useful - this:

• creates a distraction for APM users (needing to filter through duplicates)
• increases DataDog APM costs for ingested spans
• likely increases network usage and slightly increases resource usage