No metrics/traces are being sent to OpenTelemetry Collector after configuring opentelemetry-instrumentation-django

[veziak]

Describe your environment

OS: Ubuntu
Python version: 3.12
Django==5.1
opentelemetry-sdk==1.26.0
opentelemetry-exporter-otlp==1.26.0
opentelemetry-instrumentation-django==0.47b0
opentelemetry-instrumentation-wsgi==0.47b0
opentelemetry-distro==0.47b0

What happened?

opentelemetry-instrumentation-django doesn't send any metrics/traces

Steps to Reproduce

configure opentelemetry-instrumentation-django following https://opentelemetry-python.readthedocs.io/en/latest/examples/django/README.html
configure OpenTelemetry Collector to collect metrics/traces

Expected Result

metrics/traces should be sent to OpenTelemetry Collector

Actual Result

Nothing is being sent

Additional context

django app repo: https://github.com/veziak/django-sample-otel

Would you like to implement a fix?

None

[emdneto]

Unfortunately, I can't reproduce your issue. I tested the example in the docs by following the instructions, and it's working fine. Are you getting any errors? It's working for manual and auto instrumentation to send traces/metrics to console and collector through OTLP. Tested with python3.12

[veziak]

Hi @emdneto , I don't have any errors, just nothing happens. What are the metric names that are being sent when you test it?

[veziak]

I also tested the django app in the examples and traces seem to work, but there is a code in a view that is not mentioned anywhere in docs:

trace.set_tracer_provider(TracerProvider())
trace.get_tracer_provider().add_span_processor(
BatchSpanProcessor(ConsoleSpanExporter())
)
Should it be added manually even with auto instrumentation?
And metrics still are not being sent

[emdneto]

@veziak Here are the metrics I'm getting in collector using your example and the example in the docs:

ScopeMetrics SchemaURL: 
InstrumentationScope opentelemetry.instrumentation.django 0.47b0
Metric #0
Descriptor:
     -> Name: http.server.active_requests
     -> Description: Number of active HTTP server requests.
     -> Unit: {request}
     -> DataType: Sum
     -> IsMonotonic: false
     -> AggregationTemporality: Cumulative
NumberDataPoints #0
Data point attributes:
     -> http.method: Str(GET)
     -> http.server_name: Str(localhost)
     -> http.scheme: Str(http)
     -> http.host: Str(localhost:8000)
     -> http.flavor: Str(1.1)
StartTimestamp: 2024-08-16 11:49:11.375342946 +0000 UTC
Timestamp: 2024-08-16 11:51:00.722118688 +0000 UTC
Value: 0
Metric #1
Descriptor:
     -> Name: http.server.duration
     -> Description: Duration of HTTP server requests.
     -> Unit: ms
     -> DataType: Histogram
     -> AggregationTemporality: Cumulative
HistogramDataPoints #0
Data point attributes:
     -> http.method: Str(GET)
     -> http.server_name: Str(localhost)
     -> http.scheme: Str(http)
     -> net.host.name: Str(localhost:8000)
     -> http.host: Str(localhost:8000)
     -> net.host.port: Int(8000)
     -> http.flavor: Str(1.1)
     -> http.status_code: Int(200)

Are you sure you're not missing any step? Let me know if we can close this as it looks like both examples are working

[veziak]

Hi @emdneto, thanks for testing it!
Yep , it definitely looks like I'm missing something!
Could you also share your collector config pls? I used this one for my test https://github.com/open-telemetry/opentelemetry-python/blob/main/docs/examples/metrics/instruments/otel-collector-config.yaml

[emdneto]

For testing I usually use the default contrib. Closing the issue then. thanks

docker run -d -p 4318:4318 -p 4317:4317 otel/opentelemetry-collector-contrib

[veziak]

just tried it again with default collector configuration, the same result: no metrics are being sent to collector.
Also tested it on another computer with the same result.
Not sure what I could have missed:
it's the same application, I start it with python manage.py runserver, only DJANGO_SETTINGS_MODULE env variable is set.
I start collector with docker run -d -p 4318:4318 -p 4317:4317 otel/opentelemetry-collector-contrib
When I send a couple of requests to django app, nothing happens in collector logs, but collector is constantly logging it's own metrics: otelcol_process_*

[pmcollins]

Have you tried running opentelemetry-instrument? It's mentioned further down in the readme. If you don't use opentelemetry-instrument you'll have to configure OTel some other way (the set_tracer_provider stuff you mentioned above).

[veziak]

@pmcollins just tried it with opentelemetry-instrument and commented out DjangoInstrumentor().instrument()
traces looks ok, but there are still no django metrics in collector's logs

[emdneto]

@veziak here are the steps I tried:

python3.12 -m venv .django
source .django/bin/activate
pip install --upgrade pip
pip install opentelemetry-distro opentelemetry-exporter-otlp
pip install Django==5.1
opentelemetry-bootstrap -a install
export DJANGO_SETTINGS_MODULE=instrumentation_example.settings
export OTEL_EXPORTER_OTLP_INSECURE="True"
docker run -d -p 4318:4318 -p 4317:4317 otel/opentelemetry-collector-contrib:0.104.0
opentelemetry-instrument --traces_exporter console,otlp --metrics_exporter otlp  --service_name service --exporter_otlp_endpoint localhost:4317 python manage.py runserver --noreload

[veziak]

Thanks @emdneto , I managed to make it work!
Minimal configuration that worked for me:

export DJANGO_SETTINGS_MODULE=instrumentation_example.settings
export OTEL_EXPORTER_OTLP_INSECURE="True"
export OTEL_EXPORTER_OTLP_ENDPOINT=localhost:4317
export OTEL_SERVICE_NAME=test
opentelemetry-instrument python manage.py runserver --noreload

A few things that I noticed:

• OTEL_EXPORTER_OTLP_INSECURE="True" is important in case of local collector, without this env var application will log some errors about certificate
• I haven't managed to make it work with python manage.py runserver --noreload and configuration in the code. DjangoInstrumentor().instrument() doesn't do anything. I tried different combinations but none of them worked
• so it seems that opentelemetry-instrument python manage.py runserver --noreload is only way to instrument django

[emdneto]

@veziak If you are using manual instrumentation you should initialize and configure the SDK as @pmcollins said

[veziak]

@emdneto I think I did: I added DjangoInstrumentor().instrument() to manage.py , and set the same environment variables as for opentelemetry-instrument