Python auto-instrumentation returns failed traces

[muandane]

Describe your environment I'm running python applications that use Pika, tensorflow and pytorch. in a kubernetes cluster, with operator version 0.94.0 and python-auto-instrumentation version 0.44b

Steps to reproduce

Just creating this instrumentation and adding the annotation to the python deployment results in the errors bellow

---
apiVersion: opentelemetry.io/v1alpha1
kind: Instrumentation
metadata:
  name: python-traces
  namespace: caas-apps
spec:
  exporter:
    endpoint: http://caas-collector.caas-infra.svc.cluster.local:4318
  propagators:
    - tracecontext
    - baggage
    - b3
  sampler:
    type: parentbased_traceidratio
    argument: '1'
  python:
    env:
      - name: OTEL_PYTHON_DISABLED_INSTRUMENTATIONS
        value: "urllib3,urllib,flask"

What is the expected behavior?

Not getting the failed "connection refused" errors

What is the actual behavior?

getting multiple traces of failing connection establishment

Additional context

i'm running other language instrumentations using different instrumentation definitions for java and nodejs and both of them work flawlessly, but the python one gives the previously show errors, i tried setting stuff like:

OTEL_EXPORTER_OTLP_TRACES_ENDPOINT: http://python-collector.caas-infra.svc.cluster.local:4318/v1/traces
OTEL_EXPORTER_OTLP_ENDPOINT: http://python-collector.caas-infra.svc.cluster.local:4318
OTEL_PYTHON_DISABLED_INSTRUMENTATIONS: "urllib3,urllib,flask,django,kafka,redis"

[xrmx]

The failing connections are coming from requests instrumentation, you may want to disable that too.

[muandane]

The failing connections are coming from requests instrumentation, you may want to disable that too.

Thank you that fixed it @xrmx