evantorrie / k8sattributes-repro

How to reproduce k8sattributes issue in opentelemetry-collector:0.55.0 and later

Instructions to reproduce

Using opentelemetry collector 0.54.0

On a machine with docker already installed:

1. brew install kind kubernetes-cli helm
2. kind create cluster --config kind-1worker-config.yaml
3. kubectl config use-context kind-kind
4. helm repo add open-telemetry https://open-telemetry.github.io/opentelemetry-helm-charts
5. helm install my-otel-demo open-telemetry/opentelemetry-demo
6. helm install --values values.agent.yaml agent open-telemetry/opentelemetry-collector
7. helm install --values values.gateway.yaml gateway open-telemetry/opentelemetry-collector
8. kubectl patch deployment/my-otel-demo-frontend --type json --patch-file patch-my-otel-demo-1.yaml
9. kubectl set env deployment/my-otel-demo-frontend OTEL_EXPORTER_OTLP_ENDPOINT='http://$(K8S_NODE_IP):4317
10. kubectl port-forward svc/my-otel-demo-frontend 8080:8080
11. kubectl logs -f deployment/gateway-opentelemetry-collector

• Note: You may need to wait a while before step 9 to ensure all of the my-otel-demo deployments have succeeded.

• Go to your browser and access http://localhost:8080

• Watch for the traces coming in, and note that the debug logs from k8sattributesprocessor indicates that it has matched the pod IP from k8s.pod.ip and is adding the requested attributes to the resource associated with the traces.

Using opentelemetry collector 0.55.0

• Edit values.gateway.yaml and replace image.tag by "0.55.0"
• Run helm upgrade gateway --values values.gateway.yaml open-telemetry/opentelemetry-collector
• Run the same kubectl logs -f deployment/gateway-opentelemetry-collector and now note that the k8sattributesprocessor can no longer match the incoming pod IP specified in k8s.pod.ip