Troubleshooting a Radius installation

This page explains some troubleshooting steps that can help you diagnose installation failures or crashes using a custom build of Radius.

Verify all Radius pods are up and running

kubectl get pods -n radius-system

The installation includes UCP, Applications RP, Dynamic RP, Controller, Deployment Engine, and the PostgreSQL database. The exact pod names contain generated suffixes. Every pod should become Running or complete successfully; use kubectl describe pod <pod-name> -n radius-system for a pod that remains pending or repeatedly restarts.

Examine logs

Read logs by deployment name instead of copying an ephemeral pod name:

kubectl logs deployment/ucp -n radius-system
kubectl logs deployment/applications-rp -n radius-system
kubectl logs deployment/dynamic-rp -n radius-system
kubectl logs deployment/controller -n radius-system
kubectl logs deployment/bicep-de -n radius-system

For a specific pod or container, use:

kubectl logs <pod-name> -n <namespace>

Logs with severity set to error or warn could point to why the pods are not in Running state.

Every rad CLI error includes a traceId that also appears in the control-plane logs. Search all Radius pod logs for that value:

kubectl logs -n radius-system -l app.kubernetes.io/part-of=radius --all-containers=true --prefix | grep <traceId>

See the observability documentation for more logging guidance.