Troubleshooting
Start with your component's logs and events: most failures show up there first. When they don't explain it, look at the resources OpenChoreo created for the component and the traffic between them.
Viewing Kubernetes Resources
Via Backstage UI
Click View K8s Artifacts on any environment card to see the resource tree:
- Deployments and ReplicaSets
- Pods (with status and events)
- Services
- HTTPRoutes
- Kubernetes events
The tree also follows resources created beneath these, and platform engineers can extend it to custom kinds.
This view helps diagnose issues like pod crashes, image pull errors, or routing problems.
Via CLI
# Stream the component's logs
occ component logs my-service --env development -f
# Check which release each environment runs and whether it is Ready
occ releasebinding list --component my-service
# Inspect the component's conditions
occ component get my-service
For Kubernetes-level inspection, use kubectl in the data plane:
# Check the underlying deployment
kubectl get deployment -A -l openchoreo.dev/component=my-service
# Check pod status
kubectl get pods -A -l openchoreo.dev/component=my-service
# Check events
kubectl get events -A --field-selector involvedObject.name=my-service
Network Flows
The Wirelogs tab on the component streams the network flows to and from it, live, as Cilium's network observability (Hubble) sees them. Each flow shows its verdict (FORWARDED or DROPPED), its direction, its source and destination, and a summary such as the HTTP method, path and status. Select a flow for its headers and raw details.
A DROPPED flow usually means a network policy blocked the call, for example to a component that isn't declared as an endpoint dependency.
- Select an environment and start the stream. It stops after a time limit set by the platform, 15 minutes by default, and the tab keeps the newest 500 flows.
- Filter the flows by verdict, workload, IP, path or status, and download them as JSON.
- Wirelogs need the Cilium network module on the environment's data plane. Elsewhere the tab says they are unavailable.
- The default
developerrole can't stream wirelogs. Ask your platform engineer for access.
Common Issues
| Symptom | Possible Cause | What to Check |
|---|---|---|
| Component stuck in "NotReady" | Data plane connectivity | Check ClusterAgent pod logs in the data plane |
| Pods in CrashLoopBackOff | Application error | Check pod logs via Backstage or occ component logs |
| Image pull error | Wrong image reference or missing credentials | Verify container image URL and registry access |
| No endpoints accessible | HTTPRoute not created or gateway misconfigured | Check kubectl get httproute -A and gateway pod logs |
| Deployment not appearing | ReleaseBinding not created | Check occ releasebinding list and Component conditions |
What's Next
- Deploy and Promote: deploy, promote, undeploy and roll back
- Environment Overrides: customize configuration per environment