Control plane path
The API server remains the control-plane entry point for desired state. Controllers reconcile drift, the scheduler places pods and etcd preserves cluster metadata. The portal mirrors this vocabulary so operators can move between UI labels and normal Kubernetes troubleshooting without translation.
Traffic and exposure
- Pod-to-pod communication is handled by the cluster networking layer.
- Services provide stable east-west addresses.
- Ingress resources define north-south HTTP(S) routes and TLS termination points.
WarningIf ingress appears healthy while upstream checks still fail, inspect service endpoints and recent events before escalating to certificate or DNS owners.
Stateful services
Stateful services such as databases, queues and metrics stores rely on persistent volumes, predictable identity and ordered rollout behavior. Review stateful workloads together with node placement, storage class expectations and restart patterns.
Access boundaries
The portal provides read-only operational visibility. Object inspection, health review and relationship tracing happen here; mutating actions stay on the standard administration path.
| Area | Typical use | Follow-up path |
|---|---|---|
| Portal | Status review, event reading, image lookup | Continue to admin tooling for write actions |
| Observability | Related signals, warning streams, event interpretation | Use object links to return to affected workload or namespace |
| Registry | Image version, digest and ownership review | Use rollout or delivery tooling for promotion |