Home / BeaverDeck / Docs / User Guide / Insights
Start With Insights
Insights are grouped checks that help start triage from symptoms instead of raw Kubernetes tables.
Open an Insights section, select check types, and use Open Resource or Open Logs
to jump to the related object.
| Section | What it checks | Typical action |
|---|---|---|
| Node Insights | Node readiness, pressure conditions, metrics pipeline, requested-resource capacity, and low requested usage below the underutilization threshold. | Open the node, view pods on it, drain or uncordon when permitted. |
| Workload Insights | Pending pods, waiting containers, OOMKilled containers, high restarts, missing requests, request usage, and overrequested pods. | Open the workload or pod, inspect logs, edit resource requests, restart, scale, or delete when permitted. |
| GPU Insights | GPU capacity discovery, allocation pressure, pending GPU pods, GPU node scheduling, workload mix, and GPU quota signals. | Check GPU node placement, quotas, pending pods, and whether GPU metrics are available. |
| Network Insights | LoadBalancer address state, service endpoints, ingress TLS, ingress backends, service coverage, and route collisions. | Open the service or ingress, then inspect its manifest and related endpoints. |
| Storage Insights | PVC binding, PVC usage, and released PVs. | Open the PVC/PV and inspect capacity, binding, reclaim state, and storage class. |
| Security Insights | Privileged pods, root-user security context, sensitive literal pod env values, and NetworkPolicy coverage. | Open the pod or namespace resources and replace literals with Secrets or identity-backed configuration where appropriate. |
| Configuration Insights | Missing Secret or ConfigMap references used by pod env, envFrom, and volumes. | Create the missing object or fix the pod/workload reference. |
For the exact trigger, risk, remediation steps, and limitations of every check, open the Insight Checks Guide.
Use Show all checks to include passing checks and Show suppressed alerts to
review ignored alerts. Ignoring or restoring an alert updates a global suppressed-Insights ConfigMap,
so the result applies to all users.