Kubernetes
Explore a cluster
Find your way around a connected cluster: health, workloads, logs, cost, alerts and exports.
About 5 minutes
Once a cluster is connected, every page under Kubernetes in the sidebar includes it. The pages show all your clusters together; where a page has a cluster picker, use it to look at one.
The cluster page
Open Kubernetes → Clusters and click a cluster.
- Cluster Risk Score: one score out of 100 from reliability, security, cost and performance, with the reason under each.
- Read access: how Zaysa reads the cluster. Reading keyless means through the connector, with no cloud key stored.
- Cloud account: the account that owns the cluster. Autopilot reads its change history when something breaks that the cluster alone cannot explain.
- Node Health Map and Namespace Heatmap: every node and namespace, with how many pods are running.
At the top right:
| Button | What it does |
|---|---|
| Refresh live metrics | Reads the cluster again now, instead of waiting for the next automatic read. |
| Check permissions | Asks the cluster what Zaysa may do in it. It changes nothing. See Check permissions. |
| Remove | Removes the cluster from Zaysa. Nothing in the cluster is changed; the connector keeps running until you remove it. |
What each page shows
| Pages | What you see |
|---|---|
| Overview, Clusters | Health, nodes, pods, workloads and monthly cost of all clusters, and the top issues. |
| Namespaces, Workloads, Pods, Nodes, Services, Ingresses, ConfigMaps, Storage, Network, RBAC, Autoscalers, Disruption Budgets, Events | The same objects kubectl get lists, read live from the cluster. Click a row for details. |
| Secrets | Names, types and how many keys each Secret has. Values are never read, stored or shown. Clusters read through a connector list no Secrets at all, because the connector is not allowed to read them. |
| Logs | Recent logs of any pod. Turn on Live to follow new lines. |
| Monitoring, Graph | CPU and memory use, availability and cost trends, and a map of how ingresses, services, workloads, pods and nodes connect. |
| Cost | Monthly spend, idle waste and savings per cluster, at live cloud prices. |
| Alerts | Rules that tell you when something goes wrong, and the alerts they raised. |
| Security, Reports, AI Assistant | Security findings, exports, and questions in plain language about your clusters. These need the Pro or Business plan. |
Read a pod’s logs
Open Kubernetes → Logs, pick the cluster and namespace, search for the pod and click it. Live follows new lines; Refresh loads the latest ones again.
Cost
Zaysa prices every node and disk from your cloud provider’s current price list, by machine type and region. If a node’s price cannot be found, it is left out rather than guessed. Idle waste is the part of the node cost your workloads do not use, measured by the busier of CPU and memory. Potential savings is what you could reclaim from that idle spend and by rightsizing.
Get alerted
Open Kubernetes → Alerts and click New Rule. Choose what to watch, for example CrashLoopBackOff pods, ImagePullBackOff pods, OOMKilled pods, Nodes NotReady or Pending pods, which clusters, and where to send it: email, Slack, Microsoft Teams or a webhook.
Rules are checked when Zaysa reads a cluster and once an hour. When the problem is gone, the alert closes on its own and a Resolved message goes to the same places.
Export a report
Open Kubernetes → Reports and click Export clusters (CSV) (1) or Export clusters (JSON) (2). Each row is one cluster: name, provider, region, version, nodes, status, CPU and memory use, monthly cost, idle spend, potential savings, open issues and when it was last read. The savings match the Cost page.
Fixes