Monitoring
The Monitoring overview answers one question: is the environment healthy right now. It is the page you land on after sign-in, at /admin/monitoring, and the starting point for every other Monitoring page.
The Monitoring group has seven pages. This page is the Overview. The others are Platform Advisor, Cluster, Host Agents, Tasks, Logs and Live Ingestion.
The advisory banner and health summary
A banner at the top counts active problems by severity, for example “4 environment advisories need attention, 1 critical, 3 high”. Press View advisories to open the Platform Advisor.
Below it, the Health summary shows Healthy, Needs attention or Critical, next to the cluster name and mode. Treat this as a coarse signal. The advisories and the gauges below carry the detail.
The Refresh control forces an update. The page also refreshes on its own every few seconds.
Platform usage
The Platform usage cards count objects held by the analytics core. These are the same entities you manage elsewhere in the console.
| Card | Meaning |
|---|---|
| Sessions | Total sessions, with how many arrived in the last 7 and 30 days |
| Templates | Registered session template definitions |
| Projects | Registered projects |
| Node instances | Configured nodes across all sessions |
| Node outputs | Stored results across all sessions |
| Dimensions | Published dashboard results |
A Runner tasks row summarises the queue by state: Pending, In progress, Retrying, Failed and Completed. A long Pending count with idle runners points at a scheduling or capacity problem. See Tasks.
Cluster summary and capacity
Three cards summarise the cluster: Nodes, Pods running against the pod limit, and the environment Mode.
Below them, gauges show capacity in use:
| Gauge | What it measures |
|---|---|
| CPU | Cores in use against cluster capacity |
| Memory | Memory in use against cluster capacity |
| Blob store | Disk used on the managed object store. This is the headline storage figure |
| Network | Ingress throughput, where the cluster exposes it |
Each gauge carries a sparkline over a History range of 1 hour, 24 hours or 7 days, so a trend is visible even when nobody had the page open. Open in chart explorer opens the same metric at full size in the chart explorer, with Latest, Average, Minimum and Maximum readings over the retained history.
Watch the blob store gauge closely. An air-gapped environment can become unstable once storage is exhausted. Sessions shows which sessions are the heavy ones, and session retention keeps stored outputs from growing without bound.
Service latency
Where the server has collected enough samples, latency cards show the p95 round-trip time to the Core API, the v2 API and a fixed read on the object store. Compare them to locate a problem. A high Core API figure with a healthy blob read points at the database or the API itself. A high blob read with a healthy Core API points at storage or the network path to it.