Metric Panels And History
Build a grid of current readings, status pills and historical graphs.
Put readings on a grid
In Design, choose + Metrics above the grid. You can also select equipment and choose Place Metrics On Grid in Properties. Name the panel, choose equipment, then select a telemetry key, service/check reading, or port state.
A panel can hold one reading or up to 16. Add readings for a host's CPU, memory, storage and service health, or combine equipment into a grid overview. Each reading has its own label, units, display style and graph toggle. Drag the panel to move it. Select it and drag its lower-right corner to resize it. Add a separate panel in another grid to build a dashboard; the source equipment stays where it is.
Choose a source
Telemetry keys are the same keys used by Live Values and automation conditions. Reported keys appear in the selector, and you can type a custom key before its first report arrives. Unreported and expired readings show Unknown.
The standard agent reports CPU, memory, disk and system uptime. Network throughput, packet rates and GPU readings need a source that reports them: an agent script, an SNMP/API check with a metric mapping, or a telemetry post. Choosing a key does not install a new collector or enable a hardware counter.
Useful keys include cpuPercent, memoryPercent, diskPercent, gpuPercent, networkRxBytesPerSecond, networkTxBytesPerSecond, networkRxPacketsPerSecond and networkTxPacketsPerSecond. Byte rates are rates, not cumulative counters. Calculate counter differences over elapsed seconds at the collector, and discard a delta when a counter resets. Use separate keys for separate interfaces or disks.
A service reading can show health, returned value, custom script metrics, continuous uptime/downtime, or observed availability. For network uptime, choose the uptime reading of the relevant reachability check. This measures successful observations from that check's location, not a guarantee of end-to-end availability.
Port states use the same model as the equipment view. A state may be measured, manually configured or inferred. Drawing a cable does not measure traffic or prove a physical link is connected.
Status pills
Choose Green / Red Status Pill and enter the source values that mean healthy and unhealthy. For service status these are up and down. For a script returning a flag they might be 1 and 0. Missing, expired and unmatched values stay neutral rather than turning green or red.
Inspect history
Use History on a reading, Metric History in equipment Properties, or History beside a Live Value. These views query the same organization history. Ranges are 1 hour, 6 hours, 24 hours, 7 days and 30 days.
History starts collecting after this feature is installed. There is no backfill for previously discarded readings. Numeric readings and observed status are stored in one-minute buckets for 30 days. Longer ranges combine buckets, showing averages with minimum/maximum values available on graph points and in the recent-values table. The table also shows the last text or status value in each bucket.
Status graphs represent up/live as 1 and down/offline/unplugged as 0. Gaps indicate missing or expired data; the graph does not interpolate across them. A single bucket can contain multiple observations, so it is not a packet capture or a complete event log. Use Activity for individual automation events.
History collection runs on the server even when the grid is hidden or no browser is open. Simulated changes are not recorded. Organization and grid permissions apply to both current sources and history. A panel referencing equipment you cannot view is hidden.