Server metrics show host pressure and help explain pending tasks, slow applications, or storage exhaustion. Open a server’s metrics view and choose the relevant period.
Available observations
Inspect CPU, memory, network, and filesystem series using the displayed units. Host collection focuses on physical network interfaces and writable local block-backed filesystems; it is not a list of every virtual device or application volume.
Server metrics default to sampling every five seconds. Filesystem sampling is fixed at five minutes, independently of the ordinary sampling interval.
Diagnose pressure
| Symptom | Compare |
|---|---|
| CPU saturation | Application CPU, replica count, reservations and limits |
| Memory pressure | Host available memory, process failures, application limits |
| Disk nearly full | Filesystem usage, retained logs, application volumes |
| Missing reports | Server state, collector status, network and Insight |
| Traffic spike | Physical interface series and application access patterns |
Host free space and individual volume size answer different questions. Use volume management to inspect local volumes.
Verify and interpret
Check sample timestamps and time range before acting. Missing observations are not zero usage.
After adjusting collection or freeing space, wait for fresh samples at the appropriate cadence. Compare with application metrics and review collection settings if reports remain stale.