CPU
Monitor per-core and aggregated CPU usage ratios to track system load and identify process bottlenecks. This collector provides granular visibility into time spent in user space, kernel space, and waiting for I/O.
Setup
CPU monitoring is enabled by default. The agent automatically discovers available CPU cores and starts collecting metrics by reading system time statistics from /proc/stat (on Linux) or equivalent system APIs. No manual configuration is required.
Configuration
No configuration required.
Metrics
| Metric | Type | Unit | Description |
|---|---|---|---|
cpu_user_ratio | Gauge | % | Time spent in user mode. |
cpu_system_ratio | Gauge | % | Time spent in system (kernel) mode. |
cpu_idle_ratio | Gauge | % | Time spent in an idle state. |
cpu_iowait_ratio | Gauge | % | Time spent waiting for I/O operations to complete. |
cpu_steal_ratio | Gauge | % | Time stolen by the hypervisor (in virtualized environments). |
cpu_nice_ratio | Gauge | % | Time spent on low-priority processes. |
cpu_irq_ratio | Gauge | % | Time servicing hardware interrupts. |
cpu_softirq_ratio | Gauge | % | Time servicing software interrupts. |
Labels
| Label | Description | Values |
|---|---|---|
cpu | Core identifier. | total (aggregated), cpu0, cpu1, etc. |