load
Use case
Alert when the load average outpaces the host’s actual CPU count - works on Windows too (gopsutil synthesizes an equivalent value from the CPU queue length via WMI), not just Linux/macOS/BSD.
Configuration
| Field | Type | Meaning |
|---|---|---|
threshold1 |
float | default 6; alert if load1 > threshold1 x CPU count |
threshold5 |
float | default 4; alert if load5 > threshold5 x CPU count |
threshold15 |
float | default 2; alert if load15 > threshold15 x CPU count |
Check order: load15 first, then load5, then load1 - whichever trips first (critical, binary - no graduated severity) is the one reported; the others aren’t evaluated once one has fired.
Output
All datapoints:
| Item | Type | Meaning |
|---|---|---|
load_cpu |
int | available CPUs |
load1 |
float | load average, 1 minute |
load5 |
float | load average, 5 minutes |
load15 |
float | load average, 15 minutes |
Example
checks:
- module: load
name: load_local
conf:
threshold1: 6.0
threshold5: 4.0
threshold15: 2.0