Type: load_average | Runner: Agent only
Reads the 1-minute system load average from /proc/loadavg and compares it against thresholds expressed as a multiple of the host's CPU count.
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
warn_factor |
float | No | 0.8 |
Warn when 1-min load exceeds this factor × CPU count (e.g. 0.8 on a 4-core server = warn at load > 3.2) |
fail_factor |
float | No | 1.5 |
Fail when 1-min load exceeds this factor × CPU count. Must be greater than warn_factor. |
| Status | Condition |
|---|---|
| OK | Load < warn_factor × CPU count |
| Warn | Load >= warn_factor × CPU count |
| Fail | Load >= fail_factor × CPU count |
| Unknown | /proc/loadavg unavailable — only works on Linux hosts |
| Unknown | Invalid config (warn_factor >= fail_factor) |
The message always reports the raw load and CPU count (e.g. Load: 1.24 (4 CPUs, warn threshold: 3.20)).
Reads /proc/loadavg from the Linux kernel. No special volume mounts are required — the kernel exposes /proc inside containers by default.
This check only works on Linux hosts. It returns Unknown on macOS or Windows Docker Desktop environments.
fail_factor of 1.5 means the check fails at load 1.5× CPU count regardless of whether the server has 2 or 64 cores.fail_factor values on servers with known burst workloads to reduce noise.warn_factor: 0.7, fail_factor: 1.0 for servers that should run comfortably under full load.