Watchdog: Load Average
Monitors the one-minute system load relative to the CPU count using factor thresholds. Reads /proc/loadavg. Type load_average, runner Agent. Fields, statuses, messages.
Watchdog: Load Average
The check compares the system load of the last minute with the number of CPUs. It reads /proc/loadavg and counts the processor entries in /proc/cpuinfo.
| Property | Value |
|---|---|
| Type identifier | load_average |
| Runner | Agent |
| Target in the email | none |
Fields
| Field | Key | Default | Meaning |
|---|---|---|---|
| Warn factor | warn_factor |
0.8 |
WARN when the load reaches at least this factor times the CPU count |
| Fail factor | fail_factor |
1.5 |
FAIL when the load reaches at least this factor times the CPU count |
The warn factor must be smaller than the fail factor.
Behaviour
Example with 4 CPUs and the defaults: warn threshold 3.2, fail threshold 6.0.
| Situation | Status | Message |
|---|---|---|
| Load below the warn threshold | OK |
Load: <load> (<n> CPUs, warn threshold: <threshold>) |
| Load at or above the warn threshold | WARN |
Load: <load> (<n> CPUs, warn threshold: <threshold>) |
| Load at or above the fail threshold | FAIL |
Load: <load> (<n> CPUs, fail threshold: <threshold>) |
| Warn factor not smaller than fail factor | UNKNOWN |
Invalid config: warn_factor (<w>) must be less than fail_factor (<f>) |
/proc/loadavg not readable |
UNKNOWN |
/proc/loadavg unavailable — check runs only on Linux hosts |
| Read error | FAIL |
Cannot read /proc/loadavg |
The check needs no additional mounts or privileges.