What does Linux load average mean, and how is it different from CPU percentage?
Linux load average includes tasks that are runnable and tasks in uninterruptible sleep, commonly waiting on I/O. It is not a CPU-utilisation percentage. Compare it with effective CPU capacity and current task states; the same load value can describe very different performance conditions.
Before you start
- An authorised SSH session on the Linux server.
Instructions
1. Read the trend and CPU count
The three averages smooth different time windows. The CPU count is useful context but may not reflect every quota or scheduling restriction applied to a container or service.
uptime
nprocA rising one-minute average suggests recent pressure. Do not apply a universal good-or-bad threshold without checking the workload and limits.
2. Inspect runnable and blocked work
Sample during the symptom. The first vmstat report covers historical averages for several counters; subsequent reports describe each sampling interval.
vmstat 1 5Inspect r for runnable tasks, b for blocked tasks, and CPU/I/O-related columns together. High load with idle CPU can indicate waits rather than a need for more CPU cores.
Check the result
- The explanation connects load to task state, effective CPU allocation and measured application response time.
Troubleshooting
- Load is high but CPU utilisation is low
- Look for blocked I/O, storage latency or other uninterruptible waits.
- The server has many CPUs but still queues work
- Check CPU affinity, cgroup quota, single-threaded bottlenecks and whether the workload can use the available parallelism.
Official references
Use the documentation for your installed version when a flag, service name or platform feature differs.