Monitoring CPU, Memory, and Disk Usage with Prometheus and Grafana

1. CPU Usage

Start by calculating the per‑second rate for each CPU mode using PromQL's irate function on the node_cpu_seconds_total metric. Example query:

irate(node_cpu_seconds_total{job="node"}[5m])

To get an average per instance, wrap the irate call in an avg aggregation with a by (instance) clause:

avg(irate(node_cpu_seconds_total{job="node"}[5m])) by (instance)

To focus on idle time, add the mode="idle" label and compute the idle percentage:

avg(irate(node_cpu_seconds_total{job="node",mode="idle"}[5m])) by (instance) * 100

Subtract this from 100 to obtain the active CPU usage percentage:

100 - avg(irate(node_cpu_seconds_total{job="node",mode="idle"}[5m])) by (instance) * 100

The result is a metric per host showing the average CPU usage over a 5‑minute window.

2. Memory Usage

Identify the relevant node_memory metrics:

node_memory_MemTotal_bytes – total memory on the host

node_memory_MemFree_bytes – free memory

node_memory_Buffers_bytes_bytes – memory in buffers

node_memory_Cached_bytes_bytes – memory in page cache

All values are in bytes. Compute memory usage percentage with the following expression:

(node_memory_MemTotal_bytes-(node_memory_MemFree_bytes+node_memory_Cached_bytes+node_memory_Buffers_bytes)) / node_memory_MemTotal_bytes * 100

3. Disk Usage

For disk monitoring, use the node_filesystem metrics. The size metric node_filesystem_size_bytes reports the total capacity of each mounted filesystem. To calculate the used percentage for the root filesystem ( / ), apply a similar formula as for memory, adding a mountpoint="/" label:

(node_filesystem_size_bytes{mountpoint="/"} - node_filesystem_free_bytes{mountpoint="/"}) / node_filesystem_size_bytes{mountpoint="/"} * 100

Additional Notes

Grafana can be localized to Chinese and supports importing monitoring templates for services such as MySQL and Redis.

Related Articles

Two Ways to Deploy Prometheus

Deploy Prometheus + Grafana

Prometheus + Alertmanager Email Alert Configuration

Please share and follow the DevOps Operations Group