NodeHighCPUUsage

Host CPU usage is above 90%. Free rule (MIT): the complete runbook and rule definition are below.


Severity	warning
Pending (`for:`)	30m (warning)
Domain	Hosts (node_exporter)
Requires	node_exporter 1.x (default collectors; systemd and hwmon noted per alert)
Rule file	`rules/node-exporter.rules.yml` (group `ftw.node-exporter.alerts`)

Meaning

Average CPU utilisation across all cores (everything except idle) has been above 90% for 30 minutes. The long for: filters out normal bursts.

Impact

Higher latency for everything on the host; the kubelet and system daemons may be starved.

Diagnosis

Break down by mode (user, system, iowait, steal):

sum by (mode) (rate(node_cpu_seconds_total{instance="<instance>", mode!="idle"}[5m]))

High steal: noisy neighbour / burstable instance out of credits.
On the host: top -o %CPU, pidstat 5 3. Kubernetes: kubectl top pods -A --sort-by=cpu | head.

Mitigation

Scale out the workload or move it to a bigger instance.
Burstable instances (t3/t4g, B-series): switch to a non-burstable type or enable unlimited mode.
Set CPU limits/requests on runaway pods.

Escalation

warning: Notify the owning team’s alert channel. Acknowledge within business hours; create a ticket if it cannot be resolved the same day. Escalate to critical handling if it trends toward the critical condition.

Owned by whoever operates the host (platform team for Kubernetes nodes, service owner for standalone VMs).

NodeExporterDown: Host or node_exporter is down.
NodeFilesystemSpaceFillingUp: Filesystem is predicted to run out of space within 24 hours.
NodeFilesystemAlmostOutOfSpace: Filesystem has less than 10% space left.
NodeMemoryHighUtilization: Host memory utilisation is above 90%.
NodeFilesystemAlmostOutOfFiles: Filesystem has less than 10% inodes left.

Rule definition

From rules/node-exporter.rules.yml in the free repository (MIT). Unit tests for it are in tests/.

- alert: NodeHighCPUUsage
  expr: 100 * (1 - avg without (cpu, mode) (rate(node_cpu_seconds_total{mode="idle"}[5m]))) > 90
  for: 30m
  labels:
    severity: warning
  annotations:
    summary: Host CPU usage is above 90%.
    description: CPU usage on {{ $labels.instance }} has been {{ $value | printf "%.1f" }}% for more than 30m.
    runbook_url: runbooks/node-exporter/NodeHighCPUUsage.md