Grafana Kubernetes Monitoring Helm Charts

Metric Collection Optimization

1. Disable OpenCost if not needed.
2. Disable collection of tmpfs filesystem, ipvs, and veth device metrics.
3. Retain most metrics from node_exporter.

metrics:
  cost:
    enabled: false
  node-exporter:
    allowList: []
    extraMetricRelabelingRules: |
      rule {
        source_labels = ["__name__"]
        regex = "node_scrape_collector_.+"
        action = "drop"
      }
prometheus-node-exporter:
  extraArgs:
    - --no-collector.ipvs
    - --collector.netclass.ignored-devices=^(veth.*|cali.*|[a-f0-9]{15})$
    - --collector.netdev.device-exclude=^(veth.*|cali.*|[a-f0-9]{15})$
    # Add tmpfs to defaults
    - --collector.filesystem.fs-types-exclude=^(autofs|binfmt_misc|bpf|cgroup2?|configfs|debugfs|devpts|devtmpfs|tmpfs|fusectl|hugetlbfs|iso9660|mqueue|nsfs|overlay|proc|procfs|pstore|rpc_pipefs|securityfs|selinuxfs|squashfs|sysfs|tracefs)$
opencost:
  enabled: false