Host Metrics (OpenTelemetry)

Host Metrics

The OpenTelemetry hostmetrics receiver collects detailed system-level metrics from your hosts, providing comprehensive visibility into resource utilization and system performance.

Collected Metrics

The hostmetrics receiver collects:

• CPU metrics - Utilization across all cores and states
• Memory metrics - Usage, available memory, and utilization percentages
• Disk I/O metrics - Read/write operations and throughput
• Filesystem metrics - Usage and available space per mount point
• Network metrics - Interface I/O, packets, errors, and TCP connections
• Process metrics - Per-process CPU, memory, and disk I/O
• System load - Load averages and pressure metrics
• Paging/Swap - Page faults and swap usage

Setup

The OpenTelemetry hostmetrics receiver collects system-level metrics from your hosts. This guide shows you how to configure the receiver in your OpenTelemetry Collector to send metrics to Dash0.

Prerequisites

You need an OpenTelemetry Collector running with access to Dash0. We recommend using one of these deployment methods:

• Dash0 operator for Kubernetes (recommended)
• OpenTelemetry Operator for Kubernetes
• OpenTelemetry Collector Helm Chart

Configuration

Add the hostmetrics receiver to your OpenTelemetry Collector configuration:

Basic Configuration

This example enables all available scrapers with default settings:

receivers:
  hostmetrics:
    collection_interval: 60s
    scrapers:
      cpu:
      disk:
      filesystem:
      load:
      memory:
      network:
      paging:
      processes:
      process:

processors:
  batch:

exporters:
  otlp/dash0:
    endpoint: ${DASH0_ENDPOINT}
    headers:
      Authorization: "Bearer ${DASH0_AUTH_TOKEN}"

service:
  pipelines:
    metrics:
      receivers: [hostmetrics]
      processors: [batch]
      exporters: [otlp/dash0]

Kubernetes Configuration

When running in Kubernetes, additional configuration is needed for complete host metrics collection:

Enabling Utilization Metrics

To get utilization ratio metrics (e.g., system.memory.utilization), enable them explicitly:

receivers:
  hostmetrics:
    root_path: /hostfs
    collection_interval: 30s
    scrapers:
      cpu:
        metrics:
          system.cpu.utilization:
            enabled: true
      memory:
        metrics:
          system.memory.utilization:
            enabled: true
      filesystem:
        metrics:
          system.filesystem.utilization:
            enabled: true

Adding Hostname Labels

To add hostname information to your metrics, use the resource detection processor:

processors:
  resourcedetection:
    detectors: [env, system]
    system:
      hostname_sources: ["os"]

service:
  pipelines:
    metrics:
      receivers: [hostmetrics]
      processors: [resourcedetection, batch]
      exporters: [otlp/dash0]

Required Permissions

The collector needs specific permissions to access host metrics:

• Volume Mounts: Mount the host filesystem at /hostfs with read-only access
• Security Context: Run with appropriate capabilities (SYS_PTRACE, DAC_READ_SEARCH)
• Network Access: Use hostNetwork: true for complete network interface metrics

For a complete working example with all necessary Kubernetes configurations, see the Dash0 hostmetrics example.

Filtering Metrics

You can include or exclude specific metrics:

receivers:
  hostmetrics:
    scrapers:
      cpu:
        metrics:
          system.cpu.time:
            enabled: false # Disable this metric
          system.cpu.utilization:
            enabled: true # Enable this metric

Verify the Setup

1. Check that metrics are being collected by inspecting the logs of your OpenTelemetry Collector.

2. Verify metrics in Dash0

   • Search for metrics starting with system. (e.g., system.cpu.utilization, system.memory.usage)

Related Links

• OpenTelemetry hostmetrics receiver documentation
• OpenTelemetry Collector documentation
• Dash0 documentation
• Check out this local example for furhter experimentation