Option to Ingest Only Host Metrics on Instana Agents (Disable All Sensors)

Summary

Provide a configuration option to allow Instana Agents to collect only host-level metrics (CPU, memory, disk, network, etc.) while disabling all automatic sensors and application monitoring.

This mode should be configurable either:

• via Environment Variable, or

• as a selectable option during the Agent installation process

Problem Statement

Instana agents automatically activate sensors to monitor applications, services, runtimes, and infrastructure components detected on the host.

In some environments, users only need basic infrastructure monitoring (host metrics) without enabling application or technology sensors.

Current limitations:

• Sensors need to be manually deactivated in the configuration.

• Disabling sensors individually is cumbersome and not scalable.

• There is no simple host-only monitoring mode.

This creates challenges in environments such as:

• Cost-optimization

• Baseline infrastructure monitoring

• Hosts where application monitoring is not allowed

Proposed Solution

Introduce a Host Metrics Only Mode for the Instana Agent.

When enabled:

• The agent collects only host-level metrics:

  • CPU

  • Memory

  • Disk

  • Network

  • OS metrics

• All technology/application sensors are disabled

• No automatic sensor discovery occurs.

Configuration Options

Option 1 – Environment Variable

Example:

INSTANA_AGENT_HOST_MONITORING_ONLY=true

or

INSTANA_AGENT_DISABLE_SENSORS=true

Expected Behavior

When enabled:

• Agent starts normally.

• Only host infrastructure metrics are collected and ingested to the Instana backend.

• No sensors are activated for:

  • Databases

  • Middleware

  • Containers

  • Runtimes

  • Applications

Benefits

• Simplified infrastructure-only monitoring

• Reduced resource consumption

• Reduced sensor management complexity

• Better support for restricted or hardened environments

• Easier large-scale host deployments

Example Use Cases

1. Security-sensitive hosts

   • Application inspection is not permitted.

2. Cost control

   • Only infrastructure monitoring is required.

3. Baseline monitoring

   • Teams want simple host health metrics without deep tracing.

4. Pre-production environments

   • Lightweight monitoring before enabling full observability.