Ops Agent

Overview

• Ops Agent = single agent for metrics + logs on Compute Engine.

  Uses Fluent Bit (logs) + OpenTelemetry Collector (metrics).

• Why: hypervisor can’t see in-guest stuff (e.g., memory, processes), so agent fills the gap.

• Installs: per-VM, Agent Policy, or automation (Ansible/Chef/Puppet/Terraform).

• Also collects Prometheus-format custom metrics; hooks into Managed Service for Prometheus.

• Non-VM managed services (App Engine, Cloud Run, GKE 기본 옵션 등) → agent not needed.

What is Ops Agent?

• Collects telemetry from inside the VM (not hypervisor).
• Combines Logging + Monitoring in one. High throughput logs, OTEL metrics.
• 3rd-party integrations: Apache, MySQL, Oracle DB, SAP HANA, NGINX… (dashboard templates available).
• OS: CentOS, Ubuntu, Windows 등 주요 OS 지원.

Why Ops Agent?

• Hypervisor only knows allocated memory, not used memory → no mem charts without agent.
• Extra goodies beyond default GCE 80+ metrics:
  • Process-level visibility, richer CPU/Disk/Network details.
  • 3rd-party app metrics/logs out of the box.
  • Custom/Prometheus metrics ingestion.
  • 설치 후 추가 설정 없이 기본 모니터링 동작 (대부분의 케이스).

Install options

1. Per-VM (Console/CLI) — quick & manual.
2. Agent Policy — fleet 관리(자동 설치/업그레이드).
3. Automation — Ansible/Chef/Puppet/Terraform.

Install via Agent Policy (policy 예시)