Core Concepts
This page introduces the core objects and workflow used in Fleet Manager.
Why Teams Use Fleet Manager
As telemetry deployments grow, configuration drift becomes a common reliability risk. Teams often end up with:
- Different agent versions across hosts.
- Inconsistent parsing or routing rules.
- Slow and risky manual changes.
- Limited visibility into what is actually running.
Fleet Manager addresses this by giving operators a central way to define, deploy, and monitor Fluent Bit behaviour across a fleet.
In the observability domain, this follows a common fleet-management model: central policy management, controlled rollout rings, and fleet-wide status reporting.
Fleet Manager supports mixed deployment models. A single fleet can include:
- Kubernetes-based agents.
- Native Linux deployments on virtual machines (VMs) or bare metal.
- Edge or embedded deployments with intermittent connectivity.
- Windows and macOS endpoints deployed through mobile device management (MDM) workflows.
Fleet Manager also supports mixed agent channels in the same estate:
- Upstream Fluent Bit (OSS).
- Telemetry Forge Agent (commercial distribution).
This lets teams adopt commercial support where needed while keeping direct compatibility with OSS deployments.
Core Objects
Agent
A running Fluent Bit instance connected to Fleet Manager.
Group
A logical collection of agents, such as:
- Environment (
dev,staging,production) - Region (
us-east,eu-west) - Platform (
kubernetes,linux-vm,bare-metal,edge-embedded,windows-mdm,macos-mdm)
Configuration Revision
A versioned telemetry configuration that can be assigned to one or more groups.
Rollout
A controlled deployment of a configuration revision to selected agents or groups.
Typical User Journey
- Connect agents to Fleet Manager.
- Organise agents into groups.
- Create or import a Fluent Bit configuration.
- Validate in a low-risk environment.
- Roll out progressively to production.
- Monitor fleet status and version adoption.
Recommended Grouping Model
Start with clear, stable dimensions:
- First dimension: environment (
dev,staging,production) - Second dimension: platform (
kubernetes,linux-vm,windows-mdm,macos-mdm,edge-embedded) - Optional third dimension: region
This structure keeps targeting predictable during rollouts and incident response.
For larger estates, add a connectivity dimension such as always-online and intermittent so rollout and monitoring expectations match real operating conditions.