Skip to main content

Agents & Scaling – Overview

This section explains how Jenkins executes work at scale using agents. Understanding agents is critical for performance, cost control, and stability.

Why Agents Matter​

Without proper agent strategy:

  • Builds queue endlessly
  • Controllers get overloaded
  • Costs increase
  • Pipelines become flaky

Agents allow Jenkins to scale horizontally.

Controller vs Agent (High-Level)​

  • Controller: Orchestrates jobs, schedules builds, manages state
  • Agents: Execute pipeline steps (builds, tests, deployments)

Controllers should not run heavy workloads.

What Is a Jenkins Agent?​

An agent is a machine (VM, container, pod, or bare metal) that:

  • Connects to the controller
  • Executes pipeline steps
  • Provides tools and runtime environments

Agent Types​

Common agent types:

  • Static agents (VMs / bare metal)
  • Docker-based agents
  • Kubernetes pod agents
  • Cloud-based ephemeral agents

Each has different trade-offs.

Executors​

  • Executors define how many builds an agent can run concurrently
  • More executors ≠ better performance
  • Executors compete for CPU, memory, disk

Proper sizing is critical.

Label-Based Scheduling​

Agents are selected using labels:

  • OS-specific
  • Tool-specific
  • Capability-specific

Labels control where workloads run.

Scaling Strategies​

  • Horizontal scaling (more agents)
  • Ephemeral agents (on-demand)
  • Workload isolation
  • Auto-scaling via cloud/Kubernetes

Common Scaling Problems​

  • Too many executors on one agent
  • Running builds on controller
  • No isolation between workloads
  • Static agents for bursty workloads

Security Considerations​

  • Agents execute arbitrary code
  • Limit agent permissions
  • Use ephemeral agents for untrusted code
  • Separate prod and non-prod agents

Interview Focus Areas​

  • Controller vs agent responsibilities
  • Executors vs agents
  • Horizontal vs vertical scaling