Loading...

Build image with containerised self-hosted Azure DevOps agent and private Azure Container Registry

Build image with containerised self-hosted Azure DevOps agent and private Azure Container Registry

Introduction

Containerization has revolutionized the way we deploy and manage software. By packaging applications and their dependencies into a container, we can ensure that our software will run consistently across different environments. This is particularly useful when it comes to setting up self-hosted agents for Azure Pipelines.

A self-hosted agent is an agent that you set up and manage on your own to run jobs. You can use self-hosted agents in Azure Pipelines or Git Hub Actions. Self-hosted agents give you more control to install dependent software needed for your builds and deployments.

 

Why use containerised self-hosted agent

As we have seen the benefit of running the workload in container including agility, portability, high density and more. By running your self-hosted agent in a container, you can take advantage of the benefits of containerization.

  • Agility: you can easily spin up new instances of your agent as needed, without having to worry about manually installing dependencies or configuring the environment. This can save you time and effort.
  • Portable: Each engineering team can have autonomy and their own build agent in manage their pipeline. This help ensure that team’s builds and deployments are consistent.
  • High density: Running your self-hosted agent in a container can make it easier to manage your infrastructure. Since containers are lightweight and portable, you can easily move your agents between different hosts as needed. This can help you optimize resource usage and reduce costs.

 

What is the challenge of using containerised self-hosted agent?

As part of CI/CD building workload container images is a key capability. It is possible to run Docker tasks on containerised selfhost agent however it does require access to the docker daemon at host level. This has critical security concerns.

A solution is to use Azure Container Registry (ACR) Task. With ACR Tasks, you can use Azure Container Registry to build container images for various platforms in the cloud. ACR Tasks also lets you automate the process of updating your Docker containers with OS and framework patches.

By default, Azure container registry accepts connections over the internet from hosts on any network. This opens an attack surface, and it is not normally in acceptable in enterprise environment. ACR Task use an agent running on Azure and required appropriated permission to able communicate with the Azure Container registry.

 

Proposed solution

There are a wide range of options to run containerized self-hosted agents on Azure from Azure Kubernetes Service to Azure Container App. This proof of concept Demonstrates the building of container images using self-hosted agent using Azure Container Instances with Azure Container Registry with private access only.  

 

 

Architecture:

 

Architecture DiagramArchitecture Diagram

 

 

Proof of concept:

The proof of concept is all written in Bicep and can be found at GitHub repo (https://github.com/duongthaiha/containerised-selfhost-agent,). It provision the following:

  • Provision a virtual network two subnets, one for Azure Container Instance and the other for Azure Container Registry
  • Private DNS zone and entries
  • Provision Azure Container Registry with private endpoint
  • Disable public access to Azure Container Registry
  • Provision Azure DevOps self-hosted agent in Azure Container Instances
  • Provision an Azure Container Registry Task to build a container image.
  • Custom role definition and role assignment following minimum privilege principles.

 

Prerequisite:

ADO Organization and a PAT token for more details permission required please refer to Run a self-hosted agent in Docker - Azure Pipelines | Microsoft Learn

GitHub Repo and Token for more detail permission required please refer to Managing your personal access tokens - GitHub Docs

 

Deployment Steps:

1) Update the parameter file to include the correct parameters:

Ha_Duong_1-1693956972719.png

 

2) Run the command to deploy the resource:

Ha_Duong_2-1693956972723.png

 

3) To start the image build process, we can connect to self-host agent using Connect capability in Azure Container Registry

Ha_Duong_3-1693956972731.png

 

4) Run command to trigger build task.

Ha_Duong_4-1693956972733.png

 

 

 

Ha_Duong_6-1693956972750.png

 

5 ) From figure above you can see that self-host agent has successfully build the image and uploaded the image into the Azure Container Registry

 

Summary:

If you are considering setting up a self-hosted agent for Azure Pipelines, it’s definitely worth considering running it in a container and using an Azure Container Registry with public access disabled for building images with ACR Tasks.

 

 

Related Content:

Run a self-hosted agent in Docker - Azure Pipelines | Microsoft Learn

Managing your personal access tokens - GitHub Docs

Run a self-hosted agent in Docker - Azure Pipelines | Microsoft Learn

ACR Tasks overview - Azure Container Registry | Microsoft Learn

 

Published on:

Learn more
Azure Architecture Blog articles
Azure Architecture Blog articles

Azure Architecture Blog articles

Share post:

Related posts

Accelerate Your Cosmos DB Infrastructure with GitHub Copilot CLI and Azure Cosmos DB Agent Kit

Modern infrastructure work is increasingly agent driven, but only if your AI actually understands the platform you’re deploying. This guide sh...

9 hours ago

SharePoint: Migrate the Maps web part to Azure Maps

The SharePoint Maps web part will migrate from Bing Maps to Azure Maps starting March 2026, completing by mid-April. Key changes include renam...

17 hours ago

Azure Cosmos DB TV Recap: Supercharging AI Agents with the Azure Cosmos DB MCP Toolkit (Ep. 110)

In Episode 110 of Azure Cosmos DB TV, host Mark Brown is joined by Sajeetharan Sinnathurai to explore how the Azure Cosmos DB MCP Toolkit is c...

5 days ago

Introducing the Azure Cosmos DB Agent Kit: Your AI Pair Programmer Just Got Smarter

The Azure Cosmos DB Agent Kit is an open-source collection of skills that teaches your AI coding assistant (GitHub Copilot, Claude Code, Gemin...

6 days ago

Introducing Markers in Azure Maps for Power BI

We’re announcing a powerful new capability in the Azure Maps Visual for Power BI: Markers. This feature makes it easier than ever for organiza...

8 days ago
Stay up to date with latest Microsoft Dynamics 365 and Power Platform news!
* Yes, I agree to the privacy policy