[BUG]: Docker Task Failing on Kubernetes 1.30 with Self-Hosted Agent Due to Docker Unavailability

[tamilvananv-conga]

What happened?

We have created a custom image using the official Microsoft agent image and registered it as a self-hosted agent in Kubernetes 1.30. However, when we use this agent label for any Docker task or container section to pull the image from the registry and run it as a container, it fails due to Docker unavailability. We understand that Docker has been deprecated, and all supported Kubernetes versions on AKS use containerd as the default runtime starting from Kubernetes 1.19.

Is there any plan to support containerd as the default runtime for these tasks, rather than Docker, in future releases of the Azure DevOps pipeline Docker task?

Versions

4.248.0 --

Environment type (Please select at least one enviroment where you face this issue)

• Self-Hosted
• Microsoft Hosted
• VMSS Pool
• Container

Azure DevOps Server type

dev.azure.com (formerly visualstudio.com)

Azure DevOps Server Version (if applicable)

No response

Operation system

Ubuntu

Version controll system

No response

Relevant log output

Task log

container:
image: ''
endpoint: ''

[theonlyames]

I had the same issue when running on an AKS cluster.

To get around it I set up a buildkit service on the cluster, installed the docker cli tools manually onto a custom image (along with other utilities my organization needed), configured a remote buildkit instance through docker cli, and modified the pipelines to use the remote buildkit instance. It's not perfect, but it can still build images fine even if it cannot actually run containers.

It's not a solution to the lack of containerd support, but it might help you.