Docker not available on most recent al2 1.31 image

[kmentch]

Environment:

• AWS Region:
• Instance Type(s):
• EKS Platform version (use aws eks describe-cluster --name <name> --query cluster.platformVersion):
• Kubernetes version (use aws eks describe-cluster --name <name> --query cluster.version):
• AMI Version:
• Kernel (e.g. uname -a):
• Release information (run cat /etc/eks/release on a node):

$ uname -a
Linux ip-10-6-188-68.ec2.internal 5.10.239-236.958.amzn2.x86_64 #1 SMP Tue Jul 15 10:36:12 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux

AMI - ami-0154a7c55b9999c7f

$ cat /etc/os-release
NAME="Amazon Linux"
VERSION="2"
ID="amzn"
ID_LIKE="centos rhel fedora"
VERSION_ID="2"
PRETTY_NAME="Amazon Linux 2"
ANSI_COLOR="0;33"
CPE_NAME="cpe:2.3:o:amazon:amazon_linux:2"
HOME_URL="https://amazonlinux.com/"
SUPPORT_END="2026-06-30"

We had a node upgrade to this instance version this morning and in our script to bring it online it installs docker, docker is not available to install as a package in this image:

$ sudo yum search docker
Loaded plugins: priorities, update-motd, versionlock
=========================================================================================================================== N/S matched: docker ============================================================================================================================
pcp-pmda-docker.x86_64 : Performance Co-Pilot (PCP) metrics from the Docker daemon
amazon-ecr-credential-helper.x86_64 : Amazon ECR Docker Credential Helper
nerdctl.x86_64 : nerdctl is a Docker-compatible CLI for containerd.
oci-add-hooks.x86_64 : Injects OCI hooks as a Docker runtime

  Name and summary matches only, use "search all" for everything.

Is there something wrong? or did docker get removed in some way.

[Halama]

Hi,
we hit a same issue today. We are using AL2 and install docker:

amazon-linux-extras enable docker
yum install -y docker-25.0.3-1.amzn2.0.1

It works on amazon-eks-node-1.32-v20250804 but started failing on amazon-eks-node-1.32-v20250807.

After some investigation I've noticed in /etc/yum.conf:

# PUT YOUR REPOS HERE OR IN separate files named file.repo
# in /etc/yum.repos.d
exclude=containerd*,docker*

I've tried to ugprade to AL2023 version v20250807 and the issue is similar:

Was it introduced in #2339 ?
thanks

[bryantbiggs]

Why do you need the docker daemon on the host? If building images, there are better alternatives that are daemonless

[mselim00]

Yes this is from #2339, that change added the excludes to the dnf conf to maintain the containerd “version lock” even when it is installed directly as a binary (i.e. preventing dnf from overwriting it). The exclude can be removed if necessary (e.g with a sed) by user discretion, but that's discouraged because many operations may overwrite the provided (and tested) version of containerd

+1 to Bryant’s comment though, consider an alternative like nerdctl, which is already installed

[kmentch]

When working with tooling that was inherited to use docker and has many pieces integrated with it, its not just as simple as replacing it quickly. The fact is this was working and now is not without any notice is the problem.

[bryantbiggs]

AL2 is quickly reaching end of support, what about switching to AL2023 since it appears you could still install docker on that OS

[Halama]

Unfortunately the issue is same in AL2023 in latest release, details - #2363 (comment)

[cartermckinnon]

We're aware of this issue and are working on a solution.

[pen-pal]

the issue occurs also while installing aws-nitro-enclave-cli on AL2023 EKS Nodes
aws/aws-nitro-enclaves-cli#702 (comment)

[cartermckinnon]

@pen-pal yes this is discussed in #2367.

[pen-pal]

is there any alternative for use for the time being on al2023 ami or al2 ami till the issue is resolved?
@cartermckinnon

[bryantbiggs]

@pen-pal use an AMI release version prior to v20250807

[kmentch]

I see this has a fix thats been merged, any idea when it will be released? @wwvela