Docker "cgroups: memory cgroup not supported on this system" in CI

Docker tried to apply a memory limit (or just start a container) and found no memory cgroup controller available. The kernel or the nested environment does not expose the memory controller Docker needs.

What this error means

A docker run (especially with --memory) fails with cgroups: memory cgroup not supported on this system. Containers without memory limits may run, but any memory-controlled container cannot start.

docker run output

docker: Error response from daemon: cgroups: memory cgroup not supported on this
system: unknown.

Common causes

Memory cgroup controller not enabled in the kernel

Some minimal or embedded kernels (and certain ARM SBC images) ship with the memory controller disabled, so Docker cannot enforce memory limits.

Controller not delegated to a nested environment

Inside Docker-in-Docker or a constrained runner, the memory controller may not be delegated to the cgroup the daemon runs in, so it appears unsupported.

cgroup v1 with memory disabled at boot

A cgroup_enable=memory boot parameter missing on a cgroup v1 host leaves the memory controller off.

How to fix it

Enable the memory controller at boot (cgroup v1)

Add the kernel boot parameters and reboot so the memory controller is active.

Terminal

# add to the kernel cmdline (e.g. /boot/cmdline.txt or GRUB):
cgroup_enable=memory cgroup_memory=1
# then reboot the host

Delegate the controller to nested Docker

For Docker-in-Docker, run the daemon with cgroup delegation / sufficient privilege so the memory controller is available.
Prefer the host Docker daemon over deeply nested runtimes.
If you cannot enable it, run containers without --memory (no memory limit) as a stopgap.

How to prevent it

Use runner kernels/images with the memory cgroup controller enabled.
Delegate cgroup controllers when running nested Docker.
Verify docker info does not warn about a missing memory controller.