How to start docker container X86 on Rasberry PI 4 with arm64 architecture?

Question

I have Raspberry PI 4 with 8 GB memory. And I have x86 container. It is working on x86/amd64 servers correctly. I would like to port it to raspberry. It has ubuntu 22 released from Raspberry team. I read there is qemu-user-static, kvm and looks like x86 docker's container on arm64 are supported but I can't get any result. I can see my Raspberry supports kvm.

    root@ubuntu:/home/ubuntu# virt-host-validate
  QEMU: Checking if device /dev/kvm exists                                   : PASS
  QEMU: Checking if device /dev/kvm is accessible                            : PASS
  QEMU: Checking if device /dev/vhost-net exists                             : PASS
  QEMU: Checking if device /dev/net/tun exists                               : PASS
  QEMU: Checking for cgroup 'cpu' controller support                         : PASS
  QEMU: Checking for cgroup 'cpuacct' controller support                     : PASS
  QEMU: Checking for cgroup 'cpuset' controller support                      : PASS
  QEMU: Checking for cgroup 'memory' controller support                      : PASS
  QEMU: Checking for cgroup 'devices' controller support                     : PASS
  QEMU: Checking for cgroup 'blkio' controller support                       : PASS
  QEMU: Checking for device assignment IOMMU support                         : WARN (Unknown if this platform has IOMMU support)
  QEMU: Checking for secure guest support                                    : WARN (Unknown if this platform has Secure Guest support)
   LXC: Checking for Linux >= 2.6.26                                         : PASS
   LXC: Checking for namespace ipc                                           : PASS
   LXC: Checking for namespace mnt                                           : PASS
   LXC: Checking for namespace pid                                           : PASS
   LXC: Checking for namespace uts                                           : PASS
   LXC: Checking for namespace net                                           : PASS
   LXC: Checking for namespace user                                          : PASS
   LXC: Checking for cgroup 'cpu' controller support                         : PASS
   LXC: Checking for cgroup 'cpuacct' controller support                     : PASS
   LXC: Checking for cgroup 'cpuset' controller support                      : PASS
   LXC: Checking for cgroup 'memory' controller support                      : PASS
   LXC: Checking for cgroup 'devices' controller support                     : PASS
   LXC: Checking for cgroup 'freezer' controller support                     : FAIL (Enable 'freezer' in kernel Kconfig file or mount/enable cgroup controller in your system)
   LXC: Checking for cgroup 'blkio' controller support                       : PASS
   LXC: Checking if device /sys/fs/fuse/connections exists                   : PASS

Looks like I am doing something wrong but this technology is obviously not easy and there is no way to understand it quickly. Internet has not a of lot info but it is about other tasks. May be someone got this way and could push me to the right direction.

So usually I starts my container by this way:

docker run -d --privileged --restart unless-stopped --hostname mycontainer --net=host --name=mycontainer -it mycontainer:v1 /bin/bash /entrypoint.sh

and with installed qemu-user-static, qemu-system and all other qemu libraries it started but doesn't work.

I read the following info: Need to run it inside qemu container by following way:

docker run --rm -it --device=/dev/kvm:/dev/kvm --device=/dev/net/tun:/dev/net/tun     --cap-add NET_ADMIN -v /home/ubuntu/mycontainer_v1.xz:/image jkz0/qemu:latest

It gives the following error:

root@ubuntu:/home/ubuntu# docker run --rm -it     --device=/dev/kvm:/dev/kvm --device=/dev/net/tun:/dev/net/tun     --cap-add NET_ADMIN -v /home/ubuntu/readyImages/adapter2_2.xz:/image     jkz0/qemu:latest
WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested
udhcpd: started, v1.30.1
udhcpd: max_leases=235 is too big, setting to 1
udhcpd: can't open '/var/lib/misc/udhcpd.leases': No such file or directory
kvm version too old
qemu-system-x86_64: failed to initialize KVM: Function not implemented

Did someone touch this issue before? I need fresh ideas. Best regards, Dmitry

Answer 1

So, I don't know about docker – but docker is "just" a runtime for OCI containers (and frankly, I find it's sometimes too much, it does a lot that I don't need). The same "docker containers" can be run using podman (which even has a relatively compatible CLI), and podman supports running containers in a VM right out of the box; all you need is an x86_64 coreOS VM image; you will have to fiddle with the VM configuration a bit manually, podman does not (yet?) have command line options for that.

Small preparation necessary: I don't know which machines your qemu supports, so run qemu-system-x86_64 -cpu help | grep -v 'alias ', and pick one that firs your needs; I'll just use Nehalem-v2, you replace that with something sensible where it occurs below!

podman machine init \
      --image-path YOUR_DOWNLOADED_FEDORA_COREOS-qemu.x86_64.qcow2.xz_IMAGE \
      x86_64_machine

Edit $XDG_CONFIG_HOME/containers/podman/machine/qemu/x86_64_machine.json; find the "CmdLine" property, a list; in that list, find "-cpu"; the next line probably says "host", which is not what you want here; replace that with your desired qemu CPU model (Nehalem-v2 in my case). Make sure "-accel" is "kvm".

So now that this is set up correctly, configure podman to always use that by default. podman system connection default x86_64_machine does that.

Start the VM, podman machine start x86_64_machine; this might take a while on an Rpi!

After it was started successful, you can run your favorite docker containers; podman run --rm -it fedora:39 uname -a should pull a Fedora 39 container (x86_64!), and run uname -a in it; after that finishes, the container will be removed (--rm) again.

Answer 2

You just can't it's not possible as the architecture differs.

Two ways to do achieve this are :

• rebuild the container as said in comment. https://www.padok.fr/en/blog/docker-arm-architectures https://www.docker.com/blog/faster-multi-platform-builds-dockerfile-cross-compilation-guide/ Even if your container comes from a "real server" it should be easy to rebuild it. That's why docker is made for.

• run docker from a virtual machine with x86/amd64 emulation. But i'm afraid there is no way to do this easily, or even doing this at all from arm machine