I am using colima for my local develoment on MacOS. Because of emulation layer, I can run both linux/arm64 and linux/amd64 docker images:
~ docker run --platform linux/amd64 --rm busybox:latest echo -n "life is good" Mon Feb 9 15:24:17 2026
Unable to find image 'busybox:latest' locally
latest: Pulling from library/busybox
61dfb50712f5: Pull complete
Digest: sha256:b3255e7dfbcd10cb367af0d409747d511aeb66dfac98cf30e97e87e4207dd76f
Status: Downloaded newer image for busybox:latest
life is good⏎
~ docker inspect busybox:latest | grep Architecture 5.5s Mon Feb 9 15:24:57 2026
"Architecture": "amd64",
~ Mon Feb 9 15:25:08 2026
~ docker image rm busybox:latest Mon Feb 9 15:25:10 2026
Untagged: busybox:latest
Deleted: sha256:b3255e7dfbcd10cb367af0d409747d511aeb66dfac98cf30e97e87e4207dd76f
~ docker run --platform linux/arm64 --rm busybox:latest echo -n "life is good" Mon Feb 9 15:25:14 2026
Unable to find image 'busybox:latest' locally
latest: Pulling from library/busybox
b85757a5ca1a: Pull complete
Digest: sha256:b3255e7dfbcd10cb367af0d409747d511aeb66dfac98cf30e97e87e4207dd76f
Status: Downloaded newer image for busybox:latest
life is good⏎
~ docker inspect busybox:latest | grep Architecture 4011ms Mon Feb 9 15:25:26 2026
"Architecture": "arm64",
Colima can also provision a k8s cluster (k3s), which I am now playing with. It’s a one node cluster.
The issue is however that when I try to deploy a pod referencing linux/amd64 architecture pod it fails on image pull:
spec.containers{ibmmq}: Failed to pull image “ibmcom/mq:latest”: Error response from daemon: no matching manifest for linux/arm64/v8 in the manifest list entries: no match for platform in manifest: not found
And I understand that this is because the k8s cluster is configured with arm64 arch:
~ k get nodes 212ms Mon Feb 9 15:29:01 2026
NAME STATUS ROLES AGE VERSION
colima Ready control-plane,master 2d21h v1.33.4+k3s1
~ k describe node colima | grep -i architecture Mon Feb 9 15:29:04 2026
Architecture: arm64
However, since technically amd64 containers can be run, I wonder if there’s any way to configure one-node k8s cluster to handle both architectures? If that won’t fly, then I guess I should still have an option to run another node with different architecture and let the scheduler deal with this…. I need a push in right direction. Thx.