Why kind load docker-image <docker_image_name> would load to control node?

[xihajun]

kind load docker-image xxx:latest -n testgpu

I guess we don't need to load to control-plane? Image: "xxx" with ID "sha256:747d6cf14415fb5f48f8dbec4ba28dcb4b8339347540e12e57aac3438487a738" not yet present on node "testgpu-control-plane", loading...

Image: "xxx" with ID "sha256:747d6cf14415fb5f48f8dbec4ba28dcb4b8339347540e12e57aac3438487a738" not yet present on node "testgpu-control-plane", loading...
Image: "xxx" with ID "sha256:747d6cf14415fb5f48f8dbec4ba28dcb4b8339347540e12e57aac3438487a738" not yet present on node "testgpu-worker", loading...

Details:

Image: "xxx" with ID "sha256:747d6cf14415fb5f48f8dbec4ba28dcb4b8339347540e12e57aac3438487a738" not yet present on node "testgpu-control-plane", loading...
Image: "xxx" with ID "sha256:747d6cf14415fb5f48f8dbec4ba28dcb4b8339347540e12e57aac3438487a738" not yet present on node "testgpu-worker", loading...
ERROR: failed to load image: command "docker exec --privileged -i testgpu-control-plane ctr --namespace=k8s.io images import --all-platforms --digests --snapshotter=overlayfs -" failed with error: exit status 1
Command Output: unpacking docker.io/library/xxx:latest (sha256:b1834fcf8afb3d81cd58d6c7058b33f5ad1d04d8d9f9fd8ca5040ed6271a2b7f)...time="2023-12-28T16:58:16Z" level=info msg="apply failure, attempting cleanup" error="failed to extract layer sha256:b7685fd66665733a9b43bf7f3052b590af89af4eccdb69d4bd40887fe5044b9f: write /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/109/fs/usr/local/lib/python3.10/dist-packages/torch/lib/libtorch_cuda.so: no space left on device: unknown" key="extract-44196102-klKV sha256:aa10e0c10b2bdf3648d57d23b7012afc73d583005aa50b1218c11b643a9ef047"
time="2023-12-28T16:58:16Z" level=warning msg="extraction snapshot removal failed" error="write /var/lib/containerd/io.containerd.metadata.v1.bolt/meta.db: no space left on device: unknown" key="extract-44196102-klKV sha256:aa10e0c10b2bdf3648d57d23b7012afc73d583005aa50b1218c11b643a9ef047"
ctr: failed to extract layer sha256:b7685fd66665733a9b43bf7f3052b590af89af4eccdb69d4bd40887fe5044b9f: write /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/109/fs/usr/local/lib/python3.10/dist-packages/torch/lib/libtorch_cuda.so: no space left on device: unknown

[BenTheElder]

You could schedule workloads to those nodes as long as you set tolerations (which may be the case for e.g. https://kind.sigs.k8s.io/docs/user/ingress/ or other system components), you can set which nodes you load to using the flags.