What would you like to be cleaned:
Looking at the function to load images:
|
# $1 cluster |
|
# $2 image |
|
function cluster_kind_load_image { |
|
# check if the command to get worker nodes could succeeded |
|
if ! $KIND get nodes --name "$1" > /dev/null 2>&1; then |
|
echo "Failed to retrieve nodes for cluster '$1'." |
|
return 1 |
|
fi |
|
# filter out 'control-plane' node, use only worker nodes to load image |
|
worker_nodes=$($KIND get nodes --name "$1" | grep -v 'control-plane') |
|
if [[ -n "$worker_nodes" ]]; then |
|
# Use docker save + ctr import directly to avoid the --all-platforms |
|
# issue with multi-arch images in DinD environments. |
|
# See: https://github.com/kubernetes-sigs/kind/issues/3795 |
|
echo "Loading image '$2' to cluster '$1'" |
|
while IFS= read -r node; do |
|
echo " Loading image to node: $node" |
|
if ! docker save "$2" | docker exec -i "$node" ctr --namespace=k8s.io images import --digests --snapshotter=overlayfs -; then |
|
echo "Failed to load image '$2' to node '$node'" |
|
return 1 |
|
fi |
|
done <<< "$worker_nodes" |
|
fi |
|
} |
I think we could load the images in parallel.
This will require calling docker save "$2" prior to the parallel loading.
Why is this needed:
To improve time to setup the cluster for testing. This especially will matter for TAS e2e tests which create 8 worker nodes.
