Liqo can adjust the placement of the pods of your deployments with the same flexibility you have in “vanilla” Kubernetes. To do so, two straightforward steps are required:
First, you can remove the old resource:
export KUBECONFIG=$KUBECONFIG_1
kubectl delete namespaceoffloadings offloading -n liqo-test
Deleting the NamespaceOffloading object, you will delete all the remote namespaces previously created and their content. You have to be sure that everything you deployed remotely is no longer needed.
You can check that the remote namespace is correctly deleted from the cluster-3:
export KUBECONFIG=$KUBECONFIG_3
kubectl get namespaces
Now, you can create a new configuration. For instance, you can decide to have:
The NamespaceOffloading resource will be like this:
export KUBECONFIG=$KUBECONFIG_1
cat << "EOF" | kubectl apply -f - -n liqo-test
apiVersion: offloading.liqo.io/v1alpha1
kind: NamespaceOffloading
metadata:
name: offloading
namespace: liqo-test
spec:
namespaceMappingStrategy: EnforceSameName
podOffloadingStrategy: Remote
EOF
It is not necessary to specify a ClusterSelector field to select all available clusters. This is the standard behavior of its default value
There should be two remote namespaces: one inside the cluster-2 and the other inside cluster-3:
export KUBECONFIG=$KUBECONFIG_2
kubectl get namespaces liqo-test
export KUBECONFIG=$KUBECONFIG_3
kubectl get namespaces liqo-test
Both of those commands should have an output like:
NAME STATUS AGE
liqo-test Active 40s
Once the new namespaces have been created, you can find out how to contact remote pods from the home-cluster