Skip to content

create.md

Latest commit

 

History

History
134 lines (100 loc) · 4.9 KB

create.md

File metadata and controls

134 lines (100 loc) · 4.9 KB

Creating Liquid Metal clusters

If you have provisioned your Liquid Metal systems you can now create a cluster.

Configuration

Configure your cluster with your preferred settings. The following fields are required:

export CLUSTER_NAME=mvm-test
export WORKER_MACHINE_COUNT=1

Select an IP from your VLAN IP block range to use for the control plane address. (For example, I am using a 192.168.10.0/25 CIDR and have set aside 192.168.10.26 to 192.168.10.126 for my MicroVMs, so I can set 192.168.10.25 as my cluster endpoint address. This is a naive configuration for the purposes of this demonstration.):

export CONTROL_PLANE_VIP=<ADDRESS>

Set the HOST_ENDPOINT to the address that the flintlockd server is running on. (This is the one you configured when provisioning your host device.)

export HOST_ENDPOINT=<FLINTLOCK_ADDRESS>:9090

Note: that the below template generation will use the HOST_ENDPOINT var to add just one bare-metal host for static placement. If you created multiple hosts in the previous steps of the tutorial you will need to edit the cluster.yaml which will be generated in the next step and add to the static pool hosts list in MicrovmCluster.

Again, ensure that the version is compatible with that of Flintlock:

export CAPMVM_VERSION=<version>

Use clusterctl to generate a cluster manifest from the CAPMVM release template:

clusterctl generate cluster -i microvm:$CAPMVM_VERSION -f cilium $CLUSTER_NAME > cluster.yaml

Note: The CAPMVM repo is private. If you do not have access, you will not be able to download this template.

Create

Perform a dry-run:

kubectl apply -f cluster.yaml --dry-run=server

And if you are happy, apply the manifest to create your cluster:

kubectl apply -f cluster.yaml

Use your new cluster

CAPI will create a secret containing your new Liquid Metal cluster's kubeconfig in the same namespace.

The secret will be named <cluster-name>-kubeconfig of type cluster.x-k8s.io/secret.

Use jq to extract the base64 encoded config and save it to a file (set the namespace if you changed it for your cluster previously):

kubectl get secret $CLUSTER_NAME-kubeconfig -o json | jq -r .data.value | base64 -d > config.yaml

We can then use that kubeconfig to interact with our Liquid Metal cluster:

kubectl --kubeconfig config.yaml get pods -A
NAMESPACE     NAME                                                   READY   STATUS    RESTARTS   AGE
kube-system   cilium-cls92                                           1/1     Running   0          1m
kube-system   cilium-nrqsd                                           1/1     Running   0          1m
kube-system   cilium-operator-595485c9f9-n2rxw                       1/1     Running   0          1m
kube-system   cilium-rvlwg                                           1/1     Running   0          1m
kube-system   coredns-558bd4d5db-bjzz4                               1/1     Running   0          1m
kube-system   coredns-558bd4d5db-jm4jl                               1/1     Running   0          1m
kube-system   etcd-mvm-test-control-plane-2dkqr                      1/1     Running   0          1m
kube-system   kube-apiserver-mvm-test-control-plane-2dkqr            1/1     Running   0          1m
kube-system   kube-controller-manager-mvm-test-control-plane-2dkqr   1/1     Running   0          1m
kube-system   kube-proxy-7zs6s                                       1/1     Running   0          1m
kube-system   kube-proxy-ghwdr                                       1/1     Running   0          1m
kube-system   kube-proxy-xrmfv                                       1/1     Running   0          1m
kube-system   kube-scheduler-mvm-test-control-plane-2dkqr            1/1     Running   0          1m
kube-system   kube-vip-mvm-test-control-plane-2dkqr                  1/1     Running   0          1m

Note: you will either have to be connected to a VPN or be in the same network as your MicroVMs to be able to connect to your cluster, depending on your network setup.

Delete

To delete your cluster, run:

kubectl delete clusters.cluster.x-k8s.io $CLUSTER_NAME

And to delete the cilium resource set:

kubectl delete clusterresourceset.addons.cluster.x-k8s.io/crs-cilium
kubectl delete configmap/cilium-addon

Troubleshooting

To troubleshoot issues with your MicroVMs or hosts, see the troubleshooting guide.