ℹ️ If you are using OCP 4.6 please follow this README
An operator to perform lifecycle management (install/upgrade/uninstall) of Kata Runtime on Openshift as well as Kubernetes cluster.
-
Make sure that
oc
is configured to talk to the cluster -
Clone the Kata Operator repository and check out the branch matching with the Openshift version. e.g. If you are running Openshift 4.7 then,
git clone https://github.com/openshift/kata-operator git checkout -b master --track origin/master
-
Install the Kata Operator on the cluster,
make install && make deploy IMG=quay.io/isolatedcontainers/kata-operator:4.7 oc adm policy add-scc-to-user privileged -z default -n kata-operator-system
-
To begin the installation of the kata runtime on the cluster,
oc create -f config/samples/kataconfiguration_v1_kataconfig.yaml
Please follow this section if you wish to install the Kata Runtime only on selected worker nodes.
Watch the description of the Kataconfig custom resource
oc describe kataconfig example-kataconfig
and look at the field 'Completed nodes' in the status. If the value matches the number of worker nodes the installation is completed.
Once the kata runtime binaries are successfully installed on the intended workers, Kata Operator will create a runtime class kata
. This runtime class can be used to deploy the pods that will use the Kata Runtime.
oc apply -f config/samples/example-fedora.yaml
-
edit the custom resource file
config/samples/kataconfiguration_v1_kataconfig.yaml
and uncomment the kata pool selector fields in the spec as follows,apiVersion: kataconfiguration.openshift.io/v1alpha1 kind: KataConfig metadata: name: example-kataconfig spec: kataConfigPoolSelector: matchLabels: custom-kata1: test
If you wish, you can change the label "custom-kata1:test" to something of your choice.
-
Apply the chosen label to the desired nodes. e.g.
oc label node <worker_node_name> custom-kata1=test
-
Create the custom resource to start the installation,
oc create -f config/samples/kataconfiguration_v1_kataconfig.yaml
oc delete kataconfig <KataConfig_CR_Name>
e.g.
oc delete kataconfig example-kataconfig
- During the installation you can watch the values of the kataconfig CR. Do
watch oc describe kataconfig example-kataconfig
. - To check if the nodes in the machine config pool are going through a config update watch the machine config pool resource. For this do
watch oc get mcp kata-oc
- Check the logs of the kata-operator controller pod to see detailled messages about what the steps it is executing. To find out the name of the controller pod,
oc get pods -n kata-operator-system | grep kata-operator-controller-manager
and then monitor the logs of the containermanager
in that pod.
The kata-operator uses three containers:
Container image name | Description | Container repository |
---|---|---|
kata-operator | It contains the controller part of the operator that watches and manages the kataconfig custom resource. It runs as a cluster scoped container. The operator itself is build with operator-sdk. | https://quay.io/isolatedcontainers/kata-operator |
kata-operator-daemon | The daemon part of the operator that runs on the nodes and performs the actual installation. It pulls down the container kata-operator-payload image. Dockerfile and other content can be found in images/daemon/ subdirectory of this github repository | https://quay.io/isolatedcontainers/kata-operator-daemon |
kata-operator-payload | The payload that is used by the daemon to install the kata binaries and dependencies (like e.g. QEMU). It's a container image with (currently) RPMs in it that will be installed on the chosen worker nodes by the daemon. Dockerfile and other content can be found in images/payload subdirectory of this github repository. | https://quay.io/isolatedcontaineres/kata-operator-payload |
Not implemented yet
Not implemented yet
- Install operator-sdk version 1.0 or above
- make docker-build docker-push IMG=quay.io//kata-operator: