This is the multi-page printable view of this section. Click here to print.

Return to the regular view of this page.

Cluster Management

1 - Running Kubelet in Standalone Mode

This tutorial shows you how to run a standalone kubelet instance.

You may have different motivations for running a standalone kubelet. This tutorial is aimed at introducing you to Kubernetes, even if you don't have much experience with it. You can follow this tutorial and learn about node setup, basic (static) Pods, and how Kubernetes manages containers.

Once you have followed this tutorial, you could try using a cluster that has a control plane to manage pods and nodes, and other types of objects. For example, Hello, minikube.

You can also run the kubelet in standalone mode to suit production use cases, such as to run the control plane for a highly available, resiliently deployed cluster. This tutorial does not cover the details you need for running a resilient control plane.

Objectives

  • Install cri-o, and kubelet on a Linux system and run them as systemd services.
  • Launch a Pod running nginx that listens to requests on TCP port 80 on the Pod's IP address.
  • Learn how the different components of the solution interact among themselves.

Before you begin

  • Admin (root) access to a Linux system that uses systemd and iptables (or nftables with iptables emulation).
  • Access to the Internet to download the components needed for the tutorial, such as:

Prepare the system

Swap configuration

By default, kubelet fails to start if swap memory is detected on a node. This means that swap should either be disabled or tolerated by kubelet.

If you have swap memory enabled, either disable it or add failSwapOn: false to the kubelet configuration file.

To check if swap is enabled:

sudo swapon --show

If there is no output from the command, then swap memory is already disabled.

To disable swap temporarily:

sudo swapoff -a

To make this change persistent across reboots:

Make sure swap is disabled in either /etc/fstab or systemd.swap, depending on how it was configured on your system.

Enable IPv4 packet forwarding

To check if IPv4 packet forwarding is enabled:

cat /proc/sys/net/ipv4/ip_forward

If the output is 1, it is already enabled. If the output is 0, then follow next steps.

To enable IPv4 packet forwarding, create a configuration file that sets the net.ipv4.ip_forward parameter to 1:

sudo tee /etc/sysctl.d/k8s.conf <<EOF
net.ipv4.ip_forward = 1
EOF

Apply the changes to the system:

sudo sysctl --system

The output is similar to:

...
* Applying /etc/sysctl.d/k8s.conf ...
net.ipv4.ip_forward = 1
* Applying /etc/sysctl.conf ...

Download, install, and configure the components

Install a container runtime

Download the latest available versions of the required packages (recommended).

This tutorial suggests installing the CRI-O container runtime (external link).

There are several ways to install the CRI-O container runtime, depending on your particular Linux distribution. Although CRI-O recommends using either deb or rpm packages, this tutorial uses the static binary bundle script of the CRI-O Packaging project, both to streamline the overall process, and to remain distribution agnostic.

The script installs and configures additional required software, such as cni-plugins, for container networking, and crun and runc, for running containers.

The script will automatically detect your system's processor architecture (amd64 or arm64) and select and install the latest versions of the software packages.

Set up CRI-O

Visit the releases page (external link).

Download the static binary bundle script:

curl https://raw.githubusercontent.com/cri-o/packaging/main/get > crio-install

Run the installer script:

sudo bash crio-install

Enable and start the crio service:

sudo systemctl daemon-reload
sudo systemctl enable --now crio.service

Quick test:

sudo systemctl is-active crio.service

The output is similar to:

active

Detailed service check:

sudo journalctl -f -u crio.service

Install network plugins

The cri-o installer installs and configures the cni-plugins package. You can verify the installation running the following command:

/opt/cni/bin/bridge --version

The output is similar to:

CNI bridge plugin v1.5.1
CNI protocol versions supported: 0.1.0, 0.2.0, 0.3.0, 0.3.1, 0.4.0, 1.0.0

To check the default configuration:

cat /etc/cni/net.d/11-crio-ipv4-bridge.conflist

The output is similar to:

{
  "cniVersion": "1.0.0",
  "name": "crio",
  "plugins": [
    {
      "type": "bridge",
      "bridge": "cni0",
      "isGateway": true,
      "ipMasq": true,
      "hairpinMode": true,
      "ipam": {
        "type": "host-local",
        "routes": [
            { "dst": "0.0.0.0/0" }
        ],
        "ranges": [
            [{ "subnet": "10.85.0.0/16" }]
        ]
      }
    }
  ]
}

Download and set up the kubelet

Download the latest stable release of the kubelet.


curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubelet"



curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/arm64/kubelet"

Configure:

sudo mkdir -p /etc/kubernetes/manifests

sudo tee /etc/kubernetes/kubelet.yaml <<EOF
apiVersion: kubelet.config.k8s.io/v1beta1
kind: KubeletConfiguration
authentication:
  webhook:
    enabled: false # Do NOT use in production clusters!
authorization:
  mode: AlwaysAllow # Do NOT use in production clusters!
enableServer: false
logging:
  format: text
address: 127.0.0.1 # Restrict access to localhost
readOnlyPort: 10255 # Do NOT use in production clusters!
staticPodPath: /etc/kubernetes/manifests
containerRuntimeEndpoint: unix:///var/run/crio/crio.sock
EOF

Install:

chmod +x kubelet
sudo cp kubelet /usr/bin/

Create a systemd service unit file:

sudo tee /etc/systemd/system/kubelet.service <<EOF
[Unit]
Description=Kubelet

[Service]
ExecStart=/usr/bin/kubelet \
 --config=/etc/kubernetes/kubelet.yaml
Restart=always

[Install]
WantedBy=multi-user.target
EOF

The command line argument --kubeconfig has been intentionally omitted in the service configuration file. This argument sets the path to a kubeconfig file that specifies how to connect to the API server, enabling API server mode. Omitting it, enables standalone mode.

Enable and start the kubelet service:

sudo systemctl daemon-reload
sudo systemctl enable --now kubelet.service

Quick test:

sudo systemctl is-active kubelet.service

The output is similar to:

active

Detailed service check:

sudo journalctl -u kubelet.service

Check the kubelet's API /healthz endpoint:

curl http://localhost:10255/healthz?verbose

The output is similar to:

[+]ping ok
[+]log ok
[+]syncloop ok
healthz check passed

Query the kubelet's API /pods endpoint:

curl http://localhost:10255/pods | jq '.'

The output is similar to:

{
  "kind": "PodList",
  "apiVersion": "v1",
  "metadata": {},
  "items": null
}

Run a Pod in the kubelet

In standalone mode, you can run Pods using Pod manifests. The manifests can either be on the local filesystem, or fetched via HTTP from a configuration source.

Create a manifest for a Pod:

cat <<EOF > static-web.yaml
apiVersion: v1
kind: Pod
metadata:
  name: static-web
spec:
  containers:
    - name: web
      image: nginx
      ports:
        - name: web
          containerPort: 80
          protocol: TCP
EOF

Copy the static-web.yaml manifest file to the /etc/kubernetes/manifests directory.

sudo cp static-web.yaml /etc/kubernetes/manifests/

Find out information about the kubelet and the Pod

The Pod networking plugin creates a network bridge (cni0) and a pair of veth interfaces for each Pod (one of the pair is inside the newly made Pod, and the other is at the host level).

Query the kubelet's API endpoint at http://localhost:10255/pods:

curl http://localhost:10255/pods | jq '.'

To obtain the IP address of the static-web Pod:

curl http://localhost:10255/pods | jq '.items[].status.podIP'

The output is similar to:

"10.85.0.4"

Connect to the nginx server Pod on http://<IP>:<Port> (port 80 is the default), in this case:

curl http://10.85.0.4

The output is similar to:

<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
...

Where to look for more details

If you need to diagnose a problem getting this tutorial to work, you can look within the following directories for monitoring and troubleshooting:

/var/lib/cni
/var/lib/containers
/var/lib/kubelet

/var/log/containers
/var/log/pods

Clean up

kubelet

sudo systemctl disable --now kubelet.service
sudo systemctl daemon-reload
sudo rm /etc/systemd/system/kubelet.service
sudo rm /usr/bin/kubelet
sudo rm -rf /etc/kubernetes
sudo rm -rf /var/lib/kubelet
sudo rm -rf /var/log/containers
sudo rm -rf /var/log/pods

Container Runtime

sudo systemctl disable --now crio.service
sudo systemctl daemon-reload
sudo rm -rf /usr/local/bin
sudo rm -rf /usr/local/lib
sudo rm -rf /usr/local/share
sudo rm -rf /usr/libexec/crio
sudo rm -rf /etc/crio
sudo rm -rf /etc/containers

Network Plugins

sudo rm -rf /opt/cni
sudo rm -rf /etc/cni
sudo rm -rf /var/lib/cni

Conclusion

This page covered the basic aspects of deploying a kubelet in standalone mode. You are now ready to deploy Pods and test additional functionality.

Notice that in standalone mode the kubelet does not support fetching Pod configurations from the control plane (because there is no control plane connection).

You also cannot use a ConfigMap or a Secret to configure the containers in a static Pod.

What's next

2 - Install Drivers and Allocate Devices with DRA

FEATURE STATE: Kubernetes v1.32 [beta] (enabled by default: false)

This tutorial shows you how to install DRA drivers in your cluster and how to use them in conjunction with the DRA APIs to allocate devices to Pods. This page is intended for cluster administrators.

Dynamic Resource Allocation (DRA) is a Kubernetes feature that allows a cluster to manage availability and allocation of hardware resources to satisfy Pod-based claims for hardware requirements and preferences (see the DRA Concept page for more background). To support this, a mixture of Kubernetes built-in components (like the Kubernetes scheduler, kubelet, and kube-controller-manager) and third-party components (called DRA drivers) share the responsibility to advertise, allocate, prepare, mount, healthcheck, unprepare, and cleanup resources throughout the Pod lifecycle. These components share information via a series of DRA specific APIs in the resource.k8s.io API group, including DeviceClasses, ResourceSlices, ResourceClaims, as well as new fields in the Pod spec itself.

Objectives

  • Deploy an example DRA driver
  • Deploy a Pod requesting a hardware claim using DRA APIs
  • Delete a Pod that has a claim

Before you begin

Your cluster should support RBAC. You can try this tutorial with a cluster using a different authorization mechanism, but in that case you will have to adapt the steps around defining roles and permissions.

You need to have a Kubernetes cluster, and the kubectl command-line tool must be configured to communicate with your cluster. It is recommended to run this tutorial on a cluster with at least two nodes that are not acting as control plane hosts. If you do not already have a cluster, you can create one by using minikube or you can use one of these Kubernetes playgrounds:

This tutorial has been tested with Linux nodes, though it may also work with other types of nodes.

Your Kubernetes server must be at or later than version v1.32.

To check the version, enter kubectl version.

Your cluster also must be configured to use the Dynamic Resource Allocation feature. To enable the DRA feature, you must enable the following feature gates and API groups:

  1. Enable the DynamicResourceAllocation feature gate on all of the following components:

    • kube-apiserver
    • kube-controller-manager
    • kube-scheduler
    • kubelet

  2. Enable the following API groups:

    • resource.k8s.io/v1beta1: required for DRA to function.
    • resource.k8s.io/v1beta2: optional, recommended improvements to the user experience.

    For more information, see Enabling or disabling API groups.

Explore the DRA initial state

With no driver installed or Pod claims yet to satisfy, you can observe the initial state of a cluster with DRA enabled.

  1. Get a list of DeviceClasses:

    kubectl get deviceclasses

    The output is similar to this:

    No resources found

    If you set up a new blank cluster for this tutorial, it's normal to find that there are no DeviceClasses. Learn more about DeviceClasses here.

  2. Get a list of ResourceSlices:

    kubectl get resourceslices

    The output is similar to this:

    No resources found

    If you set up a new blank cluster for this tutorial, it's normal to find that there are no ResourceSlices advertised. Learn more about ResourceSlices here.

  3. View ResourceClaims and ResourceClaimTemplates

    kubectl get resourceclaims -A
kubectl get resourceclaimtemplates -A

    The output is similar to this:

    No resources found
No resources found

    If you set up a new blank cluster for this tutorial, it's normal to find that there are no ResourceClaims or ResourceClaimTemplates as you, the user, have not created any. Learn more about ResourceClaims and ResourceClaimTemplates here.

At this point, you have confirmed that DRA is enabled and configured properly in the cluster, and that no DRA drivers have advertised any resources to the DRA APIs yet.

Install an example DRA driver

DRA drivers are third-party applications that run on each node of your cluster to interface with the hardware of that node and Kubernetes' built-in DRA components. The installation procedure depends on the driver you choose, but is likely deployed as a DaemonSet to all or a selection of the nodes (using selectors or similar mechanisms) in your cluster.

Check your driver's documentation for specific installation instructions, which may include a Helm chart, a set of manifests, or other deployment tooling.

This tutorial uses an example driver which can be found in the kubernetes-sigs/dra-example-driver repository to demonstrate driver installation.

Prepare your cluster for driver installation

To make it easier to cleanup later, create a namespace called dra-tutorial in your cluster.

In a production environment, you would likely be using a previously released or qualified image from the driver vendor or your own organization, and your nodes would need to have access to the image registry where the driver image is hosted. In this tutorial, you will use a publicly released image of the dra-example-driver to simulate access to a DRA driver image.

  1. Create the namespace:

    kubectl create namespace dra-tutorial

  2. Confirm your nodes have access to the image by running the following from within one of your cluster's nodes:

    docker pull registry.k8s.io/dra-example-driver/dra-example-driver:v0.1.0

Deploy the DRA driver components

For this tutorial, you will install the critical example resource driver components individually with kubectl.

  1. Create the DeviceClass representing the device types this DRA driver supports:

    dra/driver-install/deviceclass.yaml 
    apiVersion: resource.k8s.io/v1beta1
kind: DeviceClass
metadata:
  name: gpu.example.com
spec:
  selectors:
  - cel: 
      expression: "device.driver == 'gpu.example.com'"
    kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/deviceclass.yaml

  2. Create the ServiceAccount, ClusterRole and ClusterRoleBinding that will be used by the driver to gain permissions to interact with the Kubernetes API on this cluster:

    1. Create the Service Account:

      dra/driver-install/serviceaccount.yaml 
      apiVersion: v1
kind: ServiceAccount
metadata:
  name: dra-example-driver-service-account
  namespace: dra-tutorial
  labels:
    app.kubernetes.io/name: dra-example-driver
    app.kubernetes.io/instance: dra-example-driver
      kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/serviceaccount.yaml

    2. Create the ClusterRole:

      dra/driver-install/clusterrole.yaml 
      apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: dra-example-driver-role
rules:
- apiGroups: ["resource.k8s.io"]
  resources: ["resourceclaims"]
  verbs: ["get"]
- apiGroups: [""]
  resources: ["nodes"]
  verbs: ["get"]
- apiGroups: ["resource.k8s.io"]
  resources: ["resourceslices"]
  verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]
      kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/clusterrole.yaml

    3. Create the ClusterRoleBinding:

      dra/driver-install/clusterrolebinding.yaml 
      apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: dra-example-driver-role-binding
subjects:
- kind: ServiceAccount
  name: dra-example-driver-service-account
  namespace: dra-tutorial
roleRef:
  kind: ClusterRole
  name: dra-example-driver-role
  apiGroup: rbac.authorization.k8s.io
      kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/clusterrolebinding.yaml

  3. Create a PriorityClass for the DRA driver. The DRA driver component is responsible for important lifecycle operations for Pods with claims, so you don't want it to be preempted. Learn more about pod priority and preemption here. Learn more about good practices when maintaining a DRA driver here.

    dra/driver-install/priorityclass.yaml 
    apiVersion: scheduling.k8s.io/v1
kind: PriorityClass
metadata:
  name: dra-driver-high-priority
value: 1000000
globalDefault: false
description: "This priority class should be used for DRA driver pods only."
    kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/priorityclass.yaml

  4. Deploy the actual DRA driver as a DaemonSet configured to run the example driver binary with the permissions provisioned above.

    dra/driver-install/daemonset.yaml 
    apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: dra-example-driver-kubeletplugin
  namespace: dra-tutorial
  labels:
    app.kubernetes.io/name: dra-example-driver
spec:
  selector:
    matchLabels:
      app.kubernetes.io/name: dra-example-driver
  updateStrategy:
    type: RollingUpdate
  template:
    metadata:
      labels:
        app.kubernetes.io/name: dra-example-driver
    spec:
      priorityClassName: dra-driver-high-priority
      serviceAccountName: dra-example-driver-service-account
      securityContext:
        {}
      containers:
      - name: plugin
        securityContext:
          privileged: true
        image: registry.k8s.io/dra-example-driver/dra-example-driver:v0.1.0
        imagePullPolicy: IfNotPresent
        command: ["dra-example-kubeletplugin"]
        resources:
          {}
        # Production drivers should always implement a liveness probe
        # For the tutorial we simply omit it
        # livenessProbe:
        #   grpc:
        #     port: 51515
        #     service: liveness
        #   failureThreshold: 3
        #   periodSeconds: 10
        env:
        - name: CDI_ROOT
          value: /var/run/cdi
        - name: KUBELET_REGISTRAR_DIRECTORY_PATH
          value: "/var/lib/kubelet/plugins_registry"
        - name: KUBELET_PLUGINS_DIRECTORY_PATH
          value: "/var/lib/kubelet/plugins"
        - name: NODE_NAME
          valueFrom:
            fieldRef:
              fieldPath: spec.nodeName
        - name: NAMESPACE
          valueFrom:
            fieldRef:
              fieldPath: metadata.namespace
        # Simulated number of devices the example driver will pretend to have.
        - name: NUM_DEVICES
          value: "9"
        - name: HEALTHCHECK_PORT
          value: "51515"
        volumeMounts:
        - name: plugins-registry
          mountPath: "/var/lib/kubelet/plugins_registry"
        - name: plugins
          mountPath: "/var/lib/kubelet/plugins"
        - name: cdi
          mountPath: /var/run/cdi
      volumes:
      - name: plugins-registry
        hostPath:
          path: "/var/lib/kubelet/plugins_registry"
      - name: plugins
        hostPath:
          path: "/var/lib/kubelet/plugins"
      - name: cdi
        hostPath:
          path: /var/run/cdi
    kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/daemonset.yaml

    It is configured with the volume mounts necessary to interact with the underlying Container Device Interface (CDI) directory, and to expose its socket to kubelet via the kubelet plugins directory.

Verify the DRA driver installation

  1. Observe the Pods of the DRA driver DaemonSet across all worker nodes:

    kubectl get pod -l app.kubernetes.io/name=dra-example-driver -n dra-tutorial

    The output is similar to this:

    NAME                                     READY   STATUS    RESTARTS   AGE
dra-example-driver-kubeletplugin-4sk2x   1/1     Running   0          13s
dra-example-driver-kubeletplugin-cttr2   1/1     Running   0          13s

  2. The initial reponsibility of each node's local DRA driver is to update the cluster with what devices are available to Pods on that node, by publishing its metadata to the ResourceSlices API. You can check that API to see that each node with a driver is advertising the device class it represents.

    Check for available ResourceSlices:

    kubectl get resourceslices

    The output is similar to this:

    NAME                                 NODE           DRIVER            POOL           AGE
kind-worker-gpu.example.com-k69gd    kind-worker    gpu.example.com   kind-worker    19s
kind-worker2-gpu.example.com-qdgpn   kind-worker2   gpu.example.com   kind-worker2   19s

At this point, you have successfully installed the example DRA driver, and confirmed its initial configuration. You're now ready to use DRA to schedule Pods.

Claim resources and deploy a Pod

To request resources using DRA, you create ResourceClaims or ResourceClaimTemplates that define the resources that your Pods need. In the example driver, a memory capacity attribute is exposed for mock GPU devices. This section shows you how to use Common Expression Language to express your requirements in a ResourceClaim, select that ResourceClaim in a Pod specification, and observe the resource allocation.

This tutorial showcases only one basic example of a DRA ResourceClaim. Read Dynamic Resource Allocation to learn more about ResourceClaims.

Create the ResourceClaim

The Pod manifest itself will include a reference to its relevant ResourceClaim object, which you will create now. Whatever the claim, the deviceClassName is a required field, narrowing down the scope of the request to a specific device class. The request itself can include a Common Expression Language expression that references attributes that may be advertised by the driver managing that device class.

In this example, you will create a request for any GPU advertising over 10Gi memory capacity. The attribute exposing capacity from the example driver takes the form device.capacity['gpu.example.com'].memory. Note also that the name of the claim is set to some-gpu.

dra/driver-install/example/resourceclaim.yaml 
apiVersion: resource.k8s.io/v1beta2
kind: ResourceClaim
metadata:
 name: some-gpu
 namespace: dra-tutorial
spec:
   devices:
     requests:
     - name: some-gpu
       exactly:
         deviceClassName: gpu.example.com
         selectors:
         - cel:
             expression: "device.capacity['gpu.example.com'].memory.compareTo(quantity('10Gi')) >= 0"
kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/example/resourceclaim.yaml

Create the Pod that references that ResourceClaim

Below is the Pod manifest referencing the ResourceClaim you just made, some-gpu, in the spec.resourceClaims.resourceClaimName field. The local name for that claim, gpu, is then used in the spec.containers.resources.claims.name field to allocate the claim to the Pod's underlying container.

dra/driver-install/example/pod.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: pod0
  namespace: dra-tutorial
  labels:
    app: pod
spec:
  containers:
  - name: ctr0
    image: ubuntu:24.04
    command: ["bash", "-c"]
    args: ["export; trap 'exit 0' TERM; sleep 9999 & wait"]
    resources:
      claims:
      - name: gpu
  resourceClaims:
  - name: gpu
    resourceClaimName: some-gpu
kubectl apply --server-side -f http://k8s.io/examples/dra/driver-install/example/pod.yaml

Explore the DRA state

The cluster now tries to schedule that Pod to a node where Kubernetes can satisfy the ResourceClaim. In our situation, the DRA driver is deployed on all nodes, and is advertising mock GPUs on all nodes, all of which have enough capacity advertised to satisfy the Pod's claim, so this Pod may be scheduled to any node and any of the mock GPUs on that node may be allocated.

The mock GPU driver injects environment variables in each container it is allocated to in order to indicate which GPUs would have been injected into them by a real resource driver and how they would have been configured, so you can check those environment variables to see how the Pods have been handled by the system.

  1. Confirm the pod has deployed:

    kubectl get pod pod0 -n dra-tutorial

    The output is similar to this:

    NAME   READY   STATUS    RESTARTS   AGE
pod0   1/1     Running   0          9s

  2. Observe the pod logs which report the name of the mock GPU allocated:

    kubectl logs pod0 -c ctr0 -n dra-tutorial | grep -E "GPU_DEVICE_[0-9]+=" | grep -v "RESOURCE_CLAIM"

    The output is similar to this:

    declare -x GPU_DEVICE_4="gpu-4"

  3. Observe the ResourceClaim object:

    You can observe the ResourceClaim more closely, first only to see its state is allocated and reserved.

    kubectl get resourceclaims -n dra-tutorial

    The output is similar to this:

    NAME       STATE                AGE
some-gpu   allocated,reserved   34s

    Looking deeper at the some-gpu ResourceClaim, you can see that the status stanza includes information about the device that has been allocated and for what pod it has been reserved for:

    kubectl get resourceclaim some-gpu -n dra-tutorial -o yaml

    The output is similar to this: 

     1apiVersion: v1
 2items:
 3- apiVersion: resource.k8s.io/v1beta2
 4  kind: ResourceClaim
 5  metadata:
 6    creationTimestamp: "2025-07-29T05:11:52Z"
 7    finalizers:
 8    - resource.kubernetes.io/delete-protection
 9    name: some-gpu
10    namespace: dra-tutorial
11    resourceVersion: "58357"
12    uid: 79e1e8d8-7e53-4362-aad1-eca97678339e
13  spec:
14    devices:
15      requests:
16      - exactly:
17          allocationMode: ExactCount
18          count: 1
19          deviceClassName: gpu.example.com
20          selectors:
21          - cel:
22              expression: device.capacity['gpu.example.com'].memory.compareTo(quantity('10Gi'))
23                >= 0
24        name: some-gpu
25  status:
26    allocation:
27      devices:
28        results:
29        - adminAccess: null
30          device: gpu-4
31          driver: gpu.example.com
32          pool: kind-worker
33          request: some-gpu
34      nodeSelector:
35        nodeSelectorTerms:
36        - matchFields:
37          - key: metadata.name
38            operator: In
39            values:
40            - kind-worker
41    reservedFor:
42    - name: pod0
43      resource: pods
44      uid: fa55b59b-d28d-4f7d-9e5b-ef4c8476dff5
45kind: List
46metadata:
47  resourceVersion: ""

  4. Observe the driver by checking the pod logs for pods backing the driver daemonset:

    kubectl logs -l app.kubernetes.io/name=dra-example-driver -n dra-tutorial

    The output is similar to this:

    I0729 05:11:52.679057       1 driver.go:84] NodePrepareResource is called: number of claims: 1
I0729 05:11:52.684450       1 driver.go:112] Returning newly prepared devices for claim '79e1e8d8-7e53-4362-aad1-eca97678339e': [&Device{RequestNames:[some-gpu],PoolName:kind-worker,DeviceName:gpu-4,CDIDeviceIDs:[k8s.gpu.example.com/gpu=common k8s.gpu.example.com/gpu=79e1e8d8-7e53-4362-aad1-eca97678339e-gpu-4],}]

You have now successfully deployed a Pod with a DRA based claim, and seen it scheduled to an appropriate node and the associated DRA APIs updated to reflect its status.

Remove the Pod with a claim

When a Pod with a claim is deleted, the DRA driver deallocates the resource so it can be available for future scheduling. You can observe that by deleting our pod with a claim and seeing that the state of the ResourceClaim changes.

Delete the pod using the resource claim

  1. Delete the pod directly:

    kubectl delete pod pod0 -n dra-tutorial

    The output is similar to this:

    pod "pod0" deleted

Observe the DRA state

The driver will deallocate the hardware and update the corresponding ResourceClaim resource that previously held the association.

  1. Check the ResourceClaim is now pending:

    kubectl get resourceclaims -n dra-tutorial

    The output is similar to this:

    NAME       STATE     AGE
some-gpu   pending   76s

  2. Observe the driver logs and see that it processed unpreparing the device for this claim:

    kubectl logs -l app.kubernetes.io/name=dra-example-driver -n dra-tutorial

    The output is similar to this:

    I0729 05:13:02.144623       1 driver.go:117] NodeUnPrepareResource is called: number of claims: 1

You have now deleted a Pod that had a claim, and observed that the driver took action to unprepare the underlying hardware resource and update the DRA APIs to reflect that the resource is available again for future scheduling.

Cleaning up

To cleanup the resources, delete the namespace for the tutorial which will clean up the ResourceClaims, driver components, and ServiceAccount. Then also delete the cluster level DeviceClass resource and cluster level RBAC resources.

kubectl delete namespace dra-tutorial
kubectl delete deviceclass gpu.example.com
kubectl delete clusterrole dra-example-driver-role
kubectl delete clusterrolebinding dra-example-driver-role-binding

What's next

3 - Namespaces Walkthrough

Kubernetes namespaces help different projects, teams, or customers to share a Kubernetes cluster.

It does this by providing the following:

  1. A scope for Names.
  2. A mechanism to attach authorization and policy to a subsection of the cluster.

Use of multiple namespaces is optional.

This example demonstrates how to use Kubernetes namespaces to subdivide your cluster.

Before you begin

You need to have a Kubernetes cluster, and the kubectl command-line tool must be configured to communicate with your cluster. It is recommended to run this tutorial on a cluster with at least two nodes that are not acting as control plane hosts. If you do not already have a cluster, you can create one by using minikube or you can use one of these Kubernetes playgrounds:

To check the version, enter kubectl version.

Prerequisites

This example assumes the following:

  1. You have an existing Kubernetes cluster.
  2. You have a basic understanding of Kubernetes Pods, Services, and Deployments.

Understand the default namespace

By default, a Kubernetes cluster will instantiate a default namespace when provisioning the cluster to hold the default set of Pods, Services, and Deployments used by the cluster.

Assuming you have a fresh cluster, you can inspect the available namespaces by doing the following:

kubectl get namespaces

NAME      STATUS    AGE
default   Active    13m

Create new namespaces

For this exercise, we will create two additional Kubernetes namespaces to hold our content.

Let's imagine a scenario where an organization is using a shared Kubernetes cluster for development and production use cases.

The development team would like to maintain a space in the cluster where they can get a view on the list of Pods, Services, and Deployments they use to build and run their application. In this space, Kubernetes resources come and go, and the restrictions on who can or cannot modify resources are relaxed to enable agile development.

The operations team would like to maintain a space in the cluster where they can enforce strict procedures on who can or cannot manipulate the set of Pods, Services, and Deployments that run the production site.

One pattern this organization could follow is to partition the Kubernetes cluster into two namespaces: development and production.

Let's create two new namespaces to hold our work.

Use the file namespace-dev.yaml which describes a development namespace:

admin/namespace-dev.yaml 
apiVersion: v1
kind: Namespace
metadata:
  name: development
  labels:
    name: development

Create the development namespace using kubectl.

kubectl create -f https://k8s.io/examples/admin/namespace-dev.yaml

Save the following contents into file namespace-prod.yaml which describes a production namespace:

admin/namespace-prod.yaml 
apiVersion: v1
kind: Namespace
metadata:
  name: production
  labels:
    name: production

And then let's create the production namespace using kubectl.

kubectl create -f https://k8s.io/examples/admin/namespace-prod.yaml

To be sure things are right, let's list all of the namespaces in our cluster.

kubectl get namespaces --show-labels

NAME          STATUS    AGE       LABELS
default       Active    32m       <none>
development   Active    29s       name=development
production    Active    23s       name=production

Create pods in each namespace

A Kubernetes namespace provides the scope for Pods, Services, and Deployments in the cluster.

Users interacting with one namespace do not see the content in another namespace.

To demonstrate this, let's spin up a simple Deployment and Pods in the development namespace.

We first check what is the current context:

kubectl config view

apiVersion: v1
clusters:
- cluster:
    certificate-authority-data: REDACTED
    server: https://130.211.122.180
  name: lithe-cocoa-92103_kubernetes
contexts:
- context:
    cluster: lithe-cocoa-92103_kubernetes
    user: lithe-cocoa-92103_kubernetes
  name: lithe-cocoa-92103_kubernetes
current-context: lithe-cocoa-92103_kubernetes
kind: Config
preferences: {}
users:
- name: lithe-cocoa-92103_kubernetes
  user:
    client-certificate-data: REDACTED
    client-key-data: REDACTED
    token: 65rZW78y8HbwXXtSXuUw9DbP4FLjHi4b
- name: lithe-cocoa-92103_kubernetes-basic-auth
  user:
    password: h5M0FtUUIflBSdI7
    username: admin

kubectl config current-context

lithe-cocoa-92103_kubernetes

The next step is to define a context for the kubectl client to work in each namespace. The value of "cluster" and "user" fields are copied from the current context.

kubectl config set-context dev --namespace=development \
  --cluster=lithe-cocoa-92103_kubernetes \
  --user=lithe-cocoa-92103_kubernetes

kubectl config set-context prod --namespace=production \
  --cluster=lithe-cocoa-92103_kubernetes \
  --user=lithe-cocoa-92103_kubernetes

By default, the above commands add two contexts that are saved into file .kube/config. You can now view the contexts and alternate against the two new request contexts depending on which namespace you wish to work against.

To view the new contexts:

kubectl config view

apiVersion: v1
clusters:
- cluster:
    certificate-authority-data: REDACTED
    server: https://130.211.122.180
  name: lithe-cocoa-92103_kubernetes
contexts:
- context:
    cluster: lithe-cocoa-92103_kubernetes
    user: lithe-cocoa-92103_kubernetes
  name: lithe-cocoa-92103_kubernetes
- context:
    cluster: lithe-cocoa-92103_kubernetes
    namespace: development
    user: lithe-cocoa-92103_kubernetes
  name: dev
- context:
    cluster: lithe-cocoa-92103_kubernetes
    namespace: production
    user: lithe-cocoa-92103_kubernetes
  name: prod
current-context: lithe-cocoa-92103_kubernetes
kind: Config
preferences: {}
users:
- name: lithe-cocoa-92103_kubernetes
  user:
    client-certificate-data: REDACTED
    client-key-data: REDACTED
    token: 65rZW78y8HbwXXtSXuUw9DbP4FLjHi4b
- name: lithe-cocoa-92103_kubernetes-basic-auth
  user:
    password: h5M0FtUUIflBSdI7
    username: admin

Let's switch to operate in the development namespace.

kubectl config use-context dev

You can verify your current context by doing the following:

kubectl config current-context

dev

At this point, all requests we make to the Kubernetes cluster from the command line are scoped to the development namespace.

Let's create some contents.

admin/snowflake-deployment.yaml 
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: snowflake
  name: snowflake
spec:
  replicas: 2
  selector:
    matchLabels:
      app: snowflake
  template:
    metadata:
      labels:
        app: snowflake
    spec:
      containers:
      - image: registry.k8s.io/serve_hostname
        imagePullPolicy: Always
        name: snowflake

Apply the manifest to create a Deployment

kubectl apply -f https://k8s.io/examples/admin/snowflake-deployment.yaml

We have created a deployment whose replica size is 2 that is running the pod called snowflake with a basic container that serves the hostname.

kubectl get deployment

NAME         READY   UP-TO-DATE   AVAILABLE   AGE
snowflake    2/2     2            2           2m

kubectl get pods -l app=snowflake

NAME                         READY     STATUS    RESTARTS   AGE
snowflake-3968820950-9dgr8   1/1       Running   0          2m
snowflake-3968820950-vgc4n   1/1       Running   0          2m

And this is great, developers are able to do what they want, and they do not have to worry about affecting content in the production namespace.

Let's switch to the production namespace and show how resources in one namespace are hidden from the other.

kubectl config use-context prod

The production namespace should be empty, and the following commands should return nothing.

kubectl get deployment
kubectl get pods

Production likes to run cattle, so let's create some cattle pods.

kubectl create deployment cattle --image=registry.k8s.io/serve_hostname --replicas=5

kubectl get deployment

NAME         READY   UP-TO-DATE   AVAILABLE   AGE
cattle       5/5     5            5           10s

kubectl get pods -l app=cattle

NAME                      READY     STATUS    RESTARTS   AGE
cattle-2263376956-41xy6   1/1       Running   0          34s
cattle-2263376956-kw466   1/1       Running   0          34s
cattle-2263376956-n4v97   1/1       Running   0          34s
cattle-2263376956-p5p3i   1/1       Running   0          34s
cattle-2263376956-sxpth   1/1       Running   0          34s

At this point, it should be clear that the resources users create in one namespace are hidden from the other namespace.

As the policy support in Kubernetes evolves, we will extend this scenario to show how you can provide different authorization rules for each namespace.