Deploy a Kubernetes stateless application via kubectl on your Azure Stack Edge Pro GPU device
APPLIES TO: 
Azure Stack Edge Pro - GPUAzure Stack Edge Pro 2Azure Stack Edge Pro RAzure Stack Edge Mini R
This article describes how to deploy a stateless application using kubectl commands on an existing Kubernetes cluster. This article also walks you through the process of creating and setting up pods in your stateless application.
Prerequisites
Before you can create a Kubernetes cluster and use the kubectl command-line tool, you need to ensure that:
You have sign-in credentials to a 1-node Azure Stack Edge Pro device.
Windows PowerShell 5.0 or later is installed on a Windows client system to access the Azure Stack Edge Pro device. You can have any other client with a Supported operating system as well. This article describes the procedure when using a Windows client. To download the latest version of Windows PowerShell, go to Installing Windows PowerShell.
Compute is enabled on the Azure Stack Edge Pro device. To enable compute, go to the Compute page in the local UI of the device. Then select a network interface that you want to enable for compute. Select Enable. Enabling compute results in the creation of a virtual switch on your device on that network interface. For more information, see Enable compute network on your Azure Stack Edge Pro.
Your Azure Stack Edge Pro device has a Kubernetes cluster server running that is version v1.9 or later. For more information, see Create and manage a Kubernetes cluster on Microsoft Azure Stack Edge Pro device.
You have installed
kubectl.
Deploy a stateless application
Before we begin, you should have:
- Created a Kubernetes cluster.
- Set up a namespace.
- Associated a user with the namespace.
- Saved the user configuration to
C:\Users\<username>\.kube. - Installed
kubectl.
Now you can begin running and managing stateless application deployments on an Azure Stack Edge Pro device. Before you start using kubectl, you need to verify that you have the correct version of kubectl.
Verify you have the correct version of kubectl and set up configuration
To check the version of kubectl:
Verify that the version of
kubectlis greater or equal to 1.9:kubectl versionAn example of the output is shown below:
PS C:\WINDOWS\system32> C:\windows\system32\kubectl.exe version Client Version: version.Info{Major:"1", Minor:"15", GitVersion:"v1.15.2", GitCommit:"f6278300bebbb750328ac16ee6dd3aa7d3549568", GitTreeState:"clean", BuildDate:"2019-08-05T09:23:26Z", GoVersion:"go1.12.5", Compiler:"gc", Platform:"windows/amd64"} Server Version: version.Info{Major:"1", Minor:"15", GitVersion:"v1.15.1", GitCommit:"4485c6f18cee9a5d3c3b4e523bd27972b1b53892", GitTreeState:"clean", BuildDate:"2019-07-18T09:09:21Z", GoVersion:"go1.12.5", Compiler:"gc", Platform:"linux/amd64"}In this case, the client version of kubectl is v1.15.2 and is compatible to continue.
Get a list of the pods running on your Kubernetes cluster. A pod is an application container, or process, running on your Kubernetes cluster.
kubectl get pods -n <namespace-string>An example of command usage is shown below:
PS C:\WINDOWS\system32> kubectl get pods -n "test1" No resources found. PS C:\WINDOWS\system32>The output should state that no resources (pods) are found because there are no applications running on your cluster.
The command will populate the directory structure of "C:\Users\<username>\.kube" with configuration files. The kubectl command-line tool will use these files to create and manage stateless applications on your Kubernetes cluster.
Manually check the directory structure of "C:\Users\<username>\.kube" to verify kubectl has populated it with the following subfolders:
PS C:\Users\username> ls .kube Directory: C:\Users\user\.kube Mode LastWriteTime Length Name ---- ------------- ------ ---- d----- 2/18/2020 11:05 AM cache d----- 2/18/2020 11:04 AM http-cache -a---- 2/18/2020 10:41 AM 5377 config
Note
To view a list of all kubectl commands, type kubectl --help.
Create a stateless application using a deployment
Now that you've verified that the kubectl command-line version is correct and you have the required configuration files, you can create a stateless application deployment.
A pod is the basic execution unit of a Kubernetes application, the smallest and simplest unit in the Kubernetes object model that you create or deploy. A pod also encapsulates storage resources, a unique network IP, and options that govern how the container(s) should run.
The type of stateless application that you create is an nginx web server deployment.
All kubectl commands you use to create and manage stateless application deployments need to specify the namespace associated with the configuration. You created the namespace while connected to the cluster on the Azure Stack Edge Pro device in the Create and manage a Kubernetes cluster on Microsoft Azure Stack Edge Pro device tutorial with New-HcsKubernetesNamespace.
To specify the namespace in a kubectl command, use kubectl <command> -n <namespace-string>.
Follow these steps to create an nginx deployment:
Apply a stateless application by creating a Kubernetes deployment object:
kubectl apply -f <yaml-file> -n <namespace-string>In this example, the path to the application YAML file is an external source.
Here is a sample use of the command and its output:
PS C:\WINDOWS\system32> kubectl apply -f https://k8s.io/examples/application/deployment.yaml -n "test1" deployment.apps/nginx-deployment createdAlternatively, you can save the following markdown to your local machine and substitute the path and filename in the -f parameter. For instance, "C:\Kubernetes\deployment.yaml". The configuration for the application deployment would be:
apiVersion: apps/v1 # for versions before 1.9.0 use apps/v1beta2 kind: Deployment metadata: name: nginx-deployment spec: selector: matchLabels: app: nginx replicas: 2 # tells deployment to run 2 pods matching the template template: metadata: labels: app: nginx spec: containers: - name: nginx image: nginx:1.7.9 ports: - containerPort: 80This command creates a default nginx-deployment that has two pods to run your application.
Get the description of the Kubernetes nginx-deployment you created:
kubectl describe deployment nginx-deployment -n <namespace-string>A sample use of the command, with output, is shown below:
PS C:\Users\user> kubectl describe deployment nginx-deployment -n "test1" Name: nginx-deployment Namespace: test1 CreationTimestamp: Tue, 18 Feb 2020 13:35:29 -0800 Labels: <none> Annotations: deployment.kubernetes.io/revision: 1 kubectl.kubernetes.io/last-applied-configuration: {"apiVersion":"apps/v1","kind":"Deployment","metadata":{"annotations":{},"name":"nginx-deployment","namespace":"test1"},"spec":{"repl... Selector: app=nginx Replicas: 2 desired | 2 updated | 2 total | 2 available | 0 unavailable StrategyType: RollingUpdate MinReadySeconds: 0 RollingUpdateStrategy: 25% max unavailable, 25% max surge Pod Template: Labels: app=nginx Containers: nginx: Image: nginx:1.7.9 Port: 80/TCP Host Port: 0/TCP Environment: <none> Mounts: <none> Volumes: <none> Conditions: Type Status Reason ---- ------ ------ Available True MinimumReplicasAvailable Progressing True NewReplicaSetAvailable OldReplicaSets: <none> NewReplicaSet: nginx-deployment-5754944d6c (2/2 replicas created) Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal ScalingReplicaSet 2m22s deployment-controller Scaled up replica set nginx-deployment-5754944d6c to 2For the replicas setting, you will see:
Replicas: 2 desired | 2 updated | 2 total | 2 available | 0 unavailableThe replicas setting indicates that your deployment specification requires two pods, and that those pods were created and updated and are ready for you to use.
Note
A replica set replaces pods that are deleted or terminated for any reason, such as in the case of device node failure or a disruptive device upgrade. For this reason, we recommend that you use a replica set even if your application requires only a single pod.
To list the pods in your deployment:
kubectl get pods -l app=nginx -n <namespace-string>A sample use of the command, with output, is shown below:
PS C:\Users\user> kubectl get pods -l app=nginx -n "test1" NAME READY STATUS RESTARTS AGE nginx-deployment-5754944d6c-7wqjd 1/1 Running 0 3m13s nginx-deployment-5754944d6c-nfj2h 1/1 Running 0 3m13sThe output verifies that we have two pods with unique names that we can reference using kubectl.
To view information on an individual pod in your deployment:
kubectl describe pod <podname-string> -n <namespace-string>
A sample use of the command, with output, is shown below:
PS C:\Users\user> kubectl describe pod "nginx-deployment-5754944d6c-7wqjd" -n "test1"
Name: nginx-deployment-5754944d6c-7wqjd
Namespace: test1
Priority: 0
Node: k8s-1d9qhq2cl-n1/10.128.46.184
Start Time: Tue, 18 Feb 2020 13:35:29 -0800
Labels: app=nginx
pod-template-hash=5754944d6c
Annotations: <none>
Status: Running
IP: 172.17.246.200
Controlled By: ReplicaSet/nginx-deployment-5754944d6c
Containers:
nginx:
Container ID: docker://280b0f76bfdc14cde481dc4f2b8180cf5fbfc90a084042f679d499f863c66979
Image: nginx:1.7.9
Image ID: docker-pullable://nginx@sha256:e3456c851a152494c3e4ff5fcc26f240206abac0c9d794affb40e0714846c451
Port: 80/TCP
Host Port: 0/TCP
State: Running
Started: Tue, 18 Feb 2020 13:35:35 -0800
Ready: True
Restart Count: 0
Environment: <none>
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from default-token-8gksw (ro)
Conditions:
Type Status
Initialized True
Ready True
ContainersReady True
PodScheduled True
Volumes:
default-token-8gksw:
Type: Secret (a volume populated by a Secret)
SecretName: default-token-8gksw
Optional: false
QoS Class: BestEffort
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 4m58s default-scheduler Successfully assigned test1/nginx-deployment-5754944d6c-7wqjd to k8s-1d9qhq2cl-n1
Normal Pulling 4m57s kubelet, k8s-1d9qhq2cl-n1 Pulling image "nginx:1.7.9"
Normal Pulled 4m52s kubelet, k8s-1d9qhq2cl-n1 Successfully pulled image "nginx:1.7.9"
Normal Created 4m52s kubelet, k8s-1d9qhq2cl-n1 Created container nginx
Normal Started 4m52s kubelet, k8s-1d9qhq2cl-n1 Started container nginx
Rescale the application deployment by increasing the replica count
Each pod is meant to run a single instance of a given application. If you want to scale your application horizontally to run multiple instances, you can increase the number of pods to one for each instance. In Kubernetes, this is referred to as replication. You can increase the number of pods in your application deployment by applying a new YAML file. The YAML file changes the replicas setting to 4, which increases the number of pods in your deployment to four pods. To increase the number of pods from 2 to 4:
PS C:\WINDOWS\system32> kubectl apply -f https://k8s.io/examples/application/deployment-scale.yaml -n "test1"
Alternatively, you can save the following markdown on your local machine and substitute the path and filename for the -f parameter for kubectl apply. For instance, "C:\Kubernetes\deployment-scale.yaml". The configuration for the application deployment scale would be:
apiVersion: apps/v1 # for versions before 1.9.0 use apps/v1beta2
kind: Deployment
metadata:
name: nginx-deployment
spec:
selector:
matchLabels:
app: nginx
replicas: 4 # Update the replicas from 2 to 4
template:
metadata:
labels:
app: nginx
spec:
containers:
- name: nginx
image: nginx:1.8
ports:
- containerPort: 80
To verify that the deployment has four pods:
kubectl get pods -l app=nginx
Example output for a rescaling deployment from two to four pods is shown below:
PS C:\WINDOWS\system32> kubectl get pods -l app=nginx
NAME READY STATUS RESTARTS AGE
nginx-deployment-148880595-4zdqq 1/1 Running 0 25s
nginx-deployment-148880595-6zgi1 1/1 Running 0 25s
nginx-deployment-148880595-fxcez 1/1 Running 0 2m
nginx-deployment-148880595-rwovn 1/1 Running 0 2m
As you can see from the output, you now have four pods in your deployment that can run your application.
Delete a Deployment
To delete the deployment, including all the pods, you need to run kubectl delete deployment specifying the name of the deployment nginx-deployment and the namespace name. To delete the deployment:
kubectl delete deployment nginx-deployment -n <namespace-string>
An example of command usage, with output, is shown below:
PS C:\Users\user> kubectl delete deployment nginx-deployment -n "test1"
deployment.extensions "nginx-deployment" deleted