edg3/experiment_software_csharp_12_net_8 · Datasets at Hugging Face

title stringlengths 3 46	content stringlengths 0 1.6k
20:501	metadata:
20:502	labels:
20:503	app: redis
20:504	role: master
20:505	tier: backend
20:506	spec:
20:507	containers:
20:508	- name: master
20:509	image: docker.io/redis:6.0.5
20:510	resources:
20:511	requests:
20:512	cpu: 100m
20:513	memory: 100Mi
20:514	ports:
20:515	- containerPort: 6379
20:516	---
20:517	apiVersion: v1
20:518	kind: Service
20:519	metadata:
20:520	name: redis-leader
20:521	labels:
20:522	app: redis
20:523	role: master
20:524	tier: backend
20:525	spec:
20:526	ports:
20:527	- port: 6379
20:528	targetPort: 6379
20:529	selector:
20:530	app: redis
20:531	role: master
20:532	tier: backend
20:533
20:534	The file is composed of two object definitions separated by a line containing just ---, that is, the object definition separator of .yaml files. It is common to group related objects, such as a Deployment with its associated Service, in the same file separated by the --- objects separator symbol in order to increase code readability.
20:535	The first object is a Deployment with a single replica, and the second object is a ClusterIP Service that exposes the Deployment on the 6379 port at the internal redis-leader.default.svc.cluster.local network address. The Deployment pod template defines the three app, role, and tier labels with values that are used in the selector definition of the Service to connect the Service with the unique Pod defined in the Deployment.
20:536	Let’s upload the redis-master.yaml file to Azure Cloud Shell, and then deploy it in the cluster with the following command:
20:537	kubectl create -f redis-master.yaml
20:538
20:539	Once the operation is complete, you can inspect the contents of the cluster with kubectl get all.
20:540	The slave storage is defined in the redis-slave.yaml file and is created in the same way, the only difference being that this time we have two replicas, and a different Docker image. The full code is in the GitHub repository associated with this book.
20:541	Let’s upload this file as well and deploy it with the following command:
20:542	kubectl create -f redis-slave.yaml
20:543
20:544	The code for the UI tier is contained in the frontend.yaml file. Deployment has three replicas and a different Service type. Let’s upload and deploy this file with the following command:
20:545	kubectl create -f frontend.yaml
20:546
20:547	It is worthwhile analyzing the Service code in the frontend.yaml file:
20:548	apiVersion: v1
20:549	kind: Service
20:550	metadata:
20:551	name: frontend
20:552	labels:
20:553	app: guestbook
20:554	tier: frontend
20:555	spec:
20:556	type: LoadBalancer
20:557	ports:
20:558	- port: 80
20:559	selector:
20:560	app: guestbook
20:561	tier: frontend
20:562
20:563	Again, the full code is in the GitHub repository associated with the book.
20:564	This Service is of the LoadBalancer type. Since this Pod is the application interface with the world outside of the Kubernetes cluster, its service must have a fixed IP and must be load balanced. Therefore, we must use a LoadBalancer service since this is the unique service type that satisfies those requirements. (See the Services section of this chapter for more information.)
20:565	If you are in Azure Kubernetes or any other cloud Kubernetes service, in order to get the public IP address assigned to the service, and then to the application, use the following command:
20:566	kubectl get service
20:567
20:568	The preceding command should display information on all the installed services. You should find the public IP in the EXTERNAL-IP column of the list. If you see only <none> values, please repeat the command until the public IP address is assigned to the load balancer.
20:569	If no IP is assigned after a few minutes, please verify whether there is some error or warning in any of the service descriptions. If not, please check whether all deployments are actually running using the following command:
20:570	kubectl get deployments
20:571
20:572	If, instead, you are on minikube, LoadBalancer services can be accessed by issuing this command:
20:573	minikube service <service name>
20:574
20:575	Thus, in our case:
20:576	minikube service frontend
20:577
20:578	The command should automatically open the browser.
20:579	Once you get the IP address, navigate with the browser to this address. The application’s home page should now appear!
20:580	If the page doesn’t appear, verify whether any service has an error by issuing the following command:
20:581	kubectl get service
20:582
20:583	If not, also verify that all deployments are in the running state with the following:
20:584	kubectl get deployments
20:585
20:586	If you find problems, please look for errors in the .yaml files, correct them, and then update the object defined in the file with:
20:587	kubectl update -f <file name>
20:588
20:589
20:590	Once you have finished experimenting with the application, make sure to remove the application from the cluster to avoid wasting your free Azure credit (public IP addresses cost money) with the following commands:
20:591	kubectl delete deployment frontend redis-master redis-slave
20:592	kubectl delete service frontend redis-leader redis-follower
20:593
20:594	In the next section, we will analyze other important Kubernetes features.
20:595	Advanced Kubernetes concepts
20:596	In this section, we will discuss other important Kubernetes features, including how to assign permanent storage to StatefulSets; how to store secrets such as passwords, connection strings, or certificates; how a container can inform Kubernetes about its health state; and how to handle complex Kubernetes packages with Helm. All of these subjects are organized into dedicated subsections. We will start with the problem of permanent storage.
20:597	Requiring permanent storage
20:598	Since Pods are moved between nodes, they can’t store data on the disk storage offered by the current node where they are running, or they would lose that storage as soon as they are moved to a different node. This leaves us with two options:
20:599
20:600	Using external databases: With the help of databases, ReplicaSets can also store information. However, if we need better performance in terms of write/update operations, we should use distributed sharded databases based on non-SQL engines such as Cosmos DB or MongoDB (see Chapter 12, Choosing Your Data Storage in the Cloud). In this case, in order to take maximum advantage of table sharding, we need StatefulSets, where each Pod instance takes care of a different table shard.

20:501

metadata:

20:502

labels:

20:503

app: redis

20:504

role: master

20:505

tier: backend

20:506

spec:

20:507

containers:

20:508

- name: master

20:509

image: docker.io/redis:6.0.5

20:510

resources:

20:511

requests:

20:512

cpu: 100m

20:513

memory: 100Mi

20:514

ports:

20:515

- containerPort: 6379

20:516

---

20:517

apiVersion: v1

20:518

kind: Service

20:519

metadata:

20:520

name: redis-leader

20:521

labels:

20:522

app: redis

20:523

role: master

20:524

tier: backend

20:525

spec:

20:526

ports:

20:527

- port: 6379

20:528

targetPort: 6379

20:529

selector:

20:530

app: redis

20:531

role: master

20:532

tier: backend

20:533

20:534

The file is composed of two object definitions separated by a line containing just ---, that is, the object definition separator of .yaml files. It is common to group related objects, such as a Deployment with its associated Service, in the same file separated by the --- objects separator symbol in order to increase code readability.

20:535

The first object is a Deployment with a single replica, and the second object is a ClusterIP Service that exposes the Deployment on the 6379 port at the internal redis-leader.default.svc.cluster.local network address. The Deployment pod template defines the three app, role, and tier labels with values that are used in the selector definition of the Service to connect the Service with the unique Pod defined in the Deployment.

20:536

Let’s upload the redis-master.yaml file to Azure Cloud Shell, and then deploy it in the cluster with the following command:

20:537

kubectl create -f redis-master.yaml

20:538

20:539

Once the operation is complete, you can inspect the contents of the cluster with kubectl get all.

20:540

The slave storage is defined in the redis-slave.yaml file and is created in the same way, the only difference being that this time we have two replicas, and a different Docker image. The full code is in the GitHub repository associated with this book.

20:541

Let’s upload this file as well and deploy it with the following command:

20:542

kubectl create -f redis-slave.yaml

20:543

20:544

The code for the UI tier is contained in the frontend.yaml file. Deployment has three replicas and a different Service type. Let’s upload and deploy this file with the following command:

20:545

kubectl create -f frontend.yaml

20:546

20:547

It is worthwhile analyzing the Service code in the frontend.yaml file:

20:548

apiVersion: v1

20:549

kind: Service

20:550

metadata:

20:551

name: frontend

20:552

labels:

20:553

app: guestbook

20:554

tier: frontend

20:555

spec:

20:556

type: LoadBalancer

20:557

ports:

20:558

- port: 80

20:559

selector:

20:560

app: guestbook

20:561

tier: frontend

20:562

20:563

Again, the full code is in the GitHub repository associated with the book.

20:564

This Service is of the LoadBalancer type. Since this Pod is the application interface with the world outside of the Kubernetes cluster, its service must have a fixed IP and must be load balanced. Therefore, we must use a LoadBalancer service since this is the unique service type that satisfies those requirements. (See the Services section of this chapter for more information.)

20:565

If you are in Azure Kubernetes or any other cloud Kubernetes service, in order to get the public IP address assigned to the service, and then to the application, use the following command:

20:566

kubectl get service

20:567

20:568

The preceding command should display information on all the installed services. You should find the public IP in the EXTERNAL-IP column of the list. If you see only <none> values, please repeat the command until the public IP address is assigned to the load balancer.

20:569

If no IP is assigned after a few minutes, please verify whether there is some error or warning in any of the service descriptions. If not, please check whether all deployments are actually running using the following command:

20:570

kubectl get deployments

20:571

20:572

If, instead, you are on minikube, LoadBalancer services can be accessed by issuing this command:

20:573

minikube service <service name>

20:574

20:575

Thus, in our case:

20:576

minikube service frontend

20:577

20:578

The command should automatically open the browser.

20:579

Once you get the IP address, navigate with the browser to this address. The application’s home page should now appear!

20:580

If the page doesn’t appear, verify whether any service has an error by issuing the following command:

20:581

kubectl get service

20:582

20:583

If not, also verify that all deployments are in the running state with the following:

20:584

kubectl get deployments

20:585

20:586

If you find problems, please look for errors in the .yaml files, correct them, and then update the object defined in the file with:

20:587

kubectl update -f <file name>

20:588

20:589

20:590

Once you have finished experimenting with the application, make sure to remove the application from the cluster to avoid wasting your free Azure credit (public IP addresses cost money) with the following commands:

20:591

kubectl delete deployment frontend redis-master redis-slave

20:592

kubectl delete service frontend redis-leader redis-follower

20:593

20:594

In the next section, we will analyze other important Kubernetes features.

20:595

Advanced Kubernetes concepts

20:596

In this section, we will discuss other important Kubernetes features, including how to assign permanent storage to StatefulSets; how to store secrets such as passwords, connection strings, or certificates; how a container can inform Kubernetes about its health state; and how to handle complex Kubernetes packages with Helm. All of these subjects are organized into dedicated subsections. We will start with the problem of permanent storage.

20:597

Requiring permanent storage

20:598

Since Pods are moved between nodes, they can’t store data on the disk storage offered by the current node where they are running, or they would lose that storage as soon as they are moved to a different node. This leaves us with two options:

20:599

20:600

Using external databases: With the help of databases, ReplicaSets can also store information. However, if we need better performance in terms of write/update operations, we should use distributed sharded databases based on non-SQL engines such as Cosmos DB or MongoDB (see Chapter 12, Choosing Your Data Storage in the Cloud). In this case, in order to take maximum advantage of table sharding, we need StatefulSets, where each Pod instance takes care of a different table shard.