Introduction and basic construction of kubernetes

Introduction

Microservices are very popular now , It's on fire Docker. Because microservices are very suitable for Docker Container implementation , Each container hosts a service . A computer running multiple containers at the same time , So we can easily simulate the complex microservice architecture . So Docker The popularity of , Speed , It's jaw dropping .

It's just Docker When the technology of container is so hot , We found , If you want to Docker Apply to specific business implementation , There are difficulties —— layout 、 Management and scheduling, etc , It's not easy . therefore , There is an urgent need for a management system , Yes Docker And containers for more advanced and flexible management .

Right now ,K8S There is .

K8S, It is a container based cluster management platform , Its full name , yes kubernetes.

Kubernetes It's Google open source Docker Container cluster management system , It mainly provides resource scheduling for containerized services 、 Capacity expansion 、 Capacity reduction and deployment .

Actually from Kuberbetes You can see its purpose , The word comes from ancient Greece , It means helmsman ,Docker Of logo It's a whale , that Kubernetes Is where whales live - Ocean leader ; It was born a little late ,15 year 7 The first version only appears in January , But it was very popular at birth , Many of them include IBM、 Red hat 、 Microsoft and other giants have joined in .

Now let me show you Kubernetes, And use him to achieve revolutionary hello,world!

Kubernetes Has been the predecessor of snow

Over the past decade , Google has always had a secret weapon ：Borg.Borg It has always been a large-scale cluster management system used internally by Google ,Borg It is based on container technology , It is mainly to realize the automation of resource management and improve the utilization of resources in the context of spanning multiple data centers , Google has always looked at Borg For heirloom , Even if the employee leaves the company, it cannot be disclosed to the outside world , until 2015 Google only made it public in , in other words Kubernetes It's actually Borg An open source version of .

Use Kubernetes？ Give me a reason ！

First of all , convenient . If our system design follows Kubernetes Design idea , Then the underlying code in the traditional system architecture has nothing to do with the business 、 We don't need to consider functional modules , We don't have to spend energy on the selection, deployment and implementation of load balancer , There is no need to consider introducing or even developing a complex service governance framework , Not to mention the headache ： Service monitoring 、 Troubleshooting, etc . All in all ,Kubernetes Is to liberate our energy , So that we can focus more on the business itself to do development .

second , to open up .Kubernetes Compatible with all languages , All programming interfaces are also compatible , No matter what language you write, it can be mapped to Kubernetes Of Service And pass the standard TCP Communication protocol for interaction , The bigger benefit is , Existing systems can also be upgraded and migrated to Kubernetes On the platform ;

Third , Powerful . After more than ten years of development ,Kubernetes It has become a very complete distributed system support platform , Have extremely strong cluster management capabilities ： Multi level security protection and access mechanism 、 Multi-tenant application support capability 、 Transparent service registration and service discovery mechanism 、 Built in intelligent load balancer 、 It can also find faults in time and repair itself 、 Service rolling upgrade and online capacity expansion 、 Extensible automatic resource scheduling mechanism 、 Multi granularity resource quota management …… That's too much , To name but a few ！

IT We all know that technology changes faster than you change clothes ,Docker The star of containerization technology has become so popular , From single machine to cluster, this is an irreversible trend , Cloud computing is also in full swing ,Kubernetes It is the only one recognized by the industry so far Docker Distributed system solutions .

You have to know this before you start ！

But don't worry , You must have some knowledge before you start again , It's like learning Java I didn't just write when I came up hello,world But first learn some and or not 、 Data types and so on .

stay Kubernetes in Service Is the core of the distributed cluster architecture , One Service It has the following key characteristics ：

1. Have a location specific name , Just like the primary key in the database ;

2. Have a virtual IP And port number ;

3. Have the ability to provide remote services and be able to map to a set of container applications that provide such services ;

at present ,Service It's all based on Socket Communication mode or a specific service that implements a specific service TCP Server Process to provide services , Although a Service Services are usually provided by multiple related service processes , And each service process has an independent EndPoint(IP + Port) Access points , however ！ however Kubernetes Can let us through the virtual Cluster IP + Service Port Connect to the specified Service On .

With Kubernetes Built-in transparent load balancing and failover mechanisms , Even if there are more service processes on the back end , Even more service processes will be redeployed to other machines due to failures , Will not affect our normal call to the service ;

Kubernetes Once created, it will not change , It also means that , stay Kubernetes In the cluster, we no longer need to IP The frequent change of address caused a headache .

One of the benefits of containers is that they have powerful isolation capabilities , Therefore, it must be regarded as Service The group of processes that provide services are put into containers for isolation ,Kubernetes Of Pod The object is to do this , Each service process is wrapped in a corresponding Pod Make it Pod The one running in Container( Containers ).

In order to establish Service and Pod Relationship between ,Kubernetes Will give each Pod A close one Label( label ), And then give them the corresponding Service Definition Label Selector( tag chooser ）.

that ,Pod What kind of thing is it ？

Pod Running on the Node( node ) in , This node can be either a physical machine or a virtual machine in a public cloud or a private cloud , Usually there are hundreds running on a node Pod.

Every Pod There is a program called Pause And some business containers , In order to improve the efficiency of communication and data exchange , These business containers share Pause The container's network stack and Volume Mount the volume , We can use this feature to put a group of closely related service processes into the same Pod in .

however , There is one caveat ： Not every one Pod And the container inside it can “ mapping ” To a Service On , Just a group that provides external or team services Pod can “ mapping ” Become a service .

In terms of cluster management ,Kubernetes Divide the machines in the cluster into one Master Nodes and a group of work nodes , The former runs processes related to cluster management , The work node runs the application .

stay Kubernetes In the cluster , To expand the capacity, you only need to work for Service The associated Pod Create a Replication Controller,Replication Controller The file contains at least the following information ：

1. The goal is Pod Definition ;

2. The goal is Pod Number of copies to run ;

3. The target to monitor Pod The label of ;

Create good Replication Controller after .Kubernetes Will pass Replication Controller To filter out the corresponding Pod Instance and monitor the status and quantity in real time , If the number of instances is less than the number of copies, it will be based on Replication Controller As defined in Pod Template to create a new one Pod, Put this again. Pod Dispatch to the appropriate Node Upstart operation , until Pod The number of instances reaches the predetermined target .

Everything must be done from hello,world Start

here hello,world It's a Web Message board application , It's based on PHP + Redis Two tier distributed architecture of Web application , This is the system deployment architecture diagram of the message board ：

Web System deployment architecture diagram of message board

Three Docker Mirror image ：

1. redis-master： For the front end Web Application to write a message , One of them has been saved “hello,world”.

2. guestbook-redis-slave： For the front end Web Application to read messages , and redis-master Keep your data in sync .

3. guestbook-php-frontend：PHPWeb service , Display the message content on the web page , At the same time, a text input box is provided for visitors to add comments .

This is a Kubernetes Deployment frame diagram ：

Kubernetes Deployment frame diagram

Now let's start hand-in-hand teaching you how to use k8s Building a hello,world application . Be careful , This article USES centos operating system .

First , We need to enable the routing and forwarding function , If you do not perform this step, the container cannot access the Internet ：

echo 1 > /proc/sys/net/ipv4/ip_forward 

then , We're configuring yum Source ：

yum install -y epel-release
yum clean all
yum list

We'll go on with the , We installed Docker.

yum install -y docker-io

Docker After installation , We started to build Kunernetes Running environment .

##  Turn off firewall 
systemctl disable firewalld
suystemctl stop firewalld
##  Install again  etcd  and  Kubernetes  Software 
yum install -y etcd
yum install -y kubernetes

And then modify Docker Configuration file for /etc/sysconfig/docker,OPTIONS Revised as follows :

OPTIONS='--selinux-enabled=false --insecure-registry gcr.io'

Be careful ,Docker Domestic image source is recommended . The general method is to edit /etc/docker/daemon.json：

{
  "registry-mirrors" : [
    "http://ovfftd6p.mirror.aliyuncs.com",
    "http://registry.docker-cn.com",
    "http://docker.mirrors.ustc.edu.cn",
    "http://hub-mirror.c.163.com"
  ],
  "insecure-registries" : [
    "registry.docker-cn.com",
    "docker.mirrors.ustc.edu.cn"
  ],
  "debug" : true,
  "experimental" : true
}

And then restart Docker Of daemon that will do .

systemctl daemon-reload
systemctl restart docker

Then modify Kubernetes Of apiserver The configuration file /etc/kubernetes/apiserver:

take –admission_control In the parameter ServiceAccount Delete .

Then start all the services in sequence ：

systemctl start etcd
systemctl start docker
systemctl start kube-apiserver
systemctl start kube-controller-manager
systemctl start kube-scheduler
systemctl start kubelet
systemctl start kube-proxy

Check whether the above services have been started successfully , If all have been successful kubernetes The cluster environment has been installed .

by Redis-master Service creation RC file , The grammar format is ymal.

apiVersion: v1
kind: ReplicationController
metadata:
  name: redis-master
  labels:
    name: redis-master
spec:
  replicas: 1
  selector:
    name: redis-master
  template:
    metadata:
     labels:
       name: redis-master
    spec:
      containers:
      - name: master
        image: kubeguide/redis-master
        ports:
        - containerPort: 6379

And then in Master Executing commands in nodes ：Kubectl create -f<config_file>, And publish it to Kubernetes In the cluster ：

$ kubectl create -f redis-master-controller.yaml
replicationcontrollers/redis-master

And then create the Service.service Corresponding file redis-master-service.yaml, The contents are as follows ：

apiVersion: v1
kind: Service
metadata:
  name: redis-master
  labels:
    name: redis-master
spec:
  ports:
  - port: 6379
    targetPort: 6379
  selector:
    name: redis-master

establish redis-master service.

kubectl create -f redis-master-service.yaml 

next , To create a redis-slave service , Same as the former , First create redis-slave Of RC The definition file  redis-salve-controller.yaml.

apiVersion: v1
kind: ReplicationController
metadata:
  name: redis-slave
  labels:
    name: redis-slave
spec:
  replicas: 2
  selector:
    name: redis-slave
  template:
    metadata:
     labels:
       name: redis-slave
    spec:
      containers:
      - name: slave
        image: kubeguide/guestbook-redis-slave
        env:
        - name: GET_HOSTS_FROM
          value: env
        ports:
        - containerPort: 6379

Then create redis-salve Corresponding Pod.

kubectl create -f redis-salve-controller.yaml

Pod After creation , Create the corresponding service. First create service Configuration file for redis-salve-service.yaml, The contents are as follows ：

apiVersion: v1
kind: Service
metadata:
  name: redis-slave
  labels:
    name: redis-slave
spec:
  ports:
  - port: 6379
  selector:
    name: redis-slave

Then according to the configuration file , establish service.

kubectl create -f redis-slave-service.yaml

In order to achieve Redis The master-slave data synchronization of the cluster ,redis-slave Need to know redis-master The address of , So we need to be able to redis-slave Start command of image /run.sh Add ：

redis-server --slaveof ${REDIS_MASTER_SERVICE_HOST} 6379

Last , establish frontend Of RC file frontend-controller.yaml, The contents are as follows ：

apiVersion: v1
kind: ReplicationController
metadata:
  name: frontend
  labels:
    name: frontend
spec:
  replicas: 3
  selector:
    name: frontend
  template:
    metadata:
     labels:
       name: frontend
    spec:
      containers:
      - name: frontend
        image: kubeguide/guestbook-php-frontend
        env:
        - name: GET_HOSTS_FROM
          value: env
        ports:
        - containerPort: 80

And then it's running kubectl create Command to create RC：

$ kubectl create -f frontend-controller.yaml
replcationcontrollers/frontend

Then create the associated Service,frontend-service.yaml The contents are as follows :

apiVersion: v1
kind: Service
metadata:
  name: frontend
  labels:
    name: frontend
spec:
  type: NodePort
  ports:
  - port: 80
    nodePort: 30001
  selector:
    name: frontend

The last step , establish service.

$ kubectl create -f frontend-service.yaml
services/frontend

Now you can open the browser to enter ：http:// virtual machine IP:30001  And then there will be hello,world, Success, that's it ：

Web Effect drawing of message board interface operation

summary

This is it. Kubernetes, Realized hello,world Even if a foot came in ,Kubernetesde There are so many advantages to the market , The most direct benefit is that you can develop complex systems with light equipment , Architects focus on “ Service components ”, The rest are responsible for the development of business code , and Kubernetes It is also a powerful tool for implementing microservices , Microservices are actually the process of breaking a big Mac into small units ,Kubernetes To do this is undoubtedly the best person ！