Gitlab CI + Helm + Kubernetes 構建CI/CD

Gitlab 和 Kubernetes CI/CD流程圖

Gitlab 和 Gitlab CI搭建參考往期文章

• github.com/yangpeng14/…
• github.com/yangpeng14/…

Helm安裝(gitlab runner機器上安裝)

• 安裝目前最新helm 2.16.0版本

$ wget https://get.helm.sh/helm-v2.16.0-linux-amd64.tar.gz
$ tar -zvf helm-v2.16.0-linux-amd64.tar.gz
$ cd linux-amd64/
$ cp helm /usr/local/bin
複製代碼

• 驗證Helm

helm version
Client: &version.Version{SemVer:"v2.16.0", GitTreeState:"clean"}
Server: &version.Version{SemVer:"v2.16.0", itTreeState:"clean"}
複製代碼

• 初始化Helm

$ helm init --client-only
$ helm plugin install https://github.com/chartmuseum/helm-push
複製代碼

• 更新repo爲阿里源

$ helm repo remove stable
$ helm repo add ali https://kubernetes.oss-cn-hangzhou.aliyuncs.com/charts
複製代碼

Gitlab CI 文件配置樣例

stages:
  - test
  - deploy

# test job
job 1:
  stage: test
  only: 
    - master
  tags:
    - tag-test
  script: echo "單元測試"

# deploy 階段把 docker build 和 k8s部署 放在一個階段
job 2:
  stage: deploy
  only: 
    - master
  tags:
    - tag-test
  script:
    - echo "Deploy ..."
    # deploy 本身寫的python部署腳本
    # helm.yaml helm values配置文件
    # product-line 產品線
    # project-name 項目名稱，若是多分支能夠使用 project-name-$CI_COMMIT_REF_NAME
    # $(date "+%Y%m%d%H%M%S")-${CI_COMMIT_SHA:0:6} docker tag
    # Dockerfile 構建項目dockerfile
    - deploy -f helm.yaml product-line project-name $(date "+%Y%m%d%H%M%S")-${CI_COMMIT_SHA:0:6} Dockerfile 
  when: manual
複製代碼

deploy 部署腳本

• 下面是 deploy 部分代碼，獲取腳本所有代碼，請關注個人 YP小站 公衆號並回復 獲取deploy代碼

#!/usr/bin/env python3

import sys
import os
import getopt
import yaml
import pykube

def usage():
    print("Usage: %s [ -b | --build-arg | -f | --file ] product-line project-name docker_images_version dockerfile_name --no-cache" % sys.argv[0], "\n", "or",
    "Usage: %s [ -b | --build-arg | -f | --file ] product-line project-name docker_images_version dockerfile_name" % sys.argv[0], "\n\n",
    "-b, --build-arg 聲明Dockerfile中環境變量，參數能夠指定屢次，示例 A=b, ", "\n",
    "-f, --file 指定Helm values.yaml文件，參數只能指定一次，指定多個也只會取第一個值")

def check_item_exists(project, server_name, docker_images_version, *args):
    api = pykube.HTTPClient(pykube.KubeConfig.from_file("k8s config"))
    deploy = pykube.Deployment.objects(api).filter(namespace=project)
    os.environ['project'] = str(project)
    os.environ['server_name'] = str(server_name)
    os.environ['docker_images_version'] = str(docker_images_version)
    os.environ['s_name'] = str(server_name)
    if args:
        os.environ['helm_values_file'] = str(args[0])

    if os.system('helm status $server_name 1> /dev/null 2>&1') == 0:
        os.system("echo '\033[1;32;40m' 'Helm滾動發佈、超時爲5分鐘、失敗在本次基礎上自動回滾上一個版本。' '\033[0m \n'")
        os.system('helm upgrade --timeout 300 --atomic --install $server_name --namespace $project \ --set serverName=$server_name --set image.project=$project \ --set serverFeatureNameReplace=$server_name \ --set image.tag=$docker_images_version "Helm warehouse address" $helm_values_file')
        return
    else:
        # 兼容過去沒有使用Helm部署
        for deployment in deploy:
            if server_name == str(deployment):
                # deployment進行滾動升級
                os.system('kubectl set image deployment/$server_name \ $server_name=harbor.example.com/$project/$server_name:$docker_images_version \ --namespace=$project')
                return

    # 部署新項目
    # 使用helm模板部署服務
    os.system('helm install --name $server_name --namespace $project \ --set serverName=$server_name --set image.project=$project \ --set serverFeatureNameReplace=$server_name \ --set image.tag=$docker_images_version "Helm warehouse address" $helm_values_file')
    return
複製代碼

參考文獻

blog.csdn.net/ygqygq2/art…

關注我

歡迎你們關注交流，按期分享自動化運維、DevOps、Kubernetes、Service Mesh和Cloud Native