Docker安装GitLab与Runner（网关），常规设置，自动化用到k8s+token
[转]图文详解k8s自动化持续集成之GitLab CI/CD
Windows里面使用Debian命令行工具完成
和Docker网络相关的命令
查看某一个容器的网络 docker inspect 容器ID 查看docker当前网络 docker network ls 首先创建一个网络（下面通过docker-compose的yml文件创建，这里只做命令的了解） docker network create gitlab-network 将现有的容器连到创建的网络中（每个容器都要连到这个网络里） docker network connect gitlab-network gitlab-server #容器名 docker network connect gitlab-network gitlab-runner 查看网络内的容器信息 docker network inspect gitlab-network 也可以在运行时指定镜像 docker run --network gitlab-network 镜像名称 移除指定的网络 docker network rm gitlab-network
先删掉下面要用的bridge名称
使用Docker Compose安装GitLab-ce和GitLab-runner
Docker Compose で GitLab + GitLab Runner の環境を整える(Window10) 重要参考
Win10下使用Docker搭建Gitlab CI自动构建平台
docker rm & docker rmi & docker prune 的差异
- docker rm ： 删除一个或多个 容器
- docker rmi : 删除一个或多个 镜像
- docker prune： 用来删除不再使用的 docker 对象
docker+gitlab+gitlab-runner部署(CentOs7)重要参考
Docker Compose yml for Gitlab and Gitlab Runner
SMTP settings(备用参考)
#vim /etc/gitlab/gitlab.rb
gitlab_rails['smtp_enable'] = true
gitlab_rails['smtp_address'] = "smtp.exmail.qq.com"
gitlab_rails['smtp_port'] = 465
gitlab_rails['smtp_user_name'] = "developers@aaa.net"
gitlab_rails['smtp_password'] = "Del43@\#$@1"
gitlab_rails['smtp_domain'] = "exmail.qq.com"
gitlab_rails['smtp_authentication'] = "login"
gitlab_rails['smtp_enable_starttls_auto'] = true
gitlab_rails['smtp_tls'] = true
gitlab_rails['gitlab_email_from'] = "developers@aaa.net"
测试邮件
#gitlab-ctl reconfigure
#gitlab-rails console
#Notify.test_email('315360007@qq.com', 'Message Subject', 'Message Body').deliver_now
配置本地windows的host文件：
127.0.0.1 gitlab-server
127.0.0.1 nexus3
浏览器访问：http://gitlab-server:9080/ 或者 http://ip:9080/
用户名：root，密码：自己设置
用root用户登录
创建测试项目：mvcdockerrunner1
添加SSH密钥
本地windows需要在当前登录用户的目录里面找到.ssh目录下面的****.pub文件添加公钥，同时设置config文件
Host gitlab-server HostName gitlab-server User git PreferredAuthentications publickey IdentityFile ~/.ssh/315360007 Port 23
克隆代码
手动设置激活的Runner(很重要)
1、注册gitlab-runner（按照提示一步一步做），熟悉了直接用下面第二种
docker exec -it gitlab-runner gitlab-runner register
2. 我们会输入 http://gitlab.local.net:9080 或 http://ip:port 也就是我们安装在本地的GitLab
3. Please enter the gitlab-ci token for this runner 要求输入 gitlab-ci token
在项目的 管理区域->runners中可以找到(这里注册的是share类型runner)
4. 输入描述,如：gitlab-runner
5. 输入tag（留空也可以，之后可以进行编辑 ）
6. 选择当遇到没有打标签的提交时是否会执行，我们选 true
7. 是否锁定此runner 到当前项目， 我们选 false
8. 选一个执行者 executor
这一步比较重要 (ssh, docker+machine, docker-ssh+machine, kubernetes, docker, parallels, virtualbox, docker-ssh, shell)
我们选docker
9. 选择默认使用的镜像： docker:stable
2、注册gitlab-runner（直接传参数一次完成）
# 手动设置激活的Runner docker exec -it gitlab-runner gitlab-runner register \ --non-interactive \ --executor "docker" \ --docker-image docker:stable \ --url "http://gitlab-server:9080/" \ --registration-token "K6BiurzDBuzx23a-mV-f" \ --description "gitlab-runner" \ --tag-list "docker,company" \ --run-untagged="true" \ --locked="false" \ --docker-privileged="true" \ --docker-extra-hosts "gitlab-server:172.20.0.1" \ --docker-extra-hosts "nexus3:172.20.0.1" #解释很重要 #"docker:stable"这里之所以这样是因为基于这个镜像它包含了docker等工具，可以在gitlab-runner执行的.gitlab-ci.yml中有docker指令时 #而无需再安装docker 参考地址：https://docs.gitlab.com/ee/ci/docker/using_docker_build.html #privileged=true #使用docker-in-docker时通常为true #extra-hosts #设置gitlab-server的网关地址
进入 gitlab-runner的容器里
docker exec -it runner-container-id /bin/bash
查看/etc/gitlab-runner/config.toml文件
cat /etc/gitlab-runner/config.toml
concurrent = 1
check_interval = 0
[session_server]
session_timeout = 1800
[[runners]]
name = "gitlab-runner"
url = "http://gitlab-server:9080/"
token = "mWpRt6ry7HimmAyt86T1"
executor = "docker"
[runners.custom_build_dir]
[runners.cache]
[runners.cache.s3]
[runners.cache.gcs]
[runners.cache.azure]
[runners.docker]
tls_verify = false
image = "docker:stable"
privileged = true
disable_entrypoint_overwrite = false
oom_kill_disable = false
disable_cache = false
volumes = ["/cache"]
extra_hosts = ["gitlab-server:172.20.0.1"]
shm_size = 0
五、提交项目代码完成CI/CD
Gitlab CI/CD 打包过程报 “ERROR: Uploading artifacts to coordinator... too large” 如何解决
Uploading artifacts for successful job 00:13 Uploading artifacts... mevdockerrunner1.tar: found 1 matching files and directories ERROR: Uploading artifacts as "archive" to coordinator... too large archive id=34 responseStatus=413 Request Entity Too Large status=413 token=fhcnwRyE FATAL: too large
GitLab管理中心里设置CI/CD持续集成和部署里最大产物大小
设置和使用Gitlab-ci变量连接Kubernetes：
1 .kube_deploy: 2 image: boxboat/kubectl:1.17.4 3 variables: 4 HELM_TOKEN: ${HELM_USER_TOKEN} 5 K8S_API_SERVER: ${KUBE_API_ADDRESS} 6 K8S_API_SERVER_CERT: ${KUBE_API_CA} 7 before_script: 8 - mkdir -p /home/alpine/.kube 9 - cp .ci/kube-config-template /home/alpine/.kube/config 10 - sed -i "s/{{ HELM_USER_TOKEN }}/${HELM_TOKEN}/g" /home/alpine/.kube/config 11 - sed -i "s/{{ KUBE_API_ADDRESS }}/${K8S_API_SERVER}/g" /home/alpine/.kube/config 12 - sed -i "s/{{ KUBE_API_CA }}/${K8S_API_SERVER_CERT}/g" /home/alpine/.kube/config
设置Kubernetes的config文件，kube-config-template文件内容：
1 apiVersion: v1 2 clusters: 3 - cluster: 4 certificate-authority-data: {{ KUBE_API_CA }} 5 server: https://{{ KUBE_API_ADDRESS }} 6 name: helm 7 contexts: 8 - context: 9 cluster: helm 10 namespace: meshop-dev 11 user: helm 12 name: helm 13 current-context: "helm" 14 kind: Config 15 preferences: {} 16 users: 17 - name: helm 18 user: 19 token: {{ HELM_USER_TOKEN }}
获取KUBE_API_CA和HELM_USER_TOKEN的值
#kubectl get sa -A #查看所有security account #kubectl get sa/helm -n kube-system -o yaml #查看在某一个命名空间下的某一个security account=helm的yaml，得到secrets的name #kubectl get secrets/helm-token-wstgs -o yaml -n kube-system #查看在某一个命名空间下的某一个secrets的name的yaml，得到token值和ca.crt值
使用变量
设置默认分支：
设置仓库不允许推送代码，只能自己创建分支提交，在服务器上合并分支
git流水线克隆代码
流水线触发器的设置，添加一个token
设置sourceTree自定义操作，设置token=上面的值
1 curl -X POST -F token=f21cbb2d6f8c10cad915a380c49b99 -F 'ref=dev' -F 'variables[MESHOP_BUILD_ONLY]=sso' https://git.runshopstore.com/api/v4/projects/2/trigger/pipeline
设置ci里面使用触发器
1 workflow: 2 rules: 3 - if: $MESHOP_BUILD_ONLY != null 4 when: always 5 - if: $CI_PIPELINE_SOURCE =~ /^trigger|pipeline|web|api$/ 6 when: always
合并请求必须保证编译通过：同时删除合并请求的分支
设置关闭仓库，测试可以使用的项目：
设置标记：
规范化git commit信息
团队敏捷实践 —— 使用 semantic-release 自动管理发布版本
单个项目最简单的自动化脚本（Config项目）
stages:
- restore
- compile
- version
- build
- deploy
.load_environment_variables: &load_environment_variables
- echo "init environment variables."
- for environment in $(ls -d -1 .ci/environments/*.sh);
do source $environment;
done
default:
image: docker.tidebuy.net/dotnet/sdk:5.0
tags:
- docker
- company
before_script:
- *load_environment_variables
variables:
# docker in docker
DOCKER_DRIVER: "overlay2"
DOCKER_HOST: tcp://localhost:2375
# cache
NUGET_PACKAGES_DIRECTORY: ".nuget"
OBJECTS_DIRECTORY: "obj"
SOURCE_CODE_PATH: "*/*/"
.nuget_cache:
cache:
key: "${CI_PROJECT_NAME}-${CI_COMMIT_REF_SLUG}"
paths:
- "$SOURCE_CODE_PATH$OBJECTS_DIRECTORY/project.assets.json"
- "$SOURCE_CODE_PATH$OBJECTS_DIRECTORY/*.csproj.nuget.*"
- "$NUGET_PACKAGES_DIRECTORY"
restore:
stage: restore
extends: .nuget_cache
cache:
policy: pull-push
script:
- echo "dotnet restore..."
- dotnet restore
$CI_PROJECT_DIR/MeShop.Config/MeShop.Config.sln
--packages $NUGET_PACKAGES_DIRECTORY
--runtime linux-x64
--configfile $CI_PROJECT_DIR/.ci/nuget.config
only:
refs:
- branches
- merge_requests
compile:
stage: compile
before_script:
- *load_environment_variables
extends:
- .nuget_cache
cache:
policy: pull
script:
- echo "compile project."
- dotnet publish MeShop.Config/MeShop.View.Config/MeShop.View.Config.csproj --no-restore
--runtime linux-x64 -c Release -o ${CI_PROJECT_DIR}/publish
artifacts:
name: "${CI_JOB_NAME}-${CI_COMMIT_REF_NAME}"
paths:
- publish/
expire_in: 1 days
only:
refs:
- branches
- merge_requests
version:
stage: version
image: meshop/semantic-release:17.1.1
script:
- if [ $CI_COMMIT_REF_NAME == 'alpha' -o $CI_COMMIT_REF_NAME == 'master' ];
then
semantic-release;
fi
- if [ ! -f ./.version ];
then
echo "VERSION=${MESHOP_BUILD_VERSION:-$(date +%Y%m%d%H%M%S)}">.version;
fi
- cat .version
artifacts:
paths:
- .version
expire_in: 1 days
needs:
- job: compile
artifacts: false
rules:
- if: "$CI_PIPELINE_SOURCE == 'merge_request_event'"
when: never
- if: "$CI_COMMIT_REF_NAME == 'dev'"
when: on_success
- if: "$CI_COMMIT_REF_NAME == 'alpha'"
when: on_success
- if: "$CI_COMMIT_REF_NAME == 'master'"
when: on_success
- if: "$CI_COMMIT_REF_NAME =~ /.*-bug-fix/"
when: on_success
build:
stage: build
image: "docker:stable"
services:
- docker:stable-dind
variables:
DOCKER_TLS_CERTDIR: ""
script:
- echo "Logging to GitLab Container Registry with CI credentials..."
- echo "$MESHOP_BUILD_REGISTRY_PASSWORD" | docker login -u "$MESHOP_BUILD_REGISTRY_USER" --password-stdin "$MESHOP_BUILD_REGISTRY"
- source ./.version
- image_version="${VERSION}"
- image_name="${MESHOP_BUILD_REGISTRY}/meshop/shop/config"
- docker build -f .ci/Dockerfile
--tag "$image_name:$image_version"
.
- docker push "$image_name:$image_version"
- if [ $MESHOP_BUILD_LATEST = 'true' ];
then
docker tag $image_name:$image_version $image_name:latest &&
docker push "$image_name:latest" ;
fi
needs:
- job: version
artifacts: true
- job: compile
artifacts: true
rules:
- if: "$CI_PIPELINE_SOURCE == 'merge_request_event'"
when: never
- if: "$CI_COMMIT_REF_NAME == 'dev'"
when: on_success
- if: "$CI_COMMIT_REF_NAME == 'alpha'"
when: on_success
- if: "$CI_COMMIT_REF_NAME == 'master'"
when: on_success
- if: "$CI_COMMIT_REF_NAME =~ /.*-bug-fix/"
when: on_success
deploy:
stage: deploy
image: boxboat/kubectl:1.17.4
script:
- mkdir -p /home/alpine/.kube
- cp .ci/kube-config-template /home/alpine/.kube/config
- sed -i "s/{{ HELM_USER_TOKEN }}/${HELM_USER_TOKEN}/g" /home/alpine/.kube/config
- sed -i "s/{{ KUBE_API_ADDRESS }}/${KUBE_API_ADDRESS}/g" /home/alpine/.kube/config
- sed -i "s/{{ KUBE_API_CA }}/${KUBE_API_CA}/g" /home/alpine/.kube/config
- kubectl rollout restart deployment/meshop-config
rules:
- if: "$CI_PIPELINE_SOURCE == 'merge_request_event'"
when: never
- if: "$CI_COMMIT_REF_NAME == 'dev'"
when: on_success
