GitLab-Pipeline
一、Pipeline开发工具
- 流水线视图
- 流水线编辑器
- Job作业视图
- Schedules 定时运行
- 环境变量
1.1 流水线视图
1.2 Schedule
1.3 环境变量
二、CI/CD设置
- 全局配置
- 是否公开访问Pipeline信息；
- 是否自动取消相同分支已经在构建的Pipeline；
- 部署Job成功后，跳过其他还未运行的部署作业；
- ci文件的本地或远程位置；
- 为项目设置流水线状态标志
skip ci
三、Pipeline语法
3.1 Stages
If stages is not defined in the .gitlab-ci.yml file, the default pipeline stages are:
设置settiing CI/CD General pipelines - CI/CD configure file
ci/stages.yml
pipeline editor: 编写新的pipeline
顺序运行
修改 gitlab-runner - config.toml 配置
concurrent = 10
stages:
- .pre
- build
- test
- deploy
- .post
job0:
tags:
- go
stage: .pre
script:
- echo "init"
job1:
tags:
- go
stage: build
script:
- echo "build"
job2-1:
tags:
- go
stage: test
script:
- echo "test1"
job2-2:
tags:
- go
stage: test
script:
- echo "test2"
job3:
tags:
- go
stage: deploy
script:
- echo "deploy"
job4:
tags:
- go
stage: .post
script:
- echo "end"
3.2 variables
variables:
GIT_DEPTH: 0 # 禁用浅克隆，完整拉取仓库
GIT_STRATEGY: clone # 强制完整克隆（而不是 fetch）
GIT_FETCH_EXTRA_FLAGS: "--no-tags" # 解决 fetch 兼容性问题
BUILD_TOOLS: "mvn"
RUNNER_TAG: "go"
stages:
- build
job1:
tags:
- ${RUNNER_TAG}
stage: build
variables:
BUILD_TOOLS: "gradle22255"
script:
- echo $BUILD_TOOLS
3.3 jobs
| 语法关键字 | 作用 | 备注 |
|---|---|---|
| variables | 定义作业中的环境变量 | |
| tags | 根据标签选择运行作业的构建节点 | - 多个标签匹配所有标签构建节点； - 标签可以使用变量 |
| stage | 指定当前作业所属的阶段名称 | |
| before_script | 作业在运行前执行的shell命令行 | |
| script | 作业在运行中执行的shell命令行 | 每个作业至少要包含一个script |
| after_script | 作业在运行后执行的命令行 |
stages:
- build
variables:
RUNNER_TAG: "go"
before_script:
- echo "global pipeline before script"
after_script:
- echo "all done"
job1:
tags:
- $RUNNER_TAG
stage: build
before_script:
- echo "before script ready"
script:
- echo "mvn package"
after_script:
- echo "after sciprt done"
job2:
tags:
- $RUNNER_TAG
stage: build
script:
- echo 'job2'
如果job没有 定义before ，after script，会使用全局的
3.3.1 job运行控制 allow_failure和when
| 关键字 | 作用 | 备注 |
|---|---|---|
| allow_failure | 控制作业状态，是否允许作业失败，默认值为false。启用后，如果作业运行失败，该作业将在用户界面中显示橙色警告。 | 管道认为作业成功/通过，不会别阻塞。假设所有其他作业均成功，则该作业的阶段及其管道将显示相同的橙色警告。但是，关联的提交将被标记为“通过”，而不会发出警告。 |
| when | 根据状态控制作业运行，当前面作业成功或者失败时运行。 | on_success前面阶段成功时执行(默认值); on_failure前面阶段失败时运行 always总是执行 manual手动执行 delayed延迟执行 never永不执行 |
stages:
- build
variables:
RUNNER_TAG: "go"
before_script:
- echo "global pipeline before script"
after_script:
- echo "all done"
job1:
tags:
- $RUNNER_TAG
stage: build
allow_failure: true
before_script:
- echo "before script ready"
script:
- echo "mvn package"
- mvs
after_script:
- echo "after sciprt done"
job2:
tags:
- $RUNNER_TAG
stage: build
script:
- echo 'job2'
stages:
- build
- test
variables:
RUNNER_TAG: "go"
before_script:
- echo "global pipeline before script"
after_script:
- echo "all done"
job1:
tags:
- $RUNNER_TAG
stage: build
allow_failure: false
before_script:
- echo "before script ready"
script:
- echo "mvn package"
- mvs
after_script:
- echo "after sciprt done"
job2:
tags:
- $RUNNER_TAG
stage: test
when: on_success
script:
- echo 'job2'
job3:
tags:
- $RUNNER_TAG
stage: test
when: on_failure
script:
- echo 'job3 is running job2 skipped'
manual 必须前面所有的阶段都是success
stages:
- build
- test
- deploy
variables:
RUNNER_TAG: "go"
before_script:
- echo "global pipeline before script"
after_script:
- echo "all done"
job1:
tags:
- $RUNNER_TAG
stage: build
allow_failure: true
before_script:
- echo "before script ready"
script:
- echo "mvn package"
- mvs
after_script:
- echo "after sciprt done"
job2:
tags:
- $RUNNER_TAG
stage: test
when: on_success
script:
- echo 'job2 is running'
job3:
tags:
- $RUNNER_TAG
stage: test
when: on_failure
script:
- echo 'job2 is running job3 skipped'
job4:
tags:
- $RUNNER_TAG
stage: deploy
when: manual
script:
- echo 'manual run'
3.3.2 job运行控制retry和timeout
| 关键字 | 作用 | 备注 |
|---|---|---|
| retry | 作业重新运行，遇到错误重新运行的次数 | 值为整数 等于或大于0，但小于等于2 异常分类 |
| timeout | 作业运行超时时间 | |
| rules | 根据特定的变量或文件变更来控制作业运行； | if changes exists |
| needs | 作业依赖控制 | needs:["作业名称"] |
| parallel | 生成多个作业，并行运行 | parallel：5 值2-50之间 |
stages:
- build
- test
- deploy
variables:
RUNNER_TAG: "go"
job1:
tags:
- $RUNNER_TAG
stage: build
retry: 2
script:
- echo "build task begin"
- mvs
job2:
tags:
- $RUNNER_TAG
stage: test
script:
- echo "test task begin"
配合条件使用，job2 不会retry
stages:
- build
- test
- deploy
variables:
RUNNER_TAG: "go"
job1:
tags:
- $RUNNER_TAG
stage: build
retry: 2
script:
- echo "build task begin"
- mvs
job2:
tags:
- $RUNNER_TAG
stage: test
when: on_failure
script:
- echo "test task begin"
- mvs
retry:
max: 2
when: api_failure
# when: script_failure job2 会重试
3.4 rules
GitLab CI 的 rules 是从上到下匹配，匹配到第一个条件就停止，不再继续往后判断。
job1 不会运行
stages:
- build
variables:
RUNNER_TAG: "go"
SKIP_BUILD: "false"
job1:
tags:
- $RUNNER_TAG
stage: build
rules:
- if: $SKIP_BUILD != true
when: never
- when: always
script:
- echo "mvn packages"
job2:
tags:
- $RUNNER_TAG
stage: build
script:
- echo "job2"
运行job1
stages:
- build
variables:
RUNNER_TAG: "go"
SKIP_BUILD: "false"
job1:
tags:
- $RUNNER_TAG
stage: build
rules:
- if: $SKIP_BUILD == "true"
when: never
- when: always
script:
- echo "mvn packages"
job2:
tags:
- $RUNNER_TAG
stage: build
script:
- echo "job2"
changes 文件更改了触发job
stages:
- build
- test
variables:
RUNNER_TAG: "go"
SKIP_BUILD: "false"
job1:
tags:
- $RUNNER_TAG
stage: build
rules:
- if: $SKIP_BUILD != "true"
when: never
- when: always
script:
- echo "mvn packages"
job2:
tags:
- $RUNNER_TAG
stage: test
rules:
- changes:
- README.md #文件更改了触发job
when: always
when: never
script:
- echo "job2"
job3:
tags:
- $RUNNER_TAG
stage: test
script:
- echo "job3"
变更文件触发
stages:
- build
- test
variables:
RUNNER_TAG: "go"
SKIP_BUILD: "false"
job1:
tags:
- $RUNNER_TAG
stage: build
rules:
- if: $SKIP_BUILD != "true"
when: never
- when: always
script:
- echo "mvn packages"
job2:
tags:
- $RUNNER_TAG
stage: test
rules:
- changes:
- README.md #文件更改了触发job
when: always
- when: never
script:
- echo "job2"
job3:
tags:
- $RUNNER_TAG
stage: test
rules:
- exists:
- Dockerfile
when: manual
- when: never
script:
- echo "job3"
job4:
tags:
- $RUNNER_TAG
stage: test
script:
- echo "job4"
# 只有job4运行
增加Dockerfile
手动运行job3
