Title

class: center, middle
name: cloudbuild
# CloudBuild2.0
      
## BuildProject

---

# Build as CODE

1. 配置及代码, 更加直观, 任何人都可以参与配置文件的更改
2. 对配置进行版本控制, 可追溯
3. 配置文件更容易分发, 共享，复制  
4. ...

```yaml
# master branch build.yml
version: 2.0
env:
  label: prod-env
steps:
  BUILD:
    - build_execute: 
        command: sh build.sh
```

```yaml
# test branch build.yml
version: 2.0
env:
  label: test-env
steps:
  BUILD:
    - build_execute: 
        command: sh build.sh -p test-options
```
---
class: center, middle
# Quick Demo

---
# Build Job

- Build Params
- Build Environment
- Build Steps

```yaml
version: 2.0
# Build Params
params:
  - name: param1
    value: value1
# Build Environment
env:
  label: CloudBuildVM

# Build Steps
steps:
  PRE_BUILD:
    - checkout
  BUILD:
    - build_execute:
        command: sh build.sh
  POST_BUILD:
    - sh:
        command: |
          # upload to cmc
          # upload to cloud artifact
```

---
# Build Params

构建时候用到的一些变量, 这些变量可以从环境变量中获取, 并支持多种类型

参数来源：
1. build.yml中直接书写
2. 构建的时候动态传入(具有更高优先级)

```yaml
params:
  - name: param1  # 参数名字
    value: value1 # 参数值
  - name: intParam
    value: 123
  - name: stringParam
    value: hello world
  - name: boolParam
    value: true
```

---
# Build Environment

构建环境，用于指定构建时候用到的环境：
1. 镜像
2. 工具
3. ...

如何指定BuildEnv：
1. label
2. playbook(暂不推荐)

```yaml
# use label
env:
  label: some-label
```

```yaml
# use playbook
env:
  resource:
    image:
      builder:
        provider: "build-env"
        playbook_ref: 5g_ran/playbook.xml
```

---
# Build Steps

构建过程中定义的步骤, 步骤包含几个阶段：
1. `PRE_ENV`：环境准备, 如镜像制作，机器制作
2. `CONFIG_ENV`: 环境配置, 如挂载磁盘，执行初始化脚本
3. `PRE_BUILD`: 构建准备，如代码下载，二进制下载
4. `BUILD`：构建阶段，单一入口脚本触发构建
5. `POST_BUILD`：构建后操作，构建清理，构建artifact上传

所有阶段都不是必须的，但是至少要有一个

```yaml
steps:
  PRE_BUILD:
    - checkout # 检出代码
  BUILD:
    - build_execute:
        command: sh build.sh
  POST_BUILD:
    - upload_cloud_artifact:
        path: artifact-*.bin
```

---
# BuildFlow

`Build Job`是不可分割的最小构建单元，在一些复杂场景下，`Build Job`可能不太适用:
1. 多个构建任务需要相互配合(DAG)
2. 加速构建过程，进行分布式构建
3. ...

`BuildFlow`将多个`Build Job`按照构建的依赖顺序进行编排，并按照编排顺序进行调度构建。

```yaml
version: 2.0
params:
  - name: param1
    value: value1
env:
  label: testlabel
buildflow:
  jobs:
    - job: job1
      depends_on:
        - job2
      build_ref: job1/build.yml
    - job: job2
      build_ref: job2/build.yml
```
---
class: center, middle
# Quick BuildFlow Demo
---
# BuildFlow Params
在`BuildFlow`中参数包括：

1. BuildFlow全局参数: 全部的job都能取到
2. BuildFlow的Job的局部参数: 只有定义的Job能够取到
3. Job定义参数: 只有定义的Job可以取到
4. 输入参数：所有的job都可以取到

参数优先级为：输入参数 > Job定义参数 > BuildFlow的Job局部参数 > BuildFlow的全局参数
---

# BuildFlow的全局参数和局部参数

全局参数会被所有的参数共享，局部参数只能被Job独享

```yaml
params: # BuildFlow的全局参数, job1, job2共享
  - name: globalVar1
    value: value1
env:
  label: some-label
buildflow:
  jobs:
    - job: job1
      params: # BuildFlow的Job的局部参数, job1独有
        - name: job1Var
          value: value1
      build_ref: job1/build.yml
    - job: job2
      params: # BuildFlow的Job的局部参数，job2独有
        - name: job2Var
          value: value2
      build_ref: job2/build.yml
    
```
---

# BuildFlow Environment

BuildFlow本身的运行也需要独占一台机器作为调度机, 所以每一次BuildFlow的运行，那么至少需要job数量+1台机器在资源池中。
即:
```scala
count(machine pool) >= count(buildflow jobs) + 1
```

BuildFlow Environment的和Build Job的环境定义没有区别，对环境性能配置也没有特殊要求

```yaml
env:
  label: buildflow-label
```
---

# BuildFlow jobs
BuildFlow的Jobs用于将一组Build Jobs进行编排，换句话说，即指定Build Jobs之间的依赖关系, 最终所有的Build Jobs会组成一个DAG关系图。

通过`depends_on`来指定依赖的job的名字, 通过`build_ref`来指定子构建的构建脚本入口文件

```yaml
buildflow:
  jobs:
  - job: job1
    depends_on:
      - job2
      - job3
    build_ref: job1/build.yml
  - job: job2
    build_ref: job2/build.yml
  - job: job3
    build_ref: job3/bulid.yml
```

```
                                +------+
                                | job3 |
                                +------+-------->-----+
                                                | job1|
                                +------+-------->-----+
                                | job2 |
                                +------+
```
---
# BuildFlow Shared Workspace

BuildFlow中的多个Jobs之间在构建以后可能会存在一些交互共享的操作：

1. 构建的Artifact的共享
2. 构建信息的互通

可以通过指定`attach_workspace`来为所有的BuildFlow Jobs挂载一个共享目录
```yaml
buildflow:
  attach_workspace:
    path: output
  jobs:
  - job: job1
    depends_on:
      - job2
    build_ref: job1/build.yml
  - job: job2
    build_ref: job2/build.yml
```
```
                               +-------+        +-------+
                               |  job1 +--------> job2  |
                               +--+----+        +---+---+
                         +--------v-----------------v------+
                         |        Shared Workspace         |
                         |       .../output/job1/job1.bin  |
                         |       .../output/job2/job2.bin  |
                         +---------------------------------+
```
---
class: center, middle
# Thank You
wiki: http://wiki.inhuawei.com/display/CloudBuild2/2.9+Build2.0+Reference