DevOps

A few weeks ago we covered Conditionals in YAML to show how to conditionally run tasks and jobs as well as how to make a job dependent on another job. This post is going to cover combing conditional and job dependencies. If you are new to this series you can use the following posts to catch up.

Getting Started with Azure DevOps
Pipeline Creation in Azure DevOps
Azure DevOps Publish Artifacts for ASP.NET Core
Azure DevOps Pipelines: Multiple Jobs in YAML
Azure DevOps Pipelines: Reusable YAML
Azure DevOps Pipelines: Use YAML Across Repos
Azure DevOps Pipelines: Conditionals in YAML
Azure DevOps Pipelines: Naming and Tagging
Azure DevOps Pipelines: Manual Tagging

Existing Job

As a reminder, our sample pipeline has 4 jobs. The WebApp1 job always runs, the WebApp2 job is run or skipped based on a pipeline variable, the DependentJob depends on WebApp1 and WebApp2, and finally, the TagSources job is dependent on all the previous jobs. We are going to be tweaking the TagSources job in this post. The following is the YAML for the setup of the TagSources job without its tasks.

- job: TagSources
  displayName: 'Tag Sources'
  pool:
    vmImage: 'ubuntu-latest'

  dependsOn:
  - WebApp1
  - WebApp2
  - DependentJob

With this setup WebApp1, WebApp2, and DependentJob all have to report successful or the TagSources job will be skipped. The following screenshot shows a pipeline run with the variable to build WebApp2 set to false.

As you can see the TagSources job was skipped because one of its dependent jobs was skipped.

Dependencies with Conditions

Let us say for our pipeline we want the TagSources job to run as long as all jobs were successful or if WebApp1 was successful and the WebApp2 and DependentJob jobs were skipped. To do this we are going to add a condition element and manually check the results of the dependencies as you can see in the following.

- job: TagSources
  displayName: 'Tag Sources'
  pool:
    vmImage: 'ubuntu-latest'

  dependsOn:
  - WebApp1
  - WebApp2
  - DependentJob
  condition: |
    and
    (
      eq(dependencies.WebApp1.result, 'Succeeded'),
      in(dependencies.WebApp2.result, 'Succeeded', 'Skipped'),
      in(dependencies.DependentJob.result, 'Succeeded', 'Skipped')
    )

And you can see in the results the TagSources job ran even with the two skipped jobs.

Wrapping Up

If there is a simpler way to accomplish what we did above I would love to hear about it. If I remember correctly I found the above in a GitHub issue, but I don’t have the link. I’m not sure how many of you will have Pipelines that will need this, but hopefully, this will save someone some research time.

In this week’s post, we are going to cover manually tagging instead of using the tagging feature built into Azure DevOps. This post will be using a sample Azure DevOps project built over the last few weeks of posts. If you want to see how this project has gotten to this point see the following posts.

Getting Started with Azure DevOps
Pipeline Creation in Azure DevOps
Azure DevOps Publish Artifacts for ASP.NET Core
Azure DevOps Pipelines: Multiple Jobs in YAML
Azure DevOps Pipelines: Reusable YAML
Azure DevOps Pipelines: Use YAML Across Repos
Azure DevOps Pipelines: Conditionals in YAML
Azure DevOps Pipelines: Naming and Tagging

Why?

Tags give you information about the state of your repo when an event happens, a build for our case. This can be super useful especially when you need a place to branch for a hotfix, for example. The automatic tagging covered in my Azure DevOps Pipelines: Naming and Tagging post is the way to go if it works and it has for about 95% of my projects. I have a couple of projects where the automatic tagging would not work for some reason. Since I couldn’t work out why Azure DevOps wouldn’t tag I ended up having to add a job to the Pipeline to perform the tagging.

Permissions

To perform this manual tagging the account used in our Pipeline will need to be giving contribute permission to our repo so that it will be able to push the tag. Use the gear in the lower left to open the Project Setting page.

Select the Repositories option and then click on the specific repo you would like to change the settings for, Playground is the repo we are using in the example. Note that this setting can also be changed using the top-level Git repositories option if you want to change the permission for all of your repos.

On the Security tab under Users looks for the user than has Build Service in the name and select it. When the user is selected it will show their permissions to the right of the users. Find the Contribute option and change its value to Allow.

YAML Changes

As a quick reminder, the YAML for this project currently has 3 jobs. Two web application builds (WebApp1, WebApp2) and the third job is there to show how to use job dependencies (DependentJob). I’m going to skip showing the YAML for these existing jobs, but that if you need the full existing YAML it can be found in posts linked at the top.

Since we already have multiple jobs in our Pipeline we are going to add the tagging code as a new job. This will easily allow us to only tag when all the other jobs have run successfully. The following is the full YAML for the new job.

- job: TagSources
  displayName: 'Tag Sources'
  pool:
    vmImage: 'ubuntu-latest'

  dependsOn:
  - WebApp1
  - WebApp2
  - DependentJob
 
  steps:
  - checkout: self
    persistCredentials: true
    clean: true
    fetchDepth: 1

  - task: PowerShell@2
    inputs:
      targetType: 'inline'
      script: |
        $env:GIT_REDIRECT_STDERR` = '2>&1'
        $tag = "manual_$(Build.BuildNumber)".replace(' ', '_')
        git tag $tag
        Write-Host "Successfully created tag $tag" 

        git push --tags
         Write-Host "Successfully pushed tag $tag"     

      failOnStderr: false

First off you see that this job depends on our existing three jobs to complete successfully before this new job will run.

dependsOn: 
- WebApp1 
- WebApp2 
- DependentJob

Starting in the Steps section you will see a checkout step that is normally handled automatically by the job, but in this case, we need to use the persistCredential option so the job will still be authed and allow us to push to our git repo. The self option is used to signify the current repo/branch. See the official Checkout docs for more information.

- checkout: self
  persistCredentials: true
  clean: true
  fetchDepth: 1

The final bit of the job is the PowerShell task that performs the actual tagging and push. I found most of this on stackoverflow and/or in a GitHub issue, but it has been a while so I don’t have the links handy. The StdErr stuff was to work around some git output that didn’t affect the tagging but was causing the job to be marked as failed. Other than that it is using the standard git commands to tag and push.

- task: PowerShell@2
  inputs:
    targetType: 'inline'
    script: |
      $env:GIT_REDIRECT_STDERR` = '2>&1'
      $tag = "manual_$(Build.BuildNumber)".replace(' ', '_')
      git tag $tag
      Write-Host "Successfully created tag $tag" 

      git push --tags
       Write-Host "Successfully pushed tag $tag"     

    failOnStderr: false

Wrapping Up

I honestly hope none of you have to use this. It was a huge pain to work out. It is also to work around some sort of issue with Azure DevOps built-in tagging support. To be fair I can see where in more complex Pipelines you might need a level of flexibility that the built-in tagging couldn’t provide and this would be your only option.

In this week’s post, we are going to cover changing the naming of Pipeline runs to provide more information as well as tagging our source when a pipeline is run. This post will be using a sample Azure DevOps project built over the last few weeks of posts. If you want to see how this project has progressed check out the following posts.

Getting Started with Azure DevOps
Pipeline Creation in Azure DevOps
Azure DevOps Publish Artifacts for ASP.NET Core
Azure DevOps Pipelines: Multiple Jobs in YAML
Azure DevOps Pipelines: Reusable YAML
Azure DevOps Pipelines: Use YAML Across Repos
Azure DevOps Pipelines: Conditionals in YAML

Naming

By default, Pipeline runs are naming using the current date with a number for how many times the Pipeline has run for the day. For example, the fourth build on March, 17th, 2020 would start with the name 20200317.4 plus the description of the last commit. If you have the need you can change this naming scheme by using a name element in your YAML. As with the rest of the YAML related things you have all the same information available as the rest of the Pipeline to use in building whatever name might be helpful for your situation. For our example, we are going to add the branch name to the front of the run date and count by adding the following name element to the top of our YAML file.

name: $(SourceBranchName)_$(date:yyyyMMdd)$(rev:.r)

resources:      
  repositories: 
  - repository: Shared
    name: Playground/Shared
    type: git 
    ref: master #branch name

trigger: none

The above would result in master_20200317.4 using the same example as above. The following screenshot shows the actual results from this change in the sample Pipeline.

Tagging

Tagging source code when running a Pipeline is a helpful way to know exactly what was included when a Pipeline is run. Here we are going to walk through using Azure DevOps to automatically tag on successful builds. From the Pipeline, you want to tag click the Edit button as you would if you were going to edit the Pipeline’s YAML. Then click the three dots and select Triggers.

Now click on the YAML tab, then Get sources, under Tag sources we are going to select On success so tags will only happen if the build completes successfully. Also, notice the Tag format which allows you to change how the tag is named. When done make sure and Save your changes.

After running a build with the above changes head over to the Repos area of the project. From Files click on History and from there you can see the tag on the last commit that was included in the build, which is displayed here as master_20200325.1.

Wrapping Up

Using clear naming for your builds can give you a lot of information at a glance, but it does take some thought to make sure information your including is helpful. Tagging is also super helpful when viewing history to know what went out with what release, and of course, they can also be used for branching. Come back next week for a look at how to manually tag when Azure DevOps automatic tagging doesn’t work for whatever reason.

In this week’s post, we are going to cover some ways to make tasks and jobs run conditionally. This will include options such as Pipeline variables to jobs that are dependent on other jobs. This post will be using a sample Azure DevOps project built over the last few weeks of posts. If you want to see the build-up check out the following posts.

Getting Started with Azure DevOps
Pipeline Creation in Azure DevOps
Azure DevOps Publish Artifacts for ASP.NET Core
Azure DevOps Pipelines: Multiple Jobs in YAML
Azure DevOps Pipelines: Reusable YAML
Azure DevOps Pipelines: Use YAML Across Repos

Sample YAML

The following YAML is based on the YAML from the previous posts, see links above, expanded with examples of using some ways of conditionally running some task or job. This is the full file for reference and the rest of the post will call out specific parts of the file as needed.

resources:      
  repositories: 
  - repository: Shared
    name: Playground/Shared
    type: git 
    ref: master #branch name

trigger: none

variables:
  buildConfiguration: 'Release'

jobs:
- job: WebApp1
  displayName: 'Build WebApp1'
  pool:
    vmImage: 'ubuntu-latest'

  steps:
  - template: buildCoreWebProject.yml@Shared
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp1.csproj
      artifactName: WebApp1

- job: WebApp2
  displayName: 'Build WebApp2'
  condition: and(succeeded(), eq(variables['BuildWebApp2'], 'true'))
  pool:
    vmImage: 'ubuntu-latest'

  steps:
  - template: build.yml
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp2.csproj
      artifactName: WebApp2
      
- job: DependentJob
  displayName: 'Build Dependent Job'
  pool:
    vmImage: 'ubuntu-latest'

  dependsOn:
  - WebApp1
  - WebApp2

  steps:
  - template: buildCoreWebProject.yml@Shared
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp1.csproj
      artifactName: WebApp1Again

Job Dependencies

The more complex pipelines get the more likely the pipeline will end up with a job that can’t run until other jobs have completed. The YAML above defines three different jobs, WebApp1, WebApp2, and DependentJob. I’m sure you have guessed by now that the third job is the one that has a dependency. To make a job dependent on other jobs we use the dependsOn element and list the jobs that must complete before the job in question can run. The following is the YAML for the sample DependentJob with the dependsOn section highlighted.

- job: DependentJob
  displayName: 'Build Dependent Job'
  pool:
    vmImage: 'ubuntu-latest'

  dependsOn:
  - WebApp1
  - WebApp2

  steps:
  - template: buildCoreWebProject.yml@Shared
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp1.csproj
      artifactName: WebApp1Again

With the above setup, DependentJob will only run if both the WebApp1 and WebApp2 jobs complete successfully.

Conditions

Conditions are a way to control if a Job or Task is run. The following example is at the job level, but the same concept works at the task level. Notice the highlighted condition.

- job: WebApp2
  displayName: 'Build WebApp2'
  condition: and(succeeded(), eq(variables['BuildWebApp2'], 'true'))
  pool:
    vmImage: 'ubuntu-latest'

  steps:
  - template: build.yml
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp2.csproj
      artifactName: WebApp2

The above condition will cause the WebApp2 job to be skipped if the BuildWebApp2 variable isn’t true. For more details on how to use conditions see the Conditions docs.

Creating a Pipeline Variable

The rest of the post is going to walk through creating a Pipeline variable and then running some sample builds to show how depends on and the conditions defined in the YAML above affect the Pipeline results.

We are starting from an existing pipeline that is already being edited. To add (or edit) variables click the Variables button in the top right of the screen.

The Variables pop out will show. If we had existing variables they show here. Click the New variable button to add a new variable.

We are adding a variable that will control the build of WebApp2 called BuildWebApp2 that defaults to the value of true. Also, make sure and check the Let user override this value when running this pipeline checkbox to allow us to edit this variable when doing a run of the pipeline. Then click the OK button.

Back on the Variables dialog click the Save button.

Edit Variables When Starting a Pipeline

Now that our Pipeline has a variable when running the Pipeline under Advanced options you will see the Variables section showing that our Pipeline has 1 variable defined. Click Variables to view/edit the variables that will be used for this run of the Pipeline.

From the Variables section, you will see a list of the defined variables as well as an option to add new variables that will exist only for this run of the Pipeline. Click on the BuildWebApp2 variable to edit the value that will be used for this run of the Pipeline.

From the Update variable dialog, you can change the value of the variable. When done click the Update button.

Pipeline Results from Sample YAML

The following is what our sample Pipeline looks like when queued with the BuildWebApp2 variable set to false. As you can see the job will be skipped.

Next is the completed results of the Pipeline run. You can see that the Build Dependent Job was skipped as well since both Build WebApp1 and Build WebApp2 must complete successfully before it will run.

Changing the BuildWebApp2 variable back to true and running the Pipeline again results in all the jobs running successfully.

Wrapping Up

Hopefully, this has helped introduce you to some of the ways you can control your Pipelines. As with everything else Azure DevOps related things are changing a lot and new options are popping up all the time. For example, while writing this post the team just announced Runtime Parameters which look like a much better option than variables for values that frequently vary between Pipeline runs.

In last week’s post, we refactored some YAML that was reusable into a new file. This post is going to cover moving that same reusable YAML to a new repo and then using it in our existing sample repo. This post is going to build on the Azure DevOps project created in previous posts if you are just joining this series check out the previous posts to find out how the project has progressed.

Getting Started with Azure DevOps
Pipeline Creation in Azure DevOps
Azure DevOps Publish Artifacts for ASP.NET Core
Azure DevOps Pipelines: Multiple Jobs in YAML
Azure DevOps Pipelines: Reusable YAML

Create a New Repository

First, we need to create a new repository that will be used to share the YAML in question. Using the Repos section of Azure DevOps as a starting point you click the dropdown with the currently selected repo name, Playground in this example, and then click New repository.

You will be presented with a dialog where you will need to enter the Repository name and any other of the options you want to configure. In this example, we are naming the repo Shared and adding a Git ignore file for Visual Studio. When done click the Create button.

The following steps should be taken on the new Shared repo. At this point, you could clone the repo and do the rest of the steps locally and then push the changes to the repo or you can use the web interface to make all the change which is the route this post is going to show. Either way, you go it shouldn’t be too hard to adapt the steps. For the web interface to add a new file in the root of the repo click the three-dot menu to the right of the repo name and then select New and then File.

The next prompt will ask for a file name, I’m using buildCoreWebProject.yml. Click Create to continue.

You will land in the file editor. Copy and paste the code out of build.yml that we were using from the previous post into the new file. The following is the full YAML from build.yml for reference.

parameters:
- name: buildConfiguration
  type: string
  default: 'Release'
- name: project
  type: string
  default: ''
- name: artifactName
  type: string
  default: ''

steps:
  - task: UseDotNet@2
    displayName: 'Use .NET 3.1.x'
    inputs:
      packageType: 'sdk'
      version: '3.1.x'

  - task: DotNetCoreCLI@2
    displayName: 'Build'
    inputs:
      command: 'build'
      projects: '**/${{ parameters.project }}'
      arguments: '--configuration ${{ parameters.buildConfiguration }}' 
  
  - task: DotNetCoreCLI@2
    displayName: 'Publish Application'
    inputs:
      command: 'publish'
      publishWebProjects: false
      projects: '**/${{ parameters.project }}'
      arguments: '--configuration ${{ parameters.buildConfiguration }} --output $(Build.ArtifactStagingDirectory)'

  - task: PublishPipelineArtifact@1
    displayName: 'Publish Artifacts'
    inputs:
      targetPath: '$(Build.ArtifactStagingDirectory)'
      artifact: ${{ parameters.artifactName }}
      publishLocation: 'pipeline'

Once the code is copied in click the Commit button to save the changes to the master branch.

Switching back to our original repo, Playground in this example, we need to add the Shared repo as a resource for in our azure-pipelines.yml file. The following is the resource deliration for using another Azure DevOps repo. The official docs for check out multiple repositories also show examples with GitHub and Bitbucket. I also found this stackoverflow question helpful.

resources: 
  repositories: 
  - repository: Shared 
    name: Playground/Shared 
    type: git 
    ref: master #branch name

Shared on the repository line is the name we will be using when referencing a file out of the Shared repo. Name is the Azure DevOps project and repo name. Type is the repo type which is Git in our case. Finally, ref is the branch name from the Shared repo that we want to use. Now that we have access to the files from the Shared repo we can use its buildCoreWebProject.yml instead of the local build.yml as a template.

Before:
  - template: build.yml
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp1.csproj
      artifactName: WebApp1

After:
  - template: buildCoreWebProject.yml@Shared
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp1.csproj
      artifactName: WebApp1

Notice that the only change is on the template line which changed from build.yml to buildCoreWebProject.yml@Shared. The @Shared on the end of the filename is what tells the pipeline the file’s source is the Shared repo. The following is the full azure-pipeline.yml for reference. The frist job is using the template form the Shared repo and the second one is using a local template.

resources:      
  repositories: 
  - repository: Shared
    name: Playground/Shared
    type: git 
    ref: master #branch name

trigger: none

variables:
  buildConfiguration: 'Release'

jobs:
- job: WebApp1
  displayName: 'Build WebApp1'
  pool:
    vmImage: 'ubuntu-latest'

  steps:
  - template: buildCoreWebProject.yml@Shared
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp1.csproj
      artifactName: WebApp1

- job: WebApp2
  displayName: 'Build WebApp2'
  pool:
    vmImage: 'ubuntu-latest'

  steps:
  - template: build.yml
    parameters:
      buildConFiguration: $(buildConfiguration)
      project: WebApp2.csproj
      artifactName: WebApp2

Wrapping Up

Being able to utilize YAML from different repos can help cut down on duplicated YAML and help keep your pipelines across repos cleaner. As with anything else, this is a useful tool when applied appropriately.