Specify conditions

TFS 2017

This article applies to TFS 2017.3 and later versions.

You can specify the conditions under which each stage, job, or step runs. By default, a job or stage runs if it does not depend on any other job or stage, or if all of the jobs or stages that it depends on have completed and succeeded. By default, a step runs if nothing in its job has failed yet and the step immediately preceding it has finished. You can customize this behavior by forcing a stage, job, or step to run even if a previous dependency fails or by specifying a custom condition.

Note

In Microsoft Team Foundation Server (TFS) 2018 and previous versions, build and release pipelines are called definitions, runs are called builds, service connections are called service endpoints, stages are called environments, and jobs are called phases.

YAML is not supported on TFS 2017.

Enable a custom condition

If the built-in conditions don't meet your needs, then you can specify custom conditions.

In TFS 2017.3, custom task conditions are available in the user interface only for Build pipelines. You can use the Release REST APIs to establish custom conditions for Release pipelines.

Conditions are written as expressions in YAML pipelines. The agent evaluates the expression beginning with the innermost function and works its way out. The final result is a boolean value that determines if the task, job, or stage should run or not. See the expressions topic for a full guide to the syntax.

Do any of your conditions make it possible for the task to run even after the build is canceled by a user? If so, then specify a reasonable value for cancel timeout so that these kinds of tasks have enough time to complete after the user cancels a run.

Examples

Run for the main branch, if succeeding

and(succeeded(), eq(variables['Build.SourceBranch'], 'refs/heads/main'))

Run if the branch is not main, if succeeding

and(succeeded(), ne(variables['Build.SourceBranch'], 'refs/heads/main'))

Run for user topic branches, if succeeding

and(succeeded(), startsWith(variables['Build.SourceBranch'], 'refs/heads/users/'))

Run for continuous integration (CI) builds if succeeding

and(succeeded(), in(variables['Build.Reason'], 'IndividualCI', 'BatchedCI'))

Run if the build is run by a branch policy for a pull request, if failing

and(failed(), eq(variables['Build.Reason'], 'PullRequest'))

Run if the build is scheduled, even if failing, even if canceled

and(always(), eq(variables['Build.Reason'], 'Schedule'))

Release.Artifacts.{artifact-alias}.SourceBranch is equivalent to Build.SourceBranch.

Run if a variable is set to true

condition: eq(variables['System.debug'], 'true')

Run if a variable is null (empty string)

Since all variables are treated as strings in Azure Pipelines, an empty string is equivalent to null in this pipeline.

variables:
- name: testEmpty
  value: ''

jobs:
  - job: A
    steps:
    - script: echo testEmpty is blank
    condition: eq(variables.testEmpty, '')

Use a template parameter as part of a condition

When you declare a parameter in the same pipeline that you have a condition, parameter expansion happens before conditions are considered. In this case, you can embed parameters inside conditions. The script in this YAML file will run because parameters.doThing is true.

parameters:
- name: doThing
  default: true
  type: boolean

steps:
- script: echo I did a thing
  condition: and(succeeded(), eq('${{ parameters.doThing }}', 'true'))

However, when you pass a parameter to a template, the parameter will not have a value when the condition gets evaluated. As a result, if you set the parameter value in both the template and the pipeline YAML files, the value from the template will get used in your condition.

# parameters.yml
parameters:
- name: doThing
  default: true # value passed to the condition
  type: boolean

jobs:
  - job: B
    steps:
    - script: echo I did a thing
    condition: and(succeeded(), eq('${{ parameters.doThing }}', 'true'))
# azure-pipeline.yml
parameters:
- name: doThing
  default: true 
  type: boolean

trigger:
- none

extends:
  template: parameters.yml

The output of this pipeline is I did a thing because the parameter doThing is true.

Use the output variable from a job in a condition in a subsequent job

You can make a variable available to future jobs and specify it in a condition. Variables available to future jobs must be marked as multi-job output variables using isOutput=true.

jobs:
- job: Foo
  steps:
  - bash: |
      echo "This is job Foo."
      echo "##vso[task.setvariable variable=doThing;isOutput=true]Yes" #set variable doThing to Yes
    name: DetermineResult
- job: Bar
  dependsOn: Foo
  condition: eq(dependencies.Foo.outputs['DetermineResult.doThing'], 'Yes') #map doThing and check the value
  steps:
  - script: echo "Job Foo ran and doThing is Yes."

FAQ

I've got a conditional step that runs even when a job is canceled. Does my conditional step affect a job that I canceled in the queue?

No. If you cancel a job while it's in the queue, then the entire job is canceled, including conditional steps.

I've got a conditional step that should run even when the deployment is canceled. How do I specify this?

If you defined the pipelines using a YAML file, then this is supported. This scenario is not yet supported for release pipelines.

How can I trigger a job if a previous job succeeded with issues?

You can use the result of the previous job. For example, in this YAML file, the condition eq(dependencies.A.result,'SucceededWithIssues') allows the job to run because Job A succeeded with issues.

jobs:
- job: A
  displayName: Job A
  continueOnError: true # next job starts even if this one fails
  steps:
  - script: echo Job A ran
  - script: exit 1

- job: B
  dependsOn: A
  condition: eq(dependencies.A.result,'SucceededWithIssues') # targets the result of the previous job 
  displayName: Job B
  steps:
  - script: echo Job B ran

I've got a conditional step that runs even when a job is canceled. How do I manage to cancel all jobs at once?

You'll experience this issue if the condition that's configured in the stage doesn't include a job status check function. To resolve the issue, add a job status check function to the condition. If you cancel a job while it's in the queue, the entire job is canceled, including all the other stages, with this function configured. For more information, see Job status functions.

stages:
- stage: Stage1
  displayName: Stage 1
  dependsOn: []
  condition: and(contains(variables['build.sourceBranch'], 'refs/heads/main'), succeeded())
  jobs:
  - job: ShowVariables
    displayName: Show variables
    steps:
    - task: CmdLine@2
      displayName: Show variables
      inputs:
        script: 'printenv'

- stage: Stage2
  displayName: stage 2
  dependsOn: Stage1
  condition: contains(variables['build.sourceBranch'], 'refs/heads/main')
  jobs:
  - job: ShowVariables
    displayName: Show variables 2
    steps:
    - task: CmdLine@2
      displayName: Show variables 2
      inputs:
        script: 'printenv'
          
- stage: Stage3
  displayName: stage 3
  dependsOn: Stage2
  condition: and(contains(variables['build.sourceBranch'], 'refs/heads/main'), succeeded())
  jobs:
  - job: ShowVariables
    displayName: Show variables 3
    steps:
    - task: CmdLine@2
      displayName: Show variables 3
      inputs:
        script: 'printenv'