How to Run Jobs in Parallel With a Matrix in Azure Pipelines

A matrix strategy clones one job into several parallel legs, each with its own variable values.

Add strategy.matrix to a job with one entry per leg. Each entry is a name plus a map of variables, and Azure DevOps runs the legs in parallel up to maxParallel.

Steps

• Add strategy.matrix to the job.
• Give each leg a name and a map of variables.
• Reference a matrix variable as $(name) inside the steps.
• Cap concurrency with maxParallel if needed.

Pipeline

jobs:
  - job: test
    strategy:
      maxParallel: 3
      matrix:
        linux:
          imageName: ubuntu-latest
        windows:
          imageName: windows-latest
        mac:
          imageName: macos-latest
    pool:
      vmImage: $(imageName)
    steps:
      - script: npm ci && npm test

Gotchas

• Each leg counts as a parallel job and consumes a parallelism slot.
• Matrix variables are available as $(name) in steps and in the pool selection.
• For a self-healing managed pool, parallel matrix legs retry transient agent dropouts automatically.

Related guides