Re-run Only Failed Tests

How to setup failed tests re-run on GitHub Actions

When a workflow fails in GitHub Actions you have the option to re-run the failed jobs. However, an additional setup is required for properly configure Playwright for rerunning only the failed tests.

See Re-run Only Failed Tests guide for more details on re-runs.

For GitHub Actions, we provide the Last Failed GitHub Action to simplify the re-runs.

We recommend you install @currents/cmd as a dev dependency in your package.json, and using npm ci or your package manager's frozen lockfile install method in your GitHub Actions. If you do not the Last Failed GitHub Action will instead install and use a global package, which will not respect your package's version lock files, and instead always pull in the latest of @currents/cmdand it's dependencies.

Playwright Sharding

If you're using Reporting in CI for running your tests in parallel, you can use the Last Failed GitHub Action to include the data from the last run.

Step-by-step guide:

Install the @currents/cmd package

npm i -D @currents/cmd

Add the currents-dev/playwright-last-failed step

Add a step to your workflow before you run your tests

- name: Playwright Last Failed action
  id: last-failed-action
  uses: currents-dev/playwright-last-failed@v1
  with:
    # if you're using a custom CI build id, set "previous-ci-build-id" accordingly 
    # previous-ci-build-id: default is ${{ github.repository }}-${{ github.run_id }}-<%= ${{ github.run_attempt }} - 1 %>
    pw-output-dir: basic/test-results
    matrix-index: ${{ matrix.shard }}
    matrix-total: ${{ strategy.job-total }}

See the action configuration for details.

A full example

name: failed-only-reruns

on:
  push:

jobs:
  test-reporter:
    strategy:
      fail-fast: false
      matrix:
        shard: [1, 2, 3]
    timeout-minutes: 60
    runs-on: ubuntu-latest
    container: mcr.microsoft.com/playwright:latest
    env:
      CURRENTS_PROJECT_ID: bnsqNa
      CURRENTS_RECORD_KEY: ${{ secrets.CURRENTS_RECORD_KEY }}
      CURRENTS_CI_BUILD_ID: ${{ github.repository }}-${{ github.run_id }}-${{ github.run_attempt }}
    steps:
      - uses: actions/checkout@v4
        with:
          ref: ${{ github.ref }}
      - run: |
          echo "$GITHUB_WORKSPACE"
          git config --global --add safe.directory "$GITHUB_WORKSPACE"
      - uses: actions/setup-node@v4
        with:
          node-version: "20.x"
      - name: Install dependencies
        run: |
          npm ci
          npx playwright install chrome
      - name: Playwright Last Failed action
        id: last-failed-action
        uses: currents-dev/playwright-last-failed@v1
        with:
          pw-output-dir: basic/test-results
          matrix-index: ${{ matrix.shard }}
          matrix-total: ${{ strategy.job-total }}
      - name: Playwright Tests
        working-directory: ./basic
        run: |
          COMMAND="npx playwright test --config playwright.config.reporter.ts ${{ steps.last-failed-action.outputs.extra-pw-flags }}"
          echo "Running command: $COMMAND"
          $COMMAND

Note the use of Custom CI Build ID for Reruns.

Full examples:

• rerun-shards-pwc.yml - rerun only the tests that failed in the previous run, using pwc helper command that is included in @currents/playwright package.

• rerun-shards-reporter.yml - rerun only the tests that failed in the previous run, using reporter explicitly configured in playwright.config.ts

Currents Orchestration

If you're using Currents Orchestration for running your Playwright tests you can also fetch the results of from Runs.

Currents Orchestration dynamically assigns tests to all the available CI runners, that's why you should select Re-run all jobs when using Currents Orchestration. Read more at Re-run Only Failed Tests guide.

Step-by-step guide:

Install the @currents/cmd package

npm i -D @currents/cmd

Set CURRENTS_API_KEY CI environment variable

Obtain an API key (see Authentication) from Currents Dashboard (in addition to Record Key) and set GitHub Actions Secret

env:
  CURRENTS_RECORD_KEY: ${{ secrets.CURRENTS_RECORD_KEY }}
  CURRENTS_API_KEY: ${{ secrets.CURRENTS_API_KEY }}

Add the currents-dev/playwright-last-failed step

Add a step that fetches the last-run information prior to running tests

- name: Playwright Last Failed action
  id: last-failed-action
  uses: currents-dev/playwright-last-failed@v1
  with:
    or8n: true
    # debug: true
    # previous-ci-build-id: default is ${{ github.repository }}-${{ github.run_id }}-%<= ${{ github.run_attempt }} - 1 %>
    pw-output-dir: basic/test-results

See the action configuration for details on the inputs.

A full example

name: failed-only-or8n

on:
  push:

jobs:
  test-or8n:
    strategy:
      fail-fast: false
      matrix:
        shard: [1, 2, 3]
    timeout-minutes: 60
    runs-on: ubuntu-latest
    container: mcr.microsoft.com/playwright:latest
    env:
      CURRENTS_PROJECT_ID: bnsqNa
      CURRENTS_RECORD_KEY: ${{ secrets.CURRENTS_RECORD_KEY }}
      CURRENTS_CI_BUILD_ID: ${{ github.repository }}-${{ github.run_id }}-${{ github.run_attempt }}
      CURRENTS_API_KEY: ${{ secrets.CURRENTS_API_KEY }}
    steps:
      - uses: actions/checkout@v4
        with:
          ref: ${{ github.ref }}
      - run: |
          echo "$GITHUB_WORKSPACE"
          git config --global --add safe.directory "$GITHUB_WORKSPACE"
      - uses: actions/setup-node@v4
        with:
          node-version: "20.x"
      - name: Install dependencies
        run: |
          npm ci
          npx playwright install chrome
      - name: Playwright Last Failed action
        id: last-failed-action
        uses: currents-dev/playwright-last-failed@v1
        with:
          or8n: true
          # previous-ci-build-id: default is ${{ github.repository }}-${{ github.run_id }}-<%= ${{ github.run_attempt }} - 1 %>
          pw-output-dir: basic/test-results
      - name: Playwright Tests
        working-directory: ./basic
        run: |
          COMMAND="npx pwc-p ${{ steps.last-failed-action.outputs.extra-pw-flags }}"
          echo "Running command: $COMMAND"
          $COMMAND

Example workflow:

• reruns-or8n.yml - rerun only the tests that failed in the previous orchestrated run.

Note the use of Custom CI Build ID for Reruns.

Custom CI Build ID for Reruns

The last-failed-action gets the previous run information using the default CI build ID pattern:

${{ github.repository }}-${{ github.run_id }}-${{ github.run_attempt }}

If you are using a different CI Build ID, specify the previous-ci-build-id configuration property.

Using custom CI build ID for reruns

For example:

# an example for custom value like:
# currents-${{ github.run_id }}-${{ github.run_attempt }}
with:
    # if you're using a custom CI build id, set "previous-ci-build-id" accordingly 
    previous-ci-build-id: currents-${{ github.run_id }}-<%= ${{ github.run_attempt }} - 1 %>
    pw-output-dir: basic/test-results
    matrix-index: ${{ matrix.shard }}
    matrix-total: ${{ strategy.job-total }}