load.yml

name: Load test

# Daily CSV -> Vortex -> CSV round-trip at ~10 GB (~100 M rows). The data is a
# deterministic seeded generator that doubles as the oracle, so the exported
# rows are diffed against the regenerated rows (no second copy stored).
#
# Scheduled workflows run only from the default branch, may be delayed under
# load, and are auto-disabled after 60 days of repo inactivity.
on:
  schedule:
    - cron: '0 3 * * *'   # 03:00 UTC daily
  workflow_dispatch:
    inputs:
      rows:
        description: 'Number of rows to generate'
        required: false
        default: '100000000'

# Never run two load jobs at once; a late-starting cron should yield to a newer one.
concurrency:
  group: load-test
  cancel-in-progress: false

permissions:
  contents: read

jobs:
  load:
    runs-on: ubuntu-latest
    timeout-minutes: 120

    steps:
      - uses: actions/checkout@v7

      - name: Set up Azul Zulu JDK 25
        uses: actions/setup-java@v5
        with:
          distribution: zulu
          java-version: '25'

      - name: Cache Maven repository
        uses: actions/cache@v5
        with:
          path: ~/.m2/repository
          key: ${{ runner.os }}-maven-${{ hashFiles('**/pom.xml') }}
          restore-keys: |
            ${{ runner.os }}-maven-

      # The root volume has only ~21 GB free; /mnt is the large (~70 GB) scratch
      # disk on GitHub-hosted Ubuntu runners. The 10 GB CSV + Vortex file + 10 GB
      # re-exported CSV (~25 GB peak) fit there but not on the root volume.
      - name: Prepare scratch dir on the large disk
        run: |
          sudo mkdir -p /mnt/vortex-load
          sudo chown "$USER" /mnt/vortex-load
          df -h / /mnt

      - name: Run round-trip load test
        run: >
          ./mvnw verify -pl integration -am
          -Dit.test=LargeCsvRoundTripLoadIntegrationTest
          -Dvortex.load.rows=${{ github.event.inputs.rows || '100000000' }}
          -Dvortex.load.dir=/mnt/vortex-load