|
| 1 | +name: data-generation run |
| 2 | +run-name: data-generation run - sf${{ github.event.inputs.scale_factor }} - ${{ github.event.inputs.num_steps }} steps |
| 3 | + |
| 4 | +on: |
| 5 | + workflow_dispatch: |
| 6 | + inputs: |
| 7 | + scale_factor: |
| 8 | + description: 'TPC-H scale factor' |
| 9 | + required: true |
| 10 | + default: '1.0' |
| 11 | + type: string |
| 12 | + num_steps: |
| 13 | + description: 'Number of data generation steps' |
| 14 | + required: false |
| 15 | + default: '100' |
| 16 | + type: string |
| 17 | + bucket: |
| 18 | + description: 'S3 bucket name' |
| 19 | + required: true |
| 20 | + type: string |
| 21 | + prefix: |
| 22 | + description: 'S3 key prefix for generated files' |
| 23 | + required: false |
| 24 | + default: 'data-gen/tpch' |
| 25 | + type: string |
| 26 | + max_concurrency: |
| 27 | + description: 'Maximum number of concurrent S3 writes' |
| 28 | + required: true |
| 29 | + default: '8' |
| 30 | + type: string |
| 31 | + region: |
| 32 | + description: 'AWS region' |
| 33 | + required: true |
| 34 | + default: 'us-east-1' |
| 35 | + type: string |
| 36 | + skip_initial: |
| 37 | + description: 'Skip data ingested by running --initial command' |
| 38 | + required: false |
| 39 | + default: false |
| 40 | + type: boolean |
| 41 | + |
| 42 | +jobs: |
| 43 | + run-data-generation: |
| 44 | + name: Run data generation |
| 45 | + runs-on: ${{ github.event.inputs.runner_type }} |
| 46 | + timeout-minutes: 600 |
| 47 | + steps: |
| 48 | + - uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6 |
| 49 | + |
| 50 | + - name: Install Rust |
| 51 | + uses: dtolnay/rust-toolchain@stable |
| 52 | + |
| 53 | + - name: Build data-generation |
| 54 | + run: cargo build --release -p data-generation |
| 55 | + |
| 56 | + - name: Run data generation |
| 57 | + run: | |
| 58 | + ARGS="--scale-factor ${{ github.event.inputs.scale_factor }}" |
| 59 | + ARGS="${ARGS} --num-steps ${{ github.event.inputs.num_steps }}" |
| 60 | + ARGS="${ARGS} --bucket ${{ github.event.inputs.bucket }}" |
| 61 | + ARGS="${ARGS} --prefix ${{ github.event.inputs.prefix }}" |
| 62 | + ARGS="${ARGS} --max-concurrency ${{ github.event.inputs.max_concurrency }}" |
| 63 | + ARGS="${ARGS} --region ${{ github.event.inputs.region }}" |
| 64 | +
|
| 65 | + if [ "${{ github.event.inputs.skip_initial }}" = "true" ]; then |
| 66 | + ARGS="${ARGS} --skip-initial" |
| 67 | + fi |
| 68 | +
|
| 69 | + echo "Running: data-generation run ${ARGS}" |
| 70 | + ./target/release/data-generation run ${ARGS} |
| 71 | + env: |
| 72 | + AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }} |
| 73 | + AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }} |
| 74 | + RUST_LOG: info |
0 commit comments