greptimedb/.github/workflows/nightly-jsonbench.yaml

name: Nightly JSONBench

on:
  schedule:
    # Trigger at 00:00(Asia/Shanghai) on every weekday.
    - cron: "0 16 * * 0-4"
  workflow_dispatch:

concurrency:
  group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
  cancel-in-progress: true

jobs:
  allocate-runner:
    name: Allocate runner
    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
    runs-on: ubuntu-latest
    outputs:
      linux-arm64-runner: ${{ steps.start-linux-arm64-runner.outputs.label }}

      # The following EC2 resource id will be used for resource releasing.
      linux-arm64-ec2-runner-label: ${{ steps.start-linux-arm64-runner.outputs.label }}
      linux-arm64-ec2-runner-instance-id: ${{ steps.start-linux-arm64-runner.outputs.ec2-instance-id }}
    steps:
      - name: Checkout
        uses: actions/checkout@v4
        with:
          fetch-depth: 0
          persist-credentials: false

      - name: Allocate Linux ARM64 runner
        uses: ./.github/actions/start-runner
        id: start-linux-arm64-runner
        with:
          runner: ${{ vars.DEFAULT_ARM64_RUNNER }}
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ${{ vars.EC2_RUNNER_REGION }}
          github-token: ${{ secrets.GH_PERSONAL_ACCESS_TOKEN }}
          image-id: ${{ vars.EC2_RUNNER_LINUX_ARM64_IMAGE_ID }}
          security-group-id: ${{ vars.EC2_RUNNER_SECURITY_GROUP_ID }}
          subnet-id: ${{ vars.EC2_RUNNER_SUBNET_ID }}

  jsonbench:
    name: Run JSONBench
    if: ${{ github.repository == 'GreptimeTeam/greptimedb' }}
    needs: [ allocate-runner ]
    runs-on: ${{ needs.allocate-runner.outputs.linux-arm64-runner }}
    timeout-minutes: 120
    env:
      JSONBENCH_DATA_DIR: /home/runner/data/bluesky
      JSONBENCH_OUTPUT_PREFIX: _ubuntu-latest
    steps:
      - name: Checkout
        uses: actions/checkout@v4
        with:
          fetch-depth: 0
          persist-credentials: false

      - uses: arduino/setup-protoc@v3
        with:
          repo-token: ${{ secrets.GITHUB_TOKEN }}

      - uses: actions-rust-lang/setup-rust-toolchain@v1

      - name: Rust Cache
        uses: Swatinem/rust-cache@v2
        with:
          shared-key: "nightly-jsonbench"
          cache-all-crates: "true"
          save-if: ${{ github.ref == 'refs/heads/main' }}

      - name: Build GreptimeDB
        run: cargo build --profile nightly --bin greptime

      - name: Reclaim disk space
        shell: bash
        run: |
          set -euo pipefail

          mkdir -p "${RUNNER_TEMP}/greptimedb-bin"
          cp ./target/nightly/greptime "${RUNNER_TEMP}/greptimedb-bin/greptime"
          chmod +x "${RUNNER_TEMP}/greptimedb-bin/greptime"

          rm -rf ./target

      - name: Run JSONBench
        shell: bash
        run: |
          set -euo pipefail

          cd "${RUNNER_TEMP}"
          cp "${RUNNER_TEMP}/greptimedb-bin/greptime" ./greptime
          chmod +x ./greptime

          export GREPTIMEDB_STANDALONE__WAL__DIR=greptimedb_data/wal
          export GREPTIMEDB_STANDALONE__STORAGE__DATA_HOME=greptimedb_data
          export GREPTIMEDB_STANDALONE__LOGGING__DIR=greptimedb_data/logs
          export GREPTIMEDB_STANDALONE__LOGGING__APPEND_STDOUT=false
          export GREPTIMEDB_STANDALONE__HTTP__BODY_LIMIT=1GB
          export GREPTIMEDB_STANDALONE__HTTP__TIMEOUT=500s

          ./greptime standalone start > greptimedb.log 2>&1 &
          greptime_pid=$!
          trap 'kill "${greptime_pid}" 2>/dev/null || true' EXIT

          until curl -s --fail -o /dev/null http://localhost:4000/health; do
            if ! kill -0 "${greptime_pid}" 2>/dev/null; then
              cat greptimedb.log
              exit 1
            fi
            sleep 1
          done

          git clone --branch greptimedb-new-json --depth 1 https://github.com/GreptimeTeam/JSONBench.git JSONBench
          cp ./greptime JSONBench/greptimedb/greptime

          cd JSONBench/greptimedb
          ./main.sh 3 "${JSONBENCH_DATA_DIR}" success.log error.log "${JSONBENCH_OUTPUT_PREFIX}" false

      - name: Upload JSONBench results
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: jsonbench-results
          path: |
            ${{ runner.temp }}/greptimedb.log
            ${{ runner.temp }}/JSONBench/greptimedb/*.log
            ${{ runner.temp }}/JSONBench/greptimedb/*.total_size
            ${{ runner.temp }}/JSONBench/greptimedb/*.data_size
            ${{ runner.temp }}/JSONBench/greptimedb/*.index_size
            ${{ runner.temp }}/JSONBench/greptimedb/*.count
            ${{ runner.temp }}/JSONBench/greptimedb/*.results_runtime
            ${{ runner.temp }}/JSONBench/greptimedb/*.query_results
          if-no-files-found: ignore
          retention-days: 7

  stop-linux-arm64-runner:
    name: Stop Linux ARM64 runner
    # It's always run as the last job in the workflow to make sure that the runner is released.
    if: ${{ always() }}
    runs-on: ubuntu-latest
    needs: [
      allocate-runner,
      jsonbench,
    ]
    steps:
      - name: Checkout
        uses: actions/checkout@v4
        with:
          fetch-depth: 0
          persist-credentials: false

      - name: Stop Linux ARM64 runner
        uses: ./.github/actions/stop-runner
        with:
          label: ${{ needs.allocate-runner.outputs.linux-arm64-ec2-runner-label }}
          ec2-instance-id: ${{ needs.allocate-runner.outputs.linux-arm64-ec2-runner-instance-id }}
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ${{ vars.EC2_RUNNER_REGION }}
          github-token: ${{ secrets.GH_PERSONAL_ACCESS_TOKEN }}