seaweedfs/weed/storage/blockvol/testrunner/scenarios/internal/benchmark-full.yaml

name: benchmark-full
timeout: 8m

env:
  master_url: "http://192.168.1.184:9433"
  volume_name: bench-full
  vol_size: "2147483648"

topology:
  nodes:
    m01:
      host: 192.168.1.181
      user: testdev
      key: "/opt/work/testdev_key"
    m02:
      host: 192.168.1.184
      user: testdev
      key: "/opt/work/testdev_key"

phases:
  # Phase 1: Clean environment
  - name: cleanup
    actions:
      - action: pre_run_cleanup
        node: m01
        kill_patterns: "weed,postgres"
        unmount: "/mnt/sw-bench"
        nvme_disconnect: "true"
      - action: pre_run_cleanup
        node: m02
        kill_patterns: "weed"

  # Phase 2: Start cluster (M02 master + VS, m01 VS for RF=2)
  - name: cluster
    actions:
      - action: exec
        node: m02
        cmd: "rm -rf /tmp/bench-master /tmp/bench-vs1 && mkdir -p /tmp/bench-master /tmp/bench-vs1/blocks"
        root: "true"
      - action: exec
        node: m01
        cmd: "rm -rf /tmp/bench-vs2 && mkdir -p /tmp/bench-vs2/blocks"
        root: "true"

      - action: start_weed_master
        node: m02
        port: "9433"
        dir: /tmp/bench-master
        save_as: master_pid

      - action: sleep
        duration: 3s

      - action: start_weed_volume
        node: m02
        port: "18480"
        master: "localhost:9433"
        dir: /tmp/bench-vs1
        extra_args: "-block.dir=/tmp/bench-vs1/blocks -block.listen=:3295 -block.nvme.enable=true -block.nvme.listen=10.0.0.3:4430 -ip=192.168.1.184"
        save_as: vs1_pid

      - action: start_weed_volume
        node: m01
        port: "18481"
        master: "192.168.1.184:9433"
        dir: /tmp/bench-vs2
        extra_args: "-block.dir=/tmp/bench-vs2/blocks -block.listen=:3296 -block.nvme.enable=true -block.nvme.listen=10.0.0.1:4431 -ip=192.168.1.181"
        save_as: vs2_pid

      - action: sleep
        duration: 5s

      - action: wait_cluster_ready
        node: m02
        master_url: "{{ master_url }}"

      - action: wait_block_servers
        count: "2"

  # Phase 3: Create RF=2 sync_all volume
  - name: create
    actions:
      - action: create_block_volume
        name: "{{ volume_name }}"
        size_bytes: "{{ vol_size }}"
        replica_factor: "2"
        durability_mode: sync_all
      - action: sleep
        duration: 10s

  # Phase 4: Wait for volume to be healthy (shipper InSync)
  - name: wait-healthy
    actions:
      - action: wait_volume_healthy
        name: "{{ volume_name }}"
        timeout: "60s"

  # Phase 5: Validate replication config
  - name: validate-replication
    actions:
      - action: validate_replication
        volume_name: "{{ volume_name }}"
        expected_rf: "2"
        expected_durability: sync_all

  # Phase 5: Report header
  - name: report
    actions:
      - action: benchmark_report
        volume_name: "{{ volume_name }}"
        protocol: nvme-tcp
        client_node: m01
        save_as: bench_header

  # Phase 6: Connect NVMe
  - name: connect
    actions:
      - action: nvme_connect_direct
        node: m01
        target_addr: "10.0.0.1"
        target_port: "4431"
        nqn: "nqn.2024-01.com.seaweedfs:vol.{{ volume_name }}"
        expected_size: "2G"
        save_as: device
      - action: print
        msg: "Device: {{ device }}"

  # Phase 7: mkfs + mount FIRST (before any fio)
  - name: mkfs-mount
    actions:
      - action: exec
        node: m01
        cmd: "mkfs.ext4 -F -E nodiscard {{ device }} && mkdir -p /mnt/sw-bench && mount -o nodiscard {{ device }} /mnt/sw-bench && echo MOUNTED"
        root: "true"
        save_as: mount_result
      - action: assert_contains
        actual: "{{ mount_result }}"
        expected: "MOUNTED"

  # Phase 9: Preflight (verify mount + device)
  - name: preflight
    actions:
      - action: benchmark_preflight
        node: m01
        volume_name: "{{ volume_name }}"
        mount_path: /mnt/sw-bench
        device: "{{ device }}"

  # Phase 10: pgbench
  - name: pgbench
    actions:
      - action: exec
        node: m01
        cmd: "mkdir -p /mnt/sw-bench/pgdata && chown postgres:postgres /mnt/sw-bench/pgdata && sudo -u postgres /usr/lib/postgresql/16/bin/initdb -D /mnt/sw-bench/pgdata > /dev/null 2>&1 && sudo -u postgres /usr/lib/postgresql/16/bin/pg_ctl -D /mnt/sw-bench/pgdata -o '-p 5588 -k /tmp' -l /tmp/pg.log start && sleep 2 && sudo -u postgres createdb -p 5588 -h /tmp pgbench 2>/dev/null && sudo -u postgres pgbench -p 5588 -h /tmp -i -s 10 pgbench > /dev/null 2>&1 && echo PG_READY"
        root: "true"
        save_as: pg_status

      - action: exec
        node: m01
        cmd: "sudo -u postgres pgbench -p 5588 -h /tmp -c 4 -j 2 -T 30 pgbench 2>&1 | grep 'tps = ' | awk '{print $3}'"
        root: "true"
        save_as: pgbench_tps
        timeout: 60s

      - action: print
        msg: "pgbench TPS: {{ pgbench_tps }}"

  # Phase 11: Postcheck
  - name: postcheck
    actions:
      - action: benchmark_postcheck
        node: m01
        volume_name: "{{ volume_name }}"
        mount_path: /mnt/sw-bench
        device: "{{ device }}"
        pgdata_path: /mnt/sw-bench/pgdata
        save_as: postcheck_result
      - action: print
        msg: "Postcheck: {{ postcheck_result }}"

  # Phase 12: Collect results as markdown
  - name: results
    actions:
      - action: collect_results
        volume_name: "{{ volume_name }}"
        title: "Benchmark: sync_all RF=2 NVMe/TCP"
        write_iops: write_iops
        read_iops: read_iops
        pgbench_tps: pgbench_tps
        postcheck: postcheck_result
        save_as: report_md

  # Phase 13: Teardown (always runs)
  - name: teardown
    always: true
    actions:
      - action: exec
        node: m01
        cmd: "sudo -u postgres /usr/lib/postgresql/16/bin/pg_ctl -D /mnt/sw-bench/pgdata -m fast stop 2>/dev/null; true"
        root: "true"
        ignore_error: true

      - action: pre_run_cleanup
        node: m01
        kill_patterns: "postgres"
        unmount: "/mnt/sw-bench"
        nvme_disconnect: "true"

      - action: stop_weed
        node: m01
        pid: "{{ vs2_pid }}"
        ignore_error: true

      - action: stop_weed
        node: m02
        pid: "{{ vs1_pid }}"
        ignore_error: true

      - action: stop_weed
        node: m02
        pid: "{{ master_pid }}"
        ignore_error: true