# Run from the pinned pdbench checkout after installing these dataset scripts.
datasource:
  path: s3://paradedb-benchmark/datasets/msmarco/v1/sampled/113m/parquet/msmarco/
  region: us-east-1

datastore:
  type: timescale
  url: ${PARADEDB_URL}
  # The shared cluster action uses this container name for every engine.
  container: paradedb
  dataset: datasets/msmarco
  table: msmarco
  settings:
    shared_buffers: 8GB
    maintenance_work_mem: 8GB
    max_parallel_maintenance_workers: 7
    max_wal_size: 32GB
  vars:
    index_name: msmarco_prism_idx
    with_options: "nlist = 0"

loader:
  expected_rows: 113520750
