VectorDBBench icon indicating copy to clipboard operation
VectorDBBench copied to clipboard

Package dependency missing

Open cydrain opened this issue 1 month ago • 2 comments

Miss some package dependency for client "milvus" and "pgvector"

Reproduce steps:

  1. conda create -n tmp python=3.11
  2. conda activate tmp
  3. pip install vectordb-bench
  4. run command "vectordbbench milvushnsw --help", and see following error
(tmp) [root@gdb163 offline_pkgs]# vectordbbench milvushnsw --help                                                                                                                             
Traceback (most recent call last):
  File "/root/miniforge3/envs/tmp/bin/vectordbbench", line 3, in <module>
    from vectordb_bench.cli.vectordbbench import cli
  File "/root/miniforge3/envs/tmp/lib/python3.11/site-packages/vectordb_bench/cli/vectordbbench.py", line 1, in <module>
    from ..backend.clients.alloydb.cli import AlloyDBScaNN
  File "/root/miniforge3/envs/tmp/lib/python3.11/site-packages/vectordb_bench/backend/clients/alloydb/cli.py", line 9, in <module>
    from ....cli.cli import (
  File "/root/miniforge3/envs/tmp/lib/python3.11/site-packages/vectordb_bench/cli/cli.py", line 23, in <module>
    from ..interface import benchmark_runner, global_result_future
  File "/root/miniforge3/envs/tmp/lib/python3.11/site-packages/vectordb_bench/interface.py", line 15, in <module>
    from .backend.assembler import Assembler, FilterNotSupportedError
  File "/root/miniforge3/envs/tmp/lib/python3.11/site-packages/vectordb_bench/backend/assembler.py", line 6, in <module>
    from vectordb_bench.models import TaskConfig
  File "/root/miniforge3/envs/tmp/lib/python3.11/site-packages/vectordb_bench/models.py", line 7, in <module>
    import ujson
ModuleNotFoundError: No module named 'ujson'

  1. pip install ujson (fix the issue in step 4)
  2. pip install vectordb-bench pgvector
  3. run command "vectordbbench pgvectorhnsw xxx", and see following error
(tmp) [root@gdb163 offline_pkgs]# vectordbbench pgvectorhnsw --case-type Performance768D1M --user-name admin --password admin123 --host 10.229.46.162 --port 5432 --db-name vectordb_test --m 16 --ef-construction 256 --ef-search 100
2025-11-06 09:40:01,405 | INFO: Task:
TaskConfig(db=<DB.PgVector: 'PgVector'>, db_config=PgVectorConfig(db_label='2025-11-06T09:40:01.281077', version='', note='', user_name=SecretStr('**********'), password=SecretStr('**********'), host='10.229.46.162', port=5432, db_name='vectordb_test', table_name='vdbbench_table_test'), db_case_config=PgVectorHNSWConfig(metric_type=None, create_index_before_load=False, create_index_after_load=True, iterative_scan='relaxed_order', m=16, ef_construction=256, ef_search=100, index=<IndexType.ES_HNSW: 'hnsw'>, maintenance_work_mem=None, max_parallel_workers=None, quantization_type=None, table_quantization_type=None, reranking=False, quantized_fetch_limit=None, reranking_metric='COSINE'), case_config=CaseConfig(case_id=<CaseType.Performance768D1M: 5>, custom_case={}, k=100, concurrency_search_config=ConcurrencySearchConfig(num_concurrency=[1, 5, 10, 20, 30, 40, 60, 80], concurrency_duration=30, concurrency_timeout=3600)), stages=['drop_old', 'load', 'search_serial', 'search_concurrent'])                                                                                                                                                
 (cli.py:635) (1520147)                                                                                                                                                                       
2025-11-06 09:40:01,406 | INFO: generated uuid for the tasks: 713fd5d88a294bcf8c8a4559b6b18892 (interface.py:76) (1520147)
2025-11-06 09:40:01,407 | WARNING: Please install client for database, error=No module named 'psycopg' (interface.py:92) (1520147)

cydrain avatar Nov 06 '25 01:11 cydrain

The missing packages includes: ujson-5.11.0-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl (for all) psycopg-3.2.12-py3-none-any.whl (for pgvector) psycopg_binary-3.2.12-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl (pgvector)

cydrain avatar Nov 06 '25 01:11 cydrain

ujson++

 bin/vectordbbench pgvectorhnsw --config-file pgvector-hnsw.yaml
Traceback (most recent call last):
....
.../vectordb-bench/lib/python3.12/site-packages/vectordb_bench/models.py", line 7, in <module>
    import ujson
ModuleNotFoundError: No module named 'ujson'

mxr576 avatar Nov 18 '25 16:11 mxr576