Two problems in running rag-conent sample script with Llama Stack 0.4.3

[TamiTakamiya]

Today (Feb 1, 2026) I have tried to run the sample script in README.md with the latest code based on Llama Stack 0.4.3 and found two issues:

1. Value Error in running query_rag.py script.

I followed the instructions in README.md and created vector DB with ./custom_docs/0.1/info.txt and custom_processory.py. Unlike the previous Llama Stack 0.3.5, the script terminated properly without pressing Ctrl-C. However, the following scripts/quey_rag.py failed with:

/home/ttakamiy/git/lightspeed-core/rag-content/.venv/bin/python scripts/query_rag.py -p vector_db/custom_docs/0.1 -x custom-docs-0_1 -m embeddings_model -k 5 -q how can I configure a cinder backend 
Command line used: scripts/query_rag.py -p vector_db/custom_docs/0.1 -x custom-docs-0_1 -m embeddings_model -k 5 -q how can I configure a cinder backend
Traceback (most recent call last):
  File "/home/ttakamiy/git/lightspeed-core/rag-content/scripts/query_rag.py", line 324, in <module>
    _llama_stack_query(args)
  File "/home/ttakamiy/git/lightspeed-core/rag-content/scripts/query_rag.py", line 191, in _llama_stack_query
    with lib_client.LlamaStackAsLibraryClient(cfg_file) as client:
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/library_client.py", line 154, in __init__
    loop.run_until_complete(self.async_client.initialize())
  File "/usr/lib64/python3.12/asyncio/base_events.py", line 691, in run_until_complete
    return future.result()
           ^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/library_client.py", line 290, in initialize
    await self.stack.initialize()
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/stack.py", line 596, in initialize
    await register_resources(self.run_config, impls)
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/stack.py", line 232, in register_resources
    await method(**{k: getattr(obj, k) for k in obj.model_dump().keys()})
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/routing_tables/vector_stores.py", line 95, in register_vector_store
    await self.register_object(vector_store)
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/routing_tables/common.py", line 228, in register_object
    await self.dist_registry.register(obj)
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/store/registry.py", line 168, in register
    success = await super().register(obj)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/store/registry.py", line 99, in register
    raise ValueError(
ValueError: Object of type 'vector_store' and identifier 'vs_b60854ea-e555-4775-8a58-2775eec9481c' already exists. Unregister it first if you want to replace it.

Process finished with exit code 1

Note

A workaround for this errir is to remove /tmp/kv_store.db

2. Another issue in query_rag.py. With the workaround, the script went further, but failed with:

Traceback (most recent call last):
  File "/home/ttakamiy/git/lightspeed-core/rag-content/scripts/query_rag.py", line 324, in <module>
    _llama_stack_query(args)
  File "/home/ttakamiy/git/lightspeed-core/rag-content/scripts/query_rag.py", line 197, in _llama_stack_query
    res = client.vector_io.query(
          ^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack_client/resources/vector_io.py", line 121, in query
    return self._post(
           ^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack_client/_base_client.py", line 1252, in post
    return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/library_client.py", line 225, in request
    result = loop.run_until_complete(self.async_client.request(*args, **kwargs))
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib64/python3.12/asyncio/base_events.py", line 691, in run_until_complete
    return future.result()
           ^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/library_client.py", line 390, in request
    response = await self._call_non_streaming(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/library_client.py", line 448, in _call_non_streaming
    result = await matched_func(**body)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/routers/vector_io.py", line 144, in query_chunks
    return await self.routing_table.query_chunks(vector_store_id, query, params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/routing_tables/vector_stores.py", line 114, in query_chunks
    await self.assert_action_allowed("read", "vector_store", vector_store_id)
  File "/home/ttakamiy/git/lightspeed-core/rag-content/.venv/lib64/python3.12/site-packages/llama_stack/core/routing_tables/common.py", line 240, in assert_action_allowed
    raise ValueError(f"{type.capitalize()} '{identifier}' not found")
ValueError: Vector_store 'custom-docs-0_1' not found

Process finished with exit code 1

Note

This seems to require a code change, i.e. replacing custom-docs-0_1 with the vector DB ID (e.g. vs_b60854ea-e555-4775-8a58-2775eec9481c). If vector_io.query call in query_rag.py is modified to

      vector_store_id = cfg['registered_resources']['vector_stores'][0]['vector_store_id']
      res = client.vector_io.query(
          vector_store_id=vector_store_id, query=args.query, params=query_cfg
          #           vector_store_id = args.product_index, query = args.query, params = query_cfg
      )

It could retrieve the data from the vector store.