fix: add ollama embedding config and fix sqlite_vec db

[wukaixingxp]

What does this PR do?

RAG+Ollama is not working, as I encounter errors: Embeddings are now served via Inference providers. Taking a closer look it seems that the embedding model yaml config has been removed here, this PR will add the config back so user can use the embedding model after they ollama run all-minilm:latest.

Then, sqlite_vec also give error SQLite objects created in a thread can only be used in that same thread. The object was created in thread id 8349485120 and this is thread id 6325039104, I think we need to use self.connection = sqlite3.connect(self.config.db_path, check_same_thread=False) instead in :

llama-stack/llama_stack/providers/inline/vector_io/sqlite_vec/sqlite_vec.py

Line 165 in 1a044ef

self.connection = sqlite3.connect(self.config.db_path)

Test Plan

[Describe the tests you ran to verify your changes with result summaries. Provide clear instructions so the plan can be easily re-executed.]
Tested with my DocQA app

[ashwinb]

see comment inline

[wukaixingxp]

see comment inline

Thanks for your help! I run pre-commit locally and everything looks good, not sure why pre-commit failed at CI

pre-commit run --all-files
check for merge conflicts................................................Passed
check for added large files..............................................Passed
fix end of files.........................................................Passed
Insert license in comments...............................................Passed
ruff.....................................................................Passed
ruff-format..............................................................Passed
blacken-docs.............................................................Passed
uv-export................................................................Passed
mypy.....................................................................Passed
Distribution Template Codegen............................................Passed

[wukaixingxp]

see comment inline

Thanks for your help! I run pre-commit locally and everything looks good, not sure why pre-commit failed at CI

pre-commit run --all-files
check for merge conflicts................................................Passed
check for added large files..............................................Passed
fix end of files.........................................................Passed
Insert license in comments...............................................Passed
ruff.....................................................................Passed
ruff-format..............................................................Passed
blacken-docs.............................................................Passed
uv-export................................................................Passed
mypy.....................................................................Passed
Distribution Template Codegen............................................Passed

@ashwinb Can you take another look at this PR? somehow the pre-commit is blocking it, I am not sure how to solve

[ashwinb]

actually I think the correct fix needs to be what @ehhuang did in 270d640

[ashwinb]

You need to fetch and rebase to latest origin/main to get rid of the pre-commit error. But I think the threading error needs to be taken care of in a better way (as @ehhuang's commit shows.)

[wukaixingxp]

You need to fetch and rebase to latest origin/main to get rid of the pre-commit error. But I think the threading error needs to be taken care of in a better way (as @ehhuang's commit shows.)

@ashwinb Thanks for the guidance, I followed his example and tested with my RAG app. Can you take another look?

[ashwinb]

so this isn't quite right. you cannot assign whatever value you get into an instance variable. you must use conn = self._get_connection() anywhere you need to use a connection.