feat(ml): add qdrant ingestion#38142
Conversation
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly enhances Apache Beam's machine learning capabilities by integrating Qdrant, a popular vector database. It provides a complete framework for ingesting embedded data into Qdrant collections, allowing for scalable and efficient management of vector embeddings within RAG pipelines. The changes include new configuration options, data transformation logic, and thorough testing to ensure reliability and correctness. Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. Footnotes
|
MichaelGruschke
force-pushed
the
qdrant
branch
2 times, most recently
from
3a266b5 to
efe7cf4
Compare
|
Checks are failing. Will not request review until checks are succeeding. If you'd like to override that behavior, comment |
|
Qdrant integration tests in the respective Python ML test suites all pass. |
|
assign set of reviewers |
|
Assigning reviewers: R: @jrmccluskey for label python. Note: If you would like to opt out of this review, comment Available commands:
The PR bot will only process comments in the main thread (not review comments). |
|
Assigning reviewers: R: @jrmccluskey for label python. Note: If you would like to opt out of this review, comment Available commands:
The PR bot will only process comments in the main thread (not review comments). |
|
Reminder, please take a look at this pr: @jrmccluskey |
| * Updates minimum Go version to 1.26.1 ([#37897](https://github.com/apache/beam/issues/37897)). | ||
| * (Python) Added image embedding support in `apache_beam.ml.rag` package ([#37628](https://github.com/apache/beam/issues/37628)). | ||
| * (Python) Added support for Python version 3.14 ([#37247](https://github.com/apache/beam/issues/37247)). | ||
| * (Python) Added [Qdrant](https://qdrant.tech/) VectorDatabaseWriteConfig implementation ([#38141](https://github.com/apache/beam/issues/38141)). |
There was a problem hiding this comment.
Move to the 2.74.0 release notes
|
|
||
| import logging | ||
| from dataclasses import dataclass, field | ||
| from typing import Any, Callable, Dict, Optional |
There was a problem hiding this comment.
Use collections.abc types for Callable, use the built-in dict type for Dict hints
There was a problem hiding this comment.
thanks for pointing that out, should be fixed now!
|
|
||
|
|
||
| @dataclass | ||
| class QdrantConnectionParameters: |
There was a problem hiding this comment.
A docstring outlining each field and the mandatory information required to create a valid set of parameters will make this much more user friendly
There was a problem hiding this comment.
good point! added docstring
|
|
||
|
|
||
| @dataclass | ||
| class QdrantWriteConfig(VectorDatabaseWriteConfig): |
There was a problem hiding this comment.
Similar thought here, a docstring should be provided here since this is the entrypoint for users to drop the qdrant write into their pipelines
refactor: use local qdrant implementation for tests chore: clean up imports chore: add qdrant dependency to ml_test extra chore: run precommit chore: add comment to CHANGES.md fix: guard against import error fix: import
|
Assigning new set of reviewers because Pr has gone too long without review. If you would like to opt out of this review, comment R: @damccorm for label python. Available commands:
|
|
Stopping reviewer notifications for this pull request: review requested by someone other than the bot, ceding control. If you'd like to restart, comment |
|
/gemini review |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces the Qdrant VectorDatabaseWriteConfig implementation for the Python SDK, allowing for data ingestion into Qdrant collections. The changes include new connection and write configuration classes, a dedicated Beam PTransform, and comprehensive integration tests. Review feedback highlights opportunities to improve the robustness of the implementation by moving batch initialization to start_bundle to avoid state leakage, adding validation for the batch_size parameter, and removing the override that disables client-server compatibility checks.
| def __init__(self, config: QdrantWriteConfig): | ||
| self.config = config | ||
| self._batch = [] | ||
| self._client: "Optional[QdrantClient]" = None |
There was a problem hiding this comment.
In Apache Beam, DoFn instances can be reused across bundles. To ensure that state is not leaked between bundles (especially in case of retries), it is a best practice to initialize bundle-specific state like self._batch in the start_bundle method rather than in __init__.
| def __init__(self, config: QdrantWriteConfig): | |
| self.config = config | |
| self._batch = [] | |
| self._client: "Optional[QdrantClient]" = None | |
| def __init__(self, config: QdrantWriteConfig): | |
| self.config = config | |
| self._client: "Optional[QdrantClient]" = None | |
| def start_bundle(self): | |
| self._batch = [] |
| def __post_init__(self): | ||
| if not self.collection_name: | ||
| raise ValueError("Collection name must be provided") |
There was a problem hiding this comment.
It is recommended to validate that batch_size is a positive integer in __post_init__ to prevent potential issues with empty or negative batch sizes during ingestion.
| def __post_init__(self): | |
| if not self.collection_name: | |
| raise ValueError("Collection name must be provided") | |
| def __post_init__(self): | |
| if not self.collection_name: | |
| raise ValueError("Collection name must be provided") | |
| if self.batch_size <= 0: | |
| raise ValueError("batch_size must be positive") |
| check_compatibility=False, | ||
| **params.kwargs, |
There was a problem hiding this comment.
Disabling the compatibility check (check_compatibility=False) can lead to difficult-to-debug issues if there is a version mismatch between the client and the Qdrant server. Unless there is a specific reason to disable it, it is safer to leave it enabled (which is the default).
| check_compatibility=False, | |
| **params.kwargs, | |
| **params.kwargs, |
| def process(self, element, *args, **kwargs): | ||
| self._batch.append(element) | ||
| if len(self._batch) >= self.config.batch_size: | ||
| self._flush() |
There was a problem hiding this comment.
Consider adding a byte size limit for individual batches, similar to BigQuery streaming inserts
beam/sdks/python/apache_beam/io/gcp/bigquery.py
Line 1655 in efe4e94
| return | ||
| if not self._client: | ||
| raise RuntimeError("Qdrant client is not initialized") | ||
| self._client.upsert( |
There was a problem hiding this comment.
Are there any retriable errors that we should handle?
|
|
||
|
|
||
| @dataclass | ||
| class QdrantConnectionParameters: |
There was a problem hiding this comment.
Can we add classmethod factories to make it clearer which combinations of parameters are valid? Something like
@dataclass
class QdrantConnectionParameters:
# ... existing fields unchanged ...
@classmethod
def for_cloud(
cls,
url: str,
api_key: str,
*,
prefer_grpc: bool = False,
timeout: Optional[int] = None,
**kwargs: Any,
) -> "QdrantConnectionParameters":
"""Connect to Qdrant Cloud. Requires the cluster URL and an API key."""
return cls(
url=url,
api_key=api_key,
https=True,
prefer_grpc=prefer_grpc,
timeout=timeout,
kwargs=kwargs,
)
@classmethod
def for_host(
cls,
host: str,
port: int = 6333,
*,
grpc_port: int = 6334,
prefer_grpc: bool = False,
https: bool = False,
api_key: Optional[str] = None,
timeout: Optional[int] = None,
**kwargs: Any,
) -> "QdrantConnectionParameters":
"""Connect to a self-hosted Qdrant instance by host and port."""
return cls(
host=host, port=port, grpc_port=grpc_port,
prefer_grpc=prefer_grpc, https=https,
api_key=api_key, timeout=timeout, kwargs=kwargs,
)
@classmethod
def for_url(
cls,
url: str,
*,
api_key: Optional[str] = None,
prefer_grpc: bool = False,
timeout: Optional[int] = None,
**kwargs: Any,
) -> "QdrantConnectionParameters":
"""Connect using a full URL like 'https://my-qdrant.example.com:6333'."""
return cls(url=url, api_key=api_key, prefer_grpc=prefer_grpc,
timeout=timeout, kwargs=kwargs)
@classmethod
def local(cls, path: str) -> "QdrantConnectionParameters":
"""Use an embedded Qdrant instance persisted to the given path."""
return cls(path=path)
@classmethod
def in_memory(cls) -> "QdrantConnectionParameters":
"""Use an embedded in-memory Qdrant instance. Useful for tests."""
return cls(location=":memory:")
4 participants
resolves #38141
This PR adds support for Qdrant vector database ingestion to Apache Beam's ML RAG pipeline.
Implementation details:
QdrantConnectionParameters
QdrantWriteConfig
QdrantWriteTransform
Dependency Changes
5.Tests
Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:
addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, commentfixes #<ISSUE NUMBER>instead.CHANGES.mdwith noteworthy changes.See the Contributor Guide for more tips on how to make review process smoother.
To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md
GitHub Actions Tests Status (on master branch)
See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.