fix: _handle_async_stream_response() in OpenAIChatGenerator handles asyncio.CancelledError closing the response stream
#10175
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
davidsbatista
requested review from
Amnah199 and
mpangrazzi
and removed request for
a team
Amnah199
approved these changes
Dec 3, 2025
Contributor
Amnah199
left a comment
Amnah199
left a comment
There was a problem hiding this comment.
LG to me! But I would wait for @mpangrazzi 's review
mpangrazzi
approved these changes
Dec 10, 2025
...AIChatGenerator-handles-asyncio.CancelledError-closing-response-stream-37e1e85255e1dc41.yaml
Outdated
Show resolved
Hide resolved
…celledError-closing-response-stream-37e1e85255e1dc41.yaml Co-authored-by: Michele Pangrazzi <[email protected]>
davidsbatista
deleted the
fix/close-openao-asyncio-stream-on-cancellation
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Related Issues
Proposed Changes:
The
_handle_async_stream_response()method inOpenAIChatGeneratordid not handleasyncio.CancelledErrorexceptions. When a streaming task was cancelled mid-stream, the async for loop would exit without properly closing the OpenAI stream object - leading to tokens still being sent and charged + potential resource leaks, open connections.Added exception handling in
_handle_async_stream_response()to catchasyncio.CancelledErrorand gracefully close the stream withasyncio.shield()to ensure the cleanup operation completes even during cancellation.How did you test it?
test_async_stream_closes_on_cancellation(): mocks an async stream, starts a streaming task, cancels it mid-stream, and verifies that stream.close() is called exactly once.test_run_async_cancellation_integration(): Tests against the real OpenAI API by starting a long-running streaming task (essay generation), cancelling it after 2 seconds, and verifying graceful cancellation with partial chunks received.Minor fix in
test_openai_responses.py: added missing OPENAI_API_KEY environment variable setup via monkeypatch.Notes for the reviewer
This PR also adds the same functionality for the following ChatGenerators (most are part of
haystack-core-integrationsonly the first is not):AzureOpenAIChatGeneratorNvidiaChatGeneratorAIMLAPIChatGeneratorMistralChatGeneratorSTACKITChatGeneratorOpenRouterChatGeneratorTogetherAIChatGeneratorsince they all inherit from the
OpenAIChatGenerator.Checklist
fix:,feat:,build:,chore:,ci:,docs:,style:,refactor:,perf:,test:and added!in case the PR includes breaking changes.