Fix: Properly close OpenAI SSE stream on interruption #2902
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
When an InterruptionFrame causes task cancellation, the OpenAI AsyncStream was not being properly closed, which could lead to connection leaks and inconsistent stream state. This fix wraps the stream processing in a try/except/finally block to: - Catch asyncio.CancelledError from interruptions - Ensure the stream is always closed via chunk_stream.close() - Log the cancellation for debugging purposes This prevents SSE connection leaks when users interrupt the bot mid-response.
Codecov Report❌ Patch coverage is
🚀 New features to boost your workflow:
|
jamsea
commented
Oct 22, 2025
| raise | ||
| finally: | ||
| # Ensure the SSE stream is properly closed to avoid connection leaks | ||
| await chunk_stream.close() |
There was a problem hiding this comment.
@aconchillo are we intentionally not calling this function presently? The main change I'm interested in is line 431, adding a debug log
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Outdated suggestions cannot be applied.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.
Problem
When an
InterruptionFrameis received (e.g., when a user interrupts the bot mid-response), it triggers task cancellation viaasyncio.CancelledError. However, the OpenAIAsyncStream[ChatCompletionChunk]was not being properly closed, which could lead to:Root Cause
The
_process_context()method inbase_llm.pyiterates over the SSE stream with:When an
InterruptionFramearrives, the frame processor cancels the current process task, but the stream is not explicitly closed, leaving the underlying HTTP/SSE connection in an inconsistent state.Solution
This PR wraps the entire stream processing section with proper exception handling:
asyncio.CancelledError, logs it for debugging, and re-raises it to maintain proper cancellation propagationchunk_stream.close()is always called to properly clean up the SSE connectionChanges
async for chunk in chunk_stream:loopawait chunk_stream.close()in the finally block