-
Notifications
You must be signed in to change notification settings - Fork 535
feat: implement concurrent message reading for session managers #897
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat: implement concurrent message reading for session managers #897
Conversation
Replace sequential message loading with async concurrent reading in both S3SessionManager and FileSessionManager to improve performance for long conversations. Uses asyncio.gather() with run_in_executor() to read multiple messages simultaneously while maintaining proper ordering. Resolves: strands-agents#874
Unshure
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This all looks good to me! Do you have any performance data on this? Wondering if we are seeing any latency improvement for loading ~100 messages from session?
Yes, I had done a test during implementation and there was a major improvement in the read speed. Almost 2.5x faster for local files but s3 will have the major visible improvement. |
…rs (strands-agents#897)" This reverts commit 08dc4ae.
* feat: replace kwargs with invocation_state in agent APIs * fix: handle **kwargs in stream_async. * feat: add a unit test for the change * Update src/strands/agent/agent.py Co-authored-by: Nick Clegg <[email protected]> * tool - executors - concurrent - remove no-op gather (#954) * feat(telemetry): updated traces to match OTEL v1.37 semantic conventions (#952) * event loop - handle model execution (#958) * feat: implement concurrent message reading for session managers (#897) Replace sequential message loading with async concurrent reading in both S3SessionManager and FileSessionManager to improve performance for long conversations. Uses asyncio.gather() with run_in_executor() to read multiple messages simultaneously while maintaining proper ordering. Resolves: #874 Co-authored-by: Vamil Gandhi <[email protected]> * hooks - before tool call event - cancel tool (#964) * fix(telemetry): removed double serialization for events (#977) * fix(litellm): map LiteLLM context-window errors to ContextWindowOverflowException (#994) * feat: add more tests and adjust invocation_state dic structure * Apply suggestion from @Unshure Co-authored-by: Nick Clegg <[email protected]> * fix: adjust **kwargs in multiagent primitives --------- Co-authored-by: Nick Clegg <[email protected]> Co-authored-by: Patrick Gray <[email protected]> Co-authored-by: poshinchen <[email protected]> Co-authored-by: Vamil Gandhi <[email protected]> Co-authored-by: Vamil Gandhi <[email protected]> Co-authored-by: ratish <[email protected]>
…nds-agents#897) Replace sequential message loading with async concurrent reading in both S3SessionManager and FileSessionManager to improve performance for long conversations. Uses asyncio.gather() with run_in_executor() to read multiple messages simultaneously while maintaining proper ordering. Resolves: strands-agents#874 Co-authored-by: Vamil Gandhi <[email protected]>
…nts#966) * feat: replace kwargs with invocation_state in agent APIs * fix: handle **kwargs in stream_async. * feat: add a unit test for the change * Update src/strands/agent/agent.py Co-authored-by: Nick Clegg <[email protected]> * tool - executors - concurrent - remove no-op gather (strands-agents#954) * feat(telemetry): updated traces to match OTEL v1.37 semantic conventions (strands-agents#952) * event loop - handle model execution (strands-agents#958) * feat: implement concurrent message reading for session managers (strands-agents#897) Replace sequential message loading with async concurrent reading in both S3SessionManager and FileSessionManager to improve performance for long conversations. Uses asyncio.gather() with run_in_executor() to read multiple messages simultaneously while maintaining proper ordering. Resolves: strands-agents#874 Co-authored-by: Vamil Gandhi <[email protected]> * hooks - before tool call event - cancel tool (strands-agents#964) * fix(telemetry): removed double serialization for events (strands-agents#977) * fix(litellm): map LiteLLM context-window errors to ContextWindowOverflowException (strands-agents#994) * feat: add more tests and adjust invocation_state dic structure * Apply suggestion from @Unshure Co-authored-by: Nick Clegg <[email protected]> * fix: adjust **kwargs in multiagent primitives --------- Co-authored-by: Nick Clegg <[email protected]> Co-authored-by: Patrick Gray <[email protected]> Co-authored-by: poshinchen <[email protected]> Co-authored-by: Vamil Gandhi <[email protected]> Co-authored-by: Vamil Gandhi <[email protected]> Co-authored-by: ratish <[email protected]>
Description
Replace sequential message loading with async concurrent reading in both S3SessionManager and FileSessionManager to improve performance for long conversations. Uses asyncio.gather() with run_in_executor() to read multiple messages simultaneously while maintaining proper ordering.
Related Issues
Resolves: #874
Type of Change
New feature
Testing
How have you tested the change? Verify that the changes do not break functionality or introduce warnings in consuming repositories: agents-docs, agents-tools, agents-cli
hatch run prepareChecklist
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.