You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Dec 16, 2025. It is now read-only.
runspec.py uses eventlet for concurrency, which uses green threads under a single Python process (under the GIL). Thus only one core will be used by a single runspec.py instance. To saturate the server I need to use multiple cores.
There appears to be a multi-instance mode for the client (runspec.py) but it is undocumented. It does seem to be used in run_multiple_instance.sh though.
I prefer a "one-stop shop" client like ab that handles multiple threads itself.
I created a patch that uses the multiprocess module for the requestBased mode here. Each process runs the existing client logic (eventlet pool, etc.), and runspec.py then merges the resulting log files for a comprehensive summary.