Investigation of a Workbench UI Latency Issue
Netflix's Workbench UI latency issue was traced to a JupyterLab extension, revealing performance dependencies on CPU and memory usage.
MAIN POINTS
- Workbench UI slowdown linked to specific JupyterLab notebooks, requiring detailed debugging from UI to Linux kernel.
- JupyterLab extension `jupyter-resource-usage` caused CPU contention due to inefficient memory statistics collection.
- Disabling the problematic extension resolved the UI latency, emphasizing the impact of system resource monitoring.
TAKEAWAYS
- UI responsiveness can be affected by extensions meant for monitoring, highlighting potential inefficiencies.
- CPU and memory usage have interrelated impacts on performance, contrary to common separate considerations.
- Efficient debugging requires a comprehensive understanding of both software and hardware interactions.