Change Log
New Features
Warm Start/Stop Optimization
Implemented an optimized warm start/stop mechanism, reducing cluster bootstrap time to under one minute. This improves system efficiency by minimizing resource allocation delays.
File Management in Workspaces
Added file management capabilities to workspaces. Users can now upload, delete, and edit files directly from the UI, streamlining job creation and notebook execution.
Yeedu Functions
Introduced a new job type: Yeedu Function. These jobs run in a fully managed, auto-scaling environment, accepting APIs as function parameters and supporting execution via triggers or events.
PrimeVue Integration
Integrated PrimeVue components into the UI to improve responsiveness, enhance performance, and deliver a more modern and interactive user experience.
Enhancements
Cluster Auto-Scaling
Enhanced the auto-scaling logic to dynamically scale resources as soon as a job enters the queue, significantly reducing wait times.
OSM File Synchronization
Fixed synchronization issues with OSM files having identical names, ensuring consistent and reliable file handling across containers.
Machine ID Inclusion in Data Model
Ensured machine_id is correctly included in the data model during cluster downsize operations, improving warm start/stop reliability.
Job Kill Process
Improved job termination logic to ensure PIDs and all child processes are properly terminated, preventing lingering jobs.
Memory Optimization
Updated memory configurations for Yeedu components to reduce overhead and prevent container failures due to memory constraints.
Auto Start Configuration
Users can no longer disable auto start but are now able to configure the timeout value based on workload needs.
Bug Fixes
Message Acknowledgment
Fixed an issue where messages from errored jobs were not acknowledged, ensuring accurate job state tracking and message handling.