We are pleased to announce the official release of Lucia Training Platform v1.5.0!
Lucia Training Platform v1.5.0 Release Notes
This release focuses on platform stability improvements, deployment enhancements, hardware compatibility, and security updates.
Platform Features & Stability
- Added feature to track user's historical jobs in non-existing virtual clusters including the job status and logs
- Support for assigning custom job names
- Update the workflow trigger conditions to make the workflow triggered and executed only when a pull request targeting main, dev or release/* branch is closed
Deployment & Hardware Support
- Fixed kubespray deployment on bare metal when cffi package is installed by package manager
- Fixed NUMA parsing in job exporter on GB200 hardware
- Fixed job exporter compatibility issue on ARM nodes
- Enhanced multi-architecture support across platform components
Security
- Updated Kubernetes scheduler version to 1.33.1
- Updated Go version to 1.24.9 for framework controller, hivedscheduler, and watchdog
- Updated Node.js packages for alert-handler, job-status-change-notification, rest-server
- Updated Python packages for cluster local storage, copilot-chat, and dashboard-data-backup
- Updated RPM packages for database-controller
- Updated Docker version for webportal-dind
- Fixed security issues in multiple component dependencies
- Fixed the module "logger" missing when running DCGM with higher versions
Storage & Infrastructure
- Mounted SSH key pairs for cluster local storage
- Enhanced cluster local storage security and stability
- Fixed Docker pull problem after Docker version update