M Best practices and tips

Abstract

概述

• Accounting file management
• Additional configuration settings
• Allocating CPUs as blocks for parallel jobs
• Cleaning up parallel job execution problems
• Configuring IBM Aspera as a data transfer tool
• Customizing output format and reducing network traffic for job queries
• Defining external host-based resources
• Enforcing job memory and swap with Linux cgroups
• Job access control
• Job directories
• Integration with AFS
• Maintaining cluster performance
• Managing floating software licenses
• Optimizing job processing with CPU frequency governors enabled
• OS partitioning and virtualization on Oracle Solaris and IBM AIX
• Performance metrics for mbatchd
• Placing jobs based on available job slots of hosts
• Running checksum to verify installation images
• Tracking job dependencies
• Using compute units for topology scheduling
• Using lsmake to accelerate Android builds
• Using NVIDIA DGX systems with LSF
• Using ssh X11 forwarding
• Using the LSF API Python wrapper