Systems Manual

Welcome to the Limulus Systems Manual.

Limulus Computing systems are available in either four or eight node desk-side cases or in a 4-node rack-mount unit. Multiple systems can be connected together and work as one cluster. There are two software stacks available: High Performance Computing (HPC) or scalable data analytics (Hadoop, Spark, Kafka, etc.).

The following Guide will help you navigate the Limulus system.

• Unpacking and Startup
• Essential Quick Start (HPC Systems)
• Essential Quick Start Data Analytics Systems)
• Getting Help

• Logging In
• User Software

• Using Software Modules
• Submitting Jobs to the Slurm Workload Manager
• Running Interactive Jobs on the Cluster Nodes
• Compiling and Running an MPI Application

• Running Hive and Spark from the Command line
• Using the Zeppelin Web Notebook

• Becoming the Administrative User
• Adding Users
• Powering Up/Down Nodes
• Adaptive Cooling
• Monitoring System Resources
• RAID Storage Management
• Setting Node Governors/Turbo Mode
• System-wide Administration Commands (pdsh)
• Software Management

• Open HPC Components
• HPC Logs and Log Management
• Logs and Log Management
• Slurm Workload Manager
• Warewulf Worker Node Images

• Using the Apache Ambari Cluster Manager
• HDFS Management
• Apache Logs and Log Management

• Limulus Design Philosophy
• Functional Diagram
• Desk-side Case
• Rack-mount Case
• Compute Blades
• Node Power Control
• Internal Networks
• Fan Control
• Limulus Services

• General Questions
• User Questions and Procedures
• Administrative Questions and Procedures
• Sandbox