HPE Performance Cluster Management Foundations Training in United States of America

Linux Fundamentals (GL120) U8583S or equivalent Linux or VMware experience

Module 1: Cluster Components

• Describe HPCM

• Describe clusters

• Describe clusters (flat) and distributed (hierarchical) clusters

• Define management, data and rack networks

• List cluster components

– Admin node

– Compute nodes

– Rack leader nodes

– ICE compute nodes

– Chassis management controllers (CMCs)

– Management interfaces (iLO, LO100i, BMC)

– Management of Ethernet switches

– Fabric switch switches and blades

– Smart PDUs

– Storage

Module 2: HPCM GUI

• Locate system groups

• Locate network groups

• Review image groups

• Use custom groups

• Manage nodes

• SSH to nodes

• Shutdown node

• Power off node

• Reboot node

• Illuminate the locate beacon to identify the chassis

• Run tasks on multiple systems

• Monitor the cluster

• Run commands to apply load on nodes

• View left chart to display CPU load

• View right chart to display memory usage

• Focus on single node

• Focus on group of nodes

• Switch to bar graph or table view

• Turn metrics off and on

• Use Ganglia to monitor cluster

• Use Nagios to monitor cluster

• Manage cluster power

• Monitor node power consumption

• Monitor job power consumption

• Monitor system power consumption

Module 3: Operating System Boot Modes and Root Filesystem Modes

• Describe disk mode

• Describe PXE mode

• Describe disk, nfs and tmpfs root file system mode

Module 4: Running Commands to Gather Information on the Cluster

• Describe pdsh

• Use pdsh commands to interrogate the cluster

• Describe output, interleaved stdout, stderr

• Review predefined pdsh node groups

• Add a pdsh node group

• Run a command with the new pdsh node group

• Run a command using the exclude node group format

• Format output from pdsh commands with dshbak –c

• Copy a file to all compute nodes, to all leader nodes, to all ice-compute nodes,

• Retrieve a file from all compute nodes, from all leader nodes, from all ice-compute nodes

• Work with HPCM connector to job scheduler

• Search central log repository