HPE Performance Cluster Management Administration - H8PE9S

Ideal candidate for this course:

• Attend this class if you need to learn to install, configure and administer clusters managed with the HPE Performance Cluster Manager (HPCM)
• Experienced Linux system administrators

The following knowledge is recommended for this seminar:

• H8PE8S: HPE Performance Cluster Management Foundations
• The following Linux system administration skills are prerequisites for this course:
  • Edit text with the vi editor
  • Recognize regular expression syntax
  • Access documentation with man and info file viewers
  • Monitor, manage and maintain log files
  • Enter common commands at the bash command line; create and interpret basic bash shell scripts
  • Install and configure standard software components, services and security features
  • Configure basic communication protocols
  • Create and modify crontabs
  • Monitor resources usage; be familiar with basic monitoring tools
  • Install and configure a Linux distribution on a server
  • Create, modify, and delete user accounts and group accounts
  •  Partition disks, manage filesystems and logical volumes
  •  Use RPM package management
  •  Install and use virtualized systems
  •  Understand basic hardware and hardware troubleshooting

Install Cluster

• Describe HPCM features 
• Define operating system slots
• Build cluster from ground up 
• Provision node with GUI 
• Provision node with command line
• Add nodes to the cluster
• Explore auto installation tools

Discover

• Discover nodes
• Interpret cluster configuration files
• Review cluster services

Data Networks

• Describe technologies
• Describe InfiniBand configuration
• Describe Intel Omni-Path configuration 
• Describe software components
• Use diagnostic commands

Manage Images

• Manage software repositories
• List software repositories
• Add software repositories
• Remove software repositories
• Create repository groups
• Customize an image by using RPM lists
• Create a compute node image
• Create an ICE-compute node image
• Manage image version control
• Check in an image into version control
• Compare differences between two versions of an image
• List the versions of an image
• Deploy a specific version of an image
• Push an ICE-compute image to a rack 
• Use parallel tools and inbuilt functionality to check differences between nodes
• Enable hyperthreading 
• Disable hyperthreading
• Configure array services
• Install batch scheduler server on a compute node
• Install batch scheduler client on a compute node and in ICE compute node
• Configure HPCM connectors to job schedulers
• Capture an image from a node (golden)
• Add RPMs to, remove RPMs from, and version control compute images
• Add and remove RPMs from running compute nodes
• Clone an ICE-compute image
• Clean up old images on the lead node
• Add RPMs to ICE compute image Compare when and when not to use tmpfs root
• Determine which nodes use tmpfs root
• Configure nodes to use tmpfs root
• List tmpfs quota difference (rack leader quotas do not apply when ICE-compute nodes are in tmpfs)
• Set tmpfs mode
• Set disk mode
• Show which mode a node has booted with
• Show which mode a node is scheduled to boot into 
• Perform a clone operating system slot operation

Automate Post Installation Tasks

• Review conf.d scripts
• Exclude a conf.d script
• Use pre_reconf.sh
• Use reconfig.sh
• Develop post install and per-host customization scripts

Configure Shared Filesystem, User Accounts, Applications, and Updates

• NFS Export a filesystem on a compute node
• Mount an NFS filesystem and create a user on an ICE compute node
• Manage user accounts
• Synchronize UIDs and GIDs, LDAP, etc.
• Run an application on compute and ICE compute nodes 
• Display BIOS settings
• Upgrade firmware
• Update kernel
• Update distribution
• Update HPCM

Troubleshoot Cluster 

• Backup cluster configuration
• Backup managed network switch configuration
• Use the central log repository 
• Investigate log files
• Gather system information
• Interrogate iLOs, BMCs
• Confirm resources
• Create pdsh groups
• Investigate  bond devices
• Inspect VLAN devices
• Capture a node crash dump
• Transfer an image from another slot or another system and confirm that the image can be used.
• Inject faults

Review Cluster Security

• Describe system administrator configurable security tasks 
• Describe what makes cluster security different from standalone security (how would change X break the cluster) 
• List ports used for each node role and for which interfaces 
• List components with passwords

                    – Admin node 
                    – Flat compute nodes 
                    – Rack leader nodes 
                    – ICE compute nodes 
                    – BMCs 
                    – CMCs 
                    – Ethernet network switches 
                    – InfiniBand and Omni-Path switches 
                    – IB/OPA switch BMCs 
                    – Storage controllers 

• List components that can have passwords applied