BeSTGRID Auckland Cluster install notes

From BeSTGRID

Jump to: navigation, search

Contents

[edit] Installation Plan

[edit] Preparation

[edit] Install Hardware

Andrey
  1. Hardware audit. Done. Each server has mounting rails, screws and power cables (for rack plug and wall plug)
  2. Local hardware review using Terminal trolley and temporary power. Done. All hardware came in working condition.
  3. Network configuration requests
    1. Review MAC addresses for all NICs Done. List of MAC Addresses
    2. Log network requests for firewall, DNS, etc Request to Shirley Zhou for networkin has been submitted
  4. Hardware mounted to racks Done. Hardware positions
  5. Documentation Documented
ETA: 2 days, complete 29th February

[edit] Network and Power

Andrey & Shirley
  1. Power and Network cabling 3/03/08 - Power cabling performed. Waiting for network cabling
  2. Check power up of all nodes
  3. Review base install and configuration 3/03/08 - Done. Cluster Hardware Configuration
  4. Documentation
    1. Current network diagram
ETA: 2 days, complete 6th March

[edit] SGI Technical Specialist review

Andrey & Brian O'Conner, SGI, Melbourne
  1. review system setup and configuration
    1. There was a discussion about IPMI tool and installation and configuration BMC chip.
  2. system orientation
  3. Documentation
ETA: 1 days, 12th March

[edit] Post SGI visit review

Andrey
  1. review system setup and configuration
  2. system orientation
  3. Documentation
ETA: 1 days, 1st April

[edit] Test Cluster setup

Andrey, Yuriy
  1. Setup Headnode
  2. Setup Compute nodes - how to do so, using Red Hat Kick Start
  3. Setup Torque
  4. Setup Maui
  5. Install Bioinformatics applications
    1. Review and Document Rocks Rolls and Rocks WAN kickstart
  6. Install Globus, then use web services to access schedule
  7. Reconfigure NG2 configuration files, describing the Auckland Test Cluster
  8. Setup GridMap file for Test Cluster
  9. Documentation separated out from and updated from Computational GRID page

[edit] Gateway Upgrade & Configuration

Andrey
  1. upgrade VDT Gateway

[edit] cluster Iteration 1

Andrey & Yuriy
  1. Install Baseboard Management Controller and IPMI on Headnode to allow access outside of Data Centre
    1. BMC and IPMI have been configured on the headnode: Configuring BMC on the HeadNode
    2. Done
  2. Install Rocks onto Head Node w/ CentOS 4.5
    1. Rocks 4.3 Installation
    2. Rocks 5.0 Installation
    3. Done
  3. Compute Node Operating System images
    1. Create images into mode software image database
    2. Use RedHat Kick Start to provision images out to nodes
    3. Compute Nodes Setting
    4. Done
  4. Install Torque with Maui scheduler
    1. Done
  5. Install Compilers on Head Node
  6. Monitoring and Alerting
    1. Nagios setup
    2. Ganglia
    3. INCA
    4. Grid Operations Centre
  7. Documentation
ETA: 3 - 5 days, complete ??? April

[edit] Iteration 2

Andrey & yuriy
  1. Install Rocks onto Head Node w/ CentOS 4.5
    1. Install IPMItool and drivers on the Headnode to propogate them to compute nodes
  2. Compute Node Operating System images
    1. Create images into mode software image database
    2. Use RedHat Kick Start to provision images out to nodes
    3. Assign hostnames for compute nodes according their position in racks (compute-9-30)
  3. Install Torque with Maui scheduler - They are included into Rocks Torque roll. Installed
  4. Install Compilers on Head Node - gcc and gfortran are included into Rocks hpc roll. Installed
  5. Monitoring and Alerting
    1. Nagios setup
    2. Ganglia - Ganglia Roll. Installed
    3. INCA
    4. Grid Operations Centre
  6. Documentation
ETA: 3 - 5 days, complete ??? April

[edit] Iteration 3

Andrey & yuriy
  1. Install Rocks onto Head Node w/ CentOS 4.5
    1. Install IPMItool and drivers on the Headnode to propogate them to compute nodes
  2. Compute Node Operating System images
    1. Create images into mode software image database
    2. Use RedHat Kick Start to provision images out to nodes
    3. Assign hostnames for compute nodes according their position in racks (compute-9-30)
  3. Install Torque with Maui scheduler - They are included into Rocks Torque roll.
  4. Install Compilers on Head Node - gcc and gfortran are included into Rocks hpc roll.
  5. Monitoring and Alerting
    1. Nagios setup
    2. Ganglia - Ganglia Roll.
    3. INCA
    4. Grid Operations Centre
  6. Documentation
ETA: 2 days, complete 5th May

[edit] Application and Gateway

Andrey, Yuriy, Vladimir(?)
  1. Setup central module repository
    1. Install and Configure applications in central module repository
  2. Setup Gateway integration
ETA: 3 days, complete ??? April

[edit] Grid Accounting

  1. Review Gratia for grid accounting https://twiki.grid.iu.edu/twiki/bin/view/Accounting/WebHome

[edit] System Testing

[edit] End User Testing