Home » Support » Training and Outreach » Seminar Series Archive » Using WestGrid QuickStart - Seminar Outline

Using WestGrid Systems - QuickStart for New Users

Outline of 2010-10-06 Seminar

(Please note that some material in these notes is out of date.  For example, the Matrix cluster has been retired.  See the QuickStart Guide for New Users for similar, but, more recently updated material).

 

 

Introduction

  • Speaker

    • Doug Phillips, Scientific Computing Consultant, U of Calgary

  • Scope of talk

    • Overview of WestGrid organization and resources

    • QuickStart Guide for New Users

    • Explore other parts of WestGrid web site

    • Log in to one or more WestGrid systems to monitor queues, submit jobs, etc.

WestGrid Overview

Getting Started

Choosing Which System to Use

  • See WestGrid Resources & Services or site-specific QuickStart guides.

  • Small-memory serial jobs or undemanding parallel jobs

    • Typically run on the Glacier cluster, but, Hermes is a newer alternative.

  • OpenMP-based parallel jobs and large-memory (> 4GB) serial jobs

    • Run on shared-memory or fat-node architectures
      (Cortex and Breezy nodes).

  • MPI-based parallel programs requiring high-performance interconnect

    • Try the Bugaboo, Checkers, Nestor and Orcinus clusters.

    • Bugaboo is preferred if you have large active storage requirements (100s of GB). 

    • Smaller Snowpatch and older Matrix cluster can also be used. Compare to systems available through Cortex.

  • Visualization requiring graphics/visualization hardware and software.

  • Software license considerations

    • Commercial license for the Gaussian only on Checkers cluster.

    • MATLAB available on some home-institution machines and to all on Orcinus with caveat.

    • See WestGrid software page.

 

Setting up Your Computer

  • Terminal client supporting ssh

  • File transfer client supporting scp and sftp

    • Command line programs scp and sftp on Linux or Mac OS X computers.

    • pscp and psftp come with PuTTY for Microsoft Window.

  • X Window display server for graphics

    • X11 tunnelling through ssh client.

    • X Window display servers such as Xming for Microsoft Windows, X11 for Mac OS or X for Linux.


Connecting and Logging In

  • Domain Name System (DNS)  test at http://westgrid.ca/iptest .

  • If user name different:

    • ssh your_username@matrix.westgrid.ca

  • If user name same:

    • ssh matrix.westgrid.ca

  • For X Window tunnelling:

    • ssh -X matrix.westgrid.ca

    • ssh -Y matrix.westgrid.ca  (Mac)

 

Working Interactively

  • Interactive vs. Batch

  • The UNIX environment

    • Shell - bash or tsch

    • UNIX Tutorial for Beginners (University of Surrey).

    • The UNIX man command (man for "manual")

    • Startup files: .bashrc, .bash_profile or .login and .cshrc

    • No Microsoft Windows binaries (recompile)

  • File systems

    • Home and scratch directories.

    • Avoid names with spaces.

    • UNIX is case sensitive

  • Transferring files

    • Use file transfer software that supports SCP (Secure Copy) or SFTP (SSH File Transfer Protocol).

    • Move files between directories using the UNIX mv command.

    • Use gcp (grid copy) to transfers files between WestGrid systems.

    • For long term storage of large files use the WestGrid storage facilities.

    • Line ending issue (use dos2unix if necessary).

  • Editing files

    • UNIX editors: emacs, vi.

    • PC-like editor: nedit (requires X Windows).

    • Use UNIX utilities to examine files (cat, more or less, head, tail).

    • Search for files with find and within files using grep.

  • Running interactive programs

    • Set up X Windows if running a graphical program.

    • Type executable file name (or ./name if "." not on PATH).

    • "echo $PATH" to list directories checked for commands.

    • File input and output redirection (program_name < input > output).

  • Restrictions on interactive jobs

    • See Working Interactively for more information.

    • Respect local policies on time limits and processor numbers.

Software

  • Locating installed software

  • Installing your own software

    • Set up a "bin" directory and add that directory to your PATH.

    • Control access through share groups or UNIX permissions.

  • Software licensing

    • Licensing policies vary with package.

    • Most licenses come from a host institution.

 

Programming

  • See programming pages.

  • Tables on the software page list compilers and the numerical (and other) libraries.

  • Set up compiler with modules if necessary.

  • Try your code with more than one compiler.

  • Add checkpointing and restart capability to your program.

  • WestGrid support staff would be happy to help with debugging.

 

Running Batch Jobs

  • The batch environment

    • Job scripts submitted to the TORQUE resource manager

    • Scheduler (Moab or Maui) finds time slot with free compute nodes to run jobs

  • Batch job scripts

    • UNIX shell scripts with TORQUE (PBS) directives and executable commands.

    • See the programming documentation for script examples specific to system used.

    • Example job script, diffuse.pbs, for a serial job on the Glacier cluster, to run a program diffuse:

#!/bin/bash
#PBS -S /bin/bash

# Script for running serial program, diffuse, on glacier

cd $PBS_O_WORKDIR
echo "Current working directory is `pwd`"

echo "Starting run at: `date`"
./diffuse
echo "Job finished with exit code $? at: `date`"

  • Commands for submitting, monitoring and deleting jobs (qsub, showq, qdel)

  • See the Running Jobs pages or QuickStart guides for details.

  • Access to compute nodes for monitoring memory, processes, disk usage.

 

Post-Processing

 

Usage Guidelines

  • QuickStart guides for individual systems

  • Job limits

  • How much is reasonable?

  • Job priorities and the fairshare policy

  • Resource Allocation Committee

  • Accounting -  WestGrid portal.

 

Getting Help

Updated 2011-05-11.