Chapter 1: Installing and Configuring Linux

Sections:

• Installing Linux
• Setting Up Accounts

We are setting up our grid so that it will appear like a grid client (NodeA) connecting to a grid that appears as a "cluster" of Torque (PBS) job managed machines represented by NodeB and a "cluster" of Sun Grid Engine (SGE) job managed machines represented by NodeC. The Globus toolkit will be configured to use either of these "clusters" to offload process intensive jobs to, to expedite completion of a task and show a direct benefit of a compute grid.

I'll make sure that our default installation instructions get us to these starting assumptions:

1. 3 machines running Fedora Core 4 i386 (FC4). For instructions on how to do this please see the section on Installing Linux
    
2. A working Yum configuration so that administrators are able to install, update, and remove packages using Yum.

   Verify that Yum is installed by running the command "yum --version". A version number should be returned. If Yum was not installed during the installation of FC4 instructions for retrieving and installing yum can be found on this web site: http://www.phy.duke.edu/~rgb/General/yum_article/yum_article/node10.html

   Log into each machine as root (recall the password is globusdemo) and run Yum with the command: "yum -y update".

   Note that that this first update will be a bit time consuming, but upon completion one should have a nice fresh install of the latest packages.
    
3. Root access to all the machines is necessary in order to complete some of the installation and for other administrative tasks
   If you followed the instructions in this tutorial for Installing Linux this should indeed be the case.  
4. Each of the machines should be running some type of time synchronization software so that the system times on the machines are accurate and consistent to within a few minutes
   
   If ntpd is to be used, the following /etc/ntp.conf file will probably work, however if you followed the instructions in this tutorial for Installing Linux there will already be a ntp.conf file that will more than likely contain this very information.
   
   [root@nodeD ~]# cat /etc/ntp.conf
   server 0.pool.ntp.org
   server 1.pool.ntp.org
   server 2.pool.ntp.org
   driftfile /var/lib/ntp/drift
   
   Also keep in mind that the machines need to have networking configured so that those time servers can be reached. You can test by doing this:
   
   [root@nodeD ~]# host 0.pool.ntp.org
   Which if all is working properly will yield a list of ip addresses similar to this:
   
   0.pool.ntp.org has address 80.35.31.228
   0.pool.ntp.org has address 80.163.145.206
   0.pool.ntp.org has address 142.58.206.202
   0.pool.ntp.org has address 199.103.21.233
   0.pool.ntp.org has address 202.234.64.222
   0.pool.ntp.org has address 213.210.55.91
   0.pool.ntp.org has address 213.239.193.168
   0.pool.ntp.org has address 216.27.160.99
   0.pool.ntp.org has address 220.164.192.140
   0.pool.ntp.org has address 62.152.126.5
   0.pool.ntp.org has address 62.193.225.80
   0.pool.ntp.org has address 64.172.230.138
    
5. Each of the machines should be configured so that all hostnames resolve correctly to IP addresses and vice versa. This will need to be accomplished by editing the file /etc/hosts (as root) on each (nodeA, nodeB, nodeC) of the machines.
   
   When /etc/host file is properly edited the cat command should yield output similar to this:
   
   [root@nodeD ~]# cat /etc/hosts
   # Do not remove the following line, or various programs
   # that require network functionality will fail. 127.0.0.1    localhost.localdomain    localhost
   192.168.31.39    nodeA.ps.univa.com    nodeA
   192.168.31.40    nodeB.ps.univa.com    nodeB
   192.168.31.41    nodeC.ps.univa.com    nodeC
   Note in particular the format of the entries above:
   
   IP address    FULLY QUALIFIED DOMAIN NAME    alias
   
   A common mistake is to put the alias before the FQDN. That is wrong. The FQDN should be first so that a lookup on the IP address returns the FQDN and not the short alias.
   

   For the purposes of this tutorial we will use the following names (short aliases) for the nodes:

   nodeA
   nodeB
   nodeC

6. All of the machines should be on the same subnet. If they
   are not on the same subnet the installation directions for
   Sun Grid Engine below will need to be slightly modified.
   If you followed the instructions in this tutorial for Installing Linux this should indeed be the case.
    
7. Make sure the proper user accounts are set up by following the instructions on this page.