This document has been retained for historical reasons. It contains a good overview of UVa's move from single, stand-alone Unix machines to a Cluster of computers managed by the Network Appliance.
The replacement of the general use Unix machines at UVa with the new blue.unix.virginia.edu system represents one of the few times when a staged and extended transition between ITC computer systems has been possible. The blue.unix system is a collection of IBM RS/6000 43P model 140 systems with PowerPC 604e processors (200 MHz). As a comparison, the machine faraday.clas had 112 MHz processors, so the new system will yield more processing power.
This collection of Unix machines is called a "cluster" as they all share the same file systems and accounts. The individual machines of this cluster are "nodes", and each has a distinct name within the cluster (eg node1.unix.virginia.edu, node2.unix.virginia.edu, etc).
The cluster will be more "fault-tolerant" than the previous systems. When users log into the cluster, the system will route them automatically to the computer in the cluster in the least use at the time. Redistributing the users in this fashion will alleviate the large "load" of users that previous systems experienced. In addition, if one node of the cluster fails, then users will be able to reconnect to blue.unix, and the system will assign them a new node on which they may work.
User home directories and mailboxes are on machines that the nodes of the cluster can access. This setup allows for users to log into any node of the cluster without noticing any difference from logging into a single machine. Only temporary files, which many editors and other Unix-based programs use, will be inaccessible if a particular node fails. The individual nodes keep temporary files on the local hard drive called /tmp, which the other nodes cannot access. Users should periodically save their work so that potential failures will not impede their work.
For information about cluster-specific issues, please look at the frequently asked questions (FAQ) document.
In addition, ITC has developed tools so that users can take advantage of some of the cluster's new features. The features include a tool for unsubscribing from mailing lists, retrieving deleted files, and gathering information about the system. Information about these tools is available inside of the Unix Cluster-Specific Tools document.
For many users, the only action required to complete migration will be to use blue.unix.virginia.edu when using telnet, FTP, rlogin, or an X-session to access the system. Users should reconfigure software that uses the retired machine names such as faraday.clas.virginia.edu, watt.seas.virginia.edu, poe.acc.virginia.edu, palladio.arch.virginia.edu, hobbes.itc.virginia.edu, etc. For a detailed description of e-mail changes, please read the Changes to E-mail document. This document provides information for Eudora and Netscape users on what changes they need to make to use the new system.
ITC has made arrangements so that mail sent to recently retired machine names will get delivered to your cluster account. In addition, if you have given out a web address that has one of these machine names or the machine name wsrv.clas.virginia.edu in it, those addresses will still work. ITC recommends, however, that you begin using and distributing web addresses with www.people.virginia.edu.
For more information about the names of the cluster machines, please look at the machine name summary table.
ITC has retired the machine names faraday.clas.virginia.edu, watt.seas.virginia.edu, and poe.acc.virginia.edu. Palladio.arch.virginia.edu and hobbes.itc.virginia.edu are scheduled to be retired on February 23, 1999.
As of April 22, 1998, new Unix accounts were created on the Cluster.