	Department of Engineering

University of Cambridge > Engineering Department > computing help

Introduction to the Unix Teaching System

A brief overview of the machines, their file structure and the workings of the Unix operating system. If you're just interested knowing how to use the system, see the Introduction for New Users

Machines and Access

The CUED Teaching System is a collection of about 150 machines which are mostly based on versions of the Unix operating system. For flexibility, it's a distributed system, with much use made of client-server processes. That is, the machines ask each other for services (computational power from a compute-server, files from a file-server, etc). The main machines are as follows

file-serv is our main file server - the discs containing user files and big programs are physically attached to it. You can't log into file-serv - think of it as an external hard drive
tyrfing, scimitar ... tachi are our main linux servers (they're also known as tw100, etc). They have their own system discs, but get user files and big packages from file-serv. They can be logged-into from within the University
tw101, tw102 ... tw612, tw613 are PCs running linux. They can be used as computers in their own right, or can be used as terminals to connect to the linux servers.
gate is a similar machine to tyrfing etc, except that it's not timetabled for undergraduate use. It has no terminals attached to them, but it can be logged into from other machines - including machines outside Cambridge University.
IIA-projects and IIA-projects are the same kind of machines as tyrfing etc. They're not used for timetabled teaching. They're intended for use by IIA/IIB students and researchers.
twn01, twn02 ... twn10 (in the DPO) each have a big screen attached to them. They run Windows, but you can easily log into other machines from them.
spanner, rasp, wrench, cramp, awl, etc are our infrastructure machines. Under various aliases they deal with mail and providing various services to the system (WWW services, nameserver facilities, backups). You can't log into them. It's a good idea to use the advertised alias for the machine providing the service rather than the current machine name - the alias will work in the future even if the actual machine providing the service changes.

See

for further information.

The Unix File system

There are various flavours of Unix. Our machines all have a similar file structure.

Overall Structure

The Unix file system is one big tree.

/ - the root of the tree
/bin, /sbin - programs (binaries)
/lib - libraries
/etc - System management programs and configuration files.
/tmp - Temporary space that anyone can use, within reason
/dev - Files here are interfaces to hardware - printers, terminals etc.
/usr/local - where the local stuff goes. It has /bin, /lib and /etc sub-directories
/public - Space used for miscellaneous files
/export - Big packages and things that may be useful to CUED users in general rather than just Teaching System users.

User files

User files are deeper in the tree. Typing pwd in a terminal window prints out the current directory name. Users can use the "cd" command to wander around the tree (or click on icons if they're running a file manager), but their access and ability to read/write files is controlled by file permissions. Files can have Read, Write and Execute permissions which can be granted to just the user of the file, to a group of users or to all users. See our Unix groups and file permissions page for details.

Each user is only allowed a certain amount of space. See the Quota page for local details.

Processes

The term "Program" is used to mean the text the programmer write, the resulting file on disc and the result of "running" that program. The term "process" is used to mean only the last of these. Unix is a multi-user and multi-processing operating system (i.e. lots of people can simultaneously run lots of programs). One way to get a feel for process activity is to run "top", which displays a regularly updated 'league table' of running processes

Load averages: 0.31, 0.51, 0.58
174 processes: 172 sleeping, 1 running, 1 zombie
Cpu states:
 LOAD   USER   NICE    SYS   IDLE  BLOCK  SWAIT   INTR   SSYS
 0.31   2.0%   2.0%   2.0%  94.1%   0.0%   0.0%   0.0%   0.0%

Memory: 42392K (14156K) real, 64600K (22828K) virtual, 1096K free

 TTY   PID USERNAME  PRI NI   SIZE   RES  STATE   TIME %WCPU  %CPU COMMAND
   ?   811 root      154 20  2376K  1860K sleep   0:11  8.91  8.89 xfs
  q8  4999 tpl       189 24   432K   336K run     0:00 17.71  1.69 top
  p5  2616 tpl       154 24  6848K  3840K sleep   1:21  0.95  0.95 Mosaic
  q8  4043 tpl       154 28  2948K  1352K sleep   0:19  0.90  0.90 emacs

After a few lines of system statistics (note that 174 processes are running), the top processes are listed. Each process has a unique PID (process identity number) and is owned by someone. The SIZE column shows how big the process is. Unix works by giving each process a turn at using the CPU. If the piece of code required isn't in RAM, Unix finds space for it there. If a process isn't using the CPU it's usually in a sleeping state.

You can run programs (i.e. start processes) by clicking on their icons or typing their names on the command line. You exit from them by using their Exit button (or sometimes typing "exit"). See the Programs section of the Frequently Asked Questions page for further details.