Transcript Computer Systems` Architeture and Administration

DT228-3
Computer Systems’
Architecture and
Administration
Ronan Bradley
[email protected]
Aims of the CSAA Course…


To introduce the student to the general principles of
computer systems administration.
To equip the student with a sufficient understanding
of the architecture of computer systems to support
systems administration activities.
11/04/2016
2
…Aims of the CSAA Course



To provide a complete coverage of the run-control
system used for UNIX system start-up and
shutdown
To provide an introduction to user and group
management techniques on Windows and UNIX
systems.
To provide an introduction to systems issues relating
to the integration of Windows and UNIX systems.
11/04/2016
3
Learning Outcomes
On completion of the course the student will:





11/04/2016
Describe the fundamental tasks of a UNIX or a Windows
systems administrator.
Describe the user and group management mechanisms
and tools on Windows and UNIX systems
Describe the system start-up and shutdown processes on
a UNIX system.
Describe and modify the start-and shutdown order of
services on a UNIX system.
4
Learning Outcomes
On completion of the course the student will:






11/04/2016
Implement the installation, configuration and removal of
software systems on a UNIX platform
Describe the configuration operation of basic file sharing
mechanisms on the UNIX platform.
Describe the common mechanisms used for user and
group management on UNIX and Windows platforms.
Describe the configuration operation of user and group
sharing mechanisms on UNIX and Windows platforms
Design a systems integration plan to support file sharing,
identity sharing and other resource sharing between UNIX
and Windows systems.
5
Examination

Semester 1


11/04/2016
Written Examination
Continuous Assessment
60%
40%
6
Reading List…

Essential Reading

Marty Poniatowski, 2002, UNIX User’s Handbook,
2nd ed., Prentice Hall PTR


Steve Shah, Wale Soyinka. 2005, Linux
Administration: A Beginner’s Guide.
Osborne/McGraw-Hill

11/04/2016
ISBN: 0-13-065419-1
ISBN: 0-07-226259-1
7
CSAA Overview: System
Admininstration

User Management



UNIX/Linux System Start-up and Shutdown






Run-Levels
Changing the default start-up run-level
Changing the run-level once the system has started
Shutting down the System
Designing custom run-levels
Managing System Resources


on UNIX/Linux Systems
on Windows Systems
CPU, memory, disk I/O, disk space
Security configuration

11/04/2016
Pluggable Authentication Modules (PAM)
8
CSAA Overview: Networking

Domain Naming System






using bind on a UNIX/Linux system
FTP for file transfer between any systems
DHCP for dynamic allocation of IP addresses
NFS for transparent file sharing between UNIX/linux
systems
SaMBa for file and printer sharing between
UNIX/Linux and Windows systems
Network Information System (NIS) for sharing of user
names and passwords between systems
11/04/2016
9
File Sharing with
FTP and NFS

File Transfer Protocol


Basic transfer between
Network File System (nfs)

11/04/2016
Transparent file sharing between UNIX/Linux systems
10
What is SaMBa?

Client & Server for the Common Internet File
System (CIFS)


Microsoft's name for its SMB protocol implementation
Provides a single integrated work-group spanning
Windows and Linux machines


11/04/2016
Linux based resources look identical to Windows based
resources.
Using Universal Naming conventions (UNC): //csaa/distrib
11
What is Domain Name
System?

Internet system for associating human-friendly
names with machine-friendly IP addresses


Resolution of a given hostname to an IP address
Supports the global domain naming hierarchy (i.e.
comp.dit.ie.)



11/04/2016
Allows machines to be grouped logically, by domain
name
The full name is referred to as the (FQDN) Fully Qualified
Domain Name (e.g. lugh.student.comp.dit.ie or lugh)
Host names map to IP addresses in a one-to-many
relationship, each machine may have many IP addresses,
and each IP address may be associated with many
machines
12
What is DHCP?

Dynamic Host Configuration Protocol


11/04/2016
DHCP/BOOTP Clients request and are granted IP
addresses (and other information about themselves and
the network)
Best to have only one machine on an Ethernet segment
(VLAN) is designated a DHCP server
13
Some UNIX and Linux
history
Ken Thompson and Dennis Ritchie
A Brief History of UNIX



UNIX led to development of TCP/IP
Numerous vendors sell different UNIX varieties
Ken Thompson and Dennis Ritchie developed UNIX
at Bell Labs (part of AT&T)


System V
UNIX source code was cheaply available from AT&T

Quickly distributed to many organizations
A Brief History of UNIX
(continued)

Berkeley Software Distribution (BSD):
Berkeley versions of UNIX


Added TCP/IP network subsystem to UNIX
AT&T sold rights to UNIX

Now owned by two groups:


The SCO Group owns rights to UNIX source code
The Open Group owns UNIX trademark
Varieties of UNIX

All flavors of UNIX share many features:






Support multiple, simultaneously logged-on users
Coordinate multiple, simultaneously running tasks
Mount disk partitions on demand
Apply permissions for file and directory access
and modification
Uniform method of issuing data to or receiving
data from hardware devices, files, and running
programs
Start programs without interfering running
programs
Varieties of UNIX (continued)

All flavors of UNIX share the following
features (continued):



Hundreds of subsystems, including dozens of
programming languages
Source code portability
Window interfaces (e.g., X Windows)
Unix and Linux

Linux an be considered as a full-featured Unix clone



It cannot legally be called Unix as ‘Unix’ is a registered
trademark
Research and development of the kernel is
undertaken by thousands of volunteers over the
Internet

Robust version 1 was released in 1994

Combination of Kernel, tools and Daemons make up the full
operating system
Distributions of Linux from many sources

Kernel must be packaged with commands, daemons and
other software to create a useable operating system
DT249/3 - CSAA - L01
19
User Mode Linux
User Mode Linux (UML)

A virtualisation technology for Lunix



Vmware is a common virtualisation technology
Allows many ‘instances’ of an operating system to run on a
single server
Provides a sandbox environment


11/04/2016
“User-Mode Linux is a safe, secure way of running Linux
versions and Linux processes.”
“Run buggy software, experiment with new Linux kernels or
distributions, and poke around in the internals of Linux, all
without risking your main Linux setup.”
21
User Mode Linux (UML)



UML is a useful teaching tool, especially in courses
where students need a dedicated machine in order
to get the most out of it
UML is being used to teach OS development,
network administration, and more general system
administration.
These are all cases where having a virtual machine
to practice on is far more convenient for everyone
than using physical boxes.

11/04/2016
http://user-mode-linux.sourceforge.net/
22
Why UML?



The UML kernel runs in a virtual machine that can be configured
in ways that your physical machine can't. It can have more
memory, more devices, and, soon, more processors. So you can
do development and testing of hardware capabilities even when
you don't have the relevant hardware.
Poking around inside a running system
 Since you have a full OS running outside UML, it's pretty easy for
the terminally curious to find ways of looking inside this kernel
that are impossible for a native kernel.
As a secure sandbox or jail Processes inside the user-mode
kernel have no access to the hosting machine or to the outside
world that's not explicitly provided. So, a malicious application
running inside it can do no harm to anything that matters.
11/04/2016
23
Why UML?



Virtual networking UML virtual machines are networkable, to each
other, to the host, and to other physical machines. So, UML can be
used to set up a virtual network that allows setting up and testing of
experimental services. See the networking tutorial and virtual network
screenshot for more information
As a test environment Testing of some types of software requires
booting up a machine. With UML, this can be very easily automated.
There is a small test harness available , consisting of a small perl
module implementing a UML object which provides methods to boot a
virtual machine, log in to it, run commands, and shut it down.
Disaster recovery practice Do you know what you would do if your
machine suddenly didn't boot one day? With UML, you can set up
virtual disasters and practice recovering from them. This doesn't have
to be serious business - it can also be fun. Ever wonder what happens
when you run UML# rm -rf / but been afraid to try it? Now you can with
impunity.
11/04/2016
24
UML at the School of
Computing

The School of Computing provides one
system specifically for hosting UML sessions


11/04/2016
wicklow.cs.dit.ie
Upgraded in 2008
25
Wicklow UML

The binary executable on wicklow.cs.dit.ie which
loads the Linux kernel for the UML environment is


/usr/bin/umllinux
This is a script which invokes

11/04/2016
/bin/linux.nfs
26
Wicklow UML



In UML, the entire file system is contained in
a single file
By default the file used is root_fs in the
current directory
If no command-line option is specified, the
executable linux kernel will look for a file
called root_fs in the current directory
11/04/2016
27
Copy On Write



In UML, the entire file system is contained in a single
file
By default the file used is root_fs in the current
directory
If no command-line option is specified, the
executable linux kernel will look for a file called
root_fs in the /uml directory. The Copy On Write
mechanism (COW) allows many users to use the
same base root_fs yet still make their own changes
DT249/3 - CSAA - L03
28
UML File systems




This works by keeping a read-only copy of the root file system
shared by all users and a copy-on-write file for each individual user
The copy-on-write (cow) file contains only those changes which an
individual user has made to files
From the user’s point of view, their file system consists of two parts
 root_fs (read-only and shared)
 cow_fs (read/write and not shared)
To support the sharing of a single root file system among all UML
users on wicklow, the master root file system is stored in
 /opt/uml/rootfs.debian
DT249/3 - CSAA - L03
29
Personal cow file


To allow UML users on wicklow to store their
own modifications to the root_fs file, each
student has their own cow file
This cow file is located in their home directory
on wicklow and is called


/uml/root_fs_cow_<your username>
Since this file reflects any changes you have
made to the root_fs, if you delete this file all
changes will be lost
DT249/3 - CSAA - L03
30
umllinux script

To start UML on wicklow, users use the
umllinux script
!/bin/sh
/bin/linux.nfs ubd0=/uml/root_fs_cow_`whoami`,
/opt/uml/rootfs.debian eth0=mcast


This tells the executable linux kernel /bin/linux.nfs
to use the root_fs (/opt/uml/rootfs.debian) and the
cow file specified (/uml/root_fs_cow_<your
username>)
It also tells linux to provide one ethernet card, eth0
DT249/3 - CSAA - L03
31
Backup your cow file


If the user were to create a backup copy of
their cow file at the start of each wicklow
session, then they would have the ability to
roll-back any changes that were made during
a given session
In this way users could experiment with
settings and installing software in the
knowledge that undoing their changes would
be a simple matter
DT249/3 - CSAA - L03
32
host_fs

How can you transfer files between the host
and the UML session?


As a virtual network, there is no way to access
servers not on the network
UML supports the use of a specialised file
system driver host_fs

This allows us to mount the file system of the host
machine from within the UML instance
There are a number of ways of doing this
11/04/2016
33

host_fs

The first mechanism for mounting the host file
system is to use the mount command as
follows


mount none /mnt/host –t hostfs
This will cause the root of the host’s file
system to be mounted at the /mnt/host
directory (which must exist)
11/04/2016
34