UNIT I www.padeepz.

net

OS is a program that acts as an intermediary between a user of a computer and the computer
hardware

Operating system goals:

 Execute user programs and make solving user problems easier

t
Make the computer system convenient to use

ne
 Use the computer hardware in an efficient manner

COMPUTER SYSTEM ARCHITECTURE

o Computer system can be divided into four components:

.
o Hardware – provides basic computing resources

pz
 CPU, memory, I/O devices

o Operating system
ee  Controls and coordinates use of hardware among various applications and
users

o Application programs – define the ways in which the system resources are used to
solve the computing problems of the users
ad
 Word processors, compilers, web browsers, database systems, video
games

o Users

 People, machines, other computers

.p
w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
 Depends on the point of view


ad
Users want convenience, ease of use and good performance

o Don’t care about resource utilization

 But shared computer such as mainframe or minicomputer must keep all users happy
.p

 Users of dedicate systems such as workstations have dedicated resources but frequently
use shared resources from servers

 Handheld computers are resource poor, optimized for usability and battery life
w

 Some computers have little or no user interface, such as embedded computers in devices
and automobiles
w

 OS is a resource allocator

o Manages all resources

w

o Decides between conflicting requests for efficient and fair resource use

 OS is a control program

o Controls execution of programs to prevent errors and improper use of the

computer

 No universally accepted definition

www.padeepz.net
  ―Everything a vendor ships when you order an operating system‖ iswww.padeepz.net
a good
approximation

o But varies wildly

 ―The one program running at all times on the computer‖ is the kernel.

 Everything else is either

o a system program (ships with the operating system) , or

t
ne
o an application program.

COMPUTER STARTUP

 Bootstrap program is loaded at power-up or reboot

.
o Typically stored in ROM or EPROM, generally known as firmware

pz
o Initializes all aspects of system

o Loads operating system kernel and starts execution

COMPUTER-SYSTEM OPERATION
ee
 One or more CPUs, device controllers connect through common bus providing access to
shared memory
ad
.p
w
w
w

 I/O devices and the CPU can execute concurrently

 Each device controller is in charge of a particular device type

 Each device controller has a local buffer

 CPU moves data from/to main memory to/from local buffers

 I/O is from the device to local buffer of controller

www.padeepz.net
INTERRUPTS www.padeepz.net

 Device controller informs CPU that it has finished its operation by causing an
interrupt

 Interrupt transfers control to the interrupt service routine generally, through the
interrupt vector, which contains the addresses of all the service routines

 Interrupt architecture must save the address of the interrupted instruction

t
 A trap or exception is a software-generated interrupt caused either by an error or a

ne
user request

 An operating system is interrupt driven

INTERRUPT HANDLING

.


pz
The OS preserves the state of the CPU by storing registers and the program counter

 Determines which type of interrupt has occurred:

o polling
ee  The interrupt controller polls (send a signal out to) each device to
determine which one made the request

o vectored interrupt system


ad
Separate segments of code determine what action should be taken for each type of
interrupt

INTERRUPT TIMELINE
.p
w
w
w

www.padeepz.net
I/O STRUCTURE www.padeepz.net

 Synchronous (blocking) I/O

o Waiting for I/O to complete

o Easy to program, not always efficient

o Wait instruction idles the CPU until the next interrupt

t
o At most one I/O request is outstanding at a time

ne
 no simultaneous I/O processing

 Asynchronous (nonblocking) I/O

o After I/O starts, control returns to user program without waiting for I/O

.
completion

pz
o Harder to program, more efficient

o System call – request to the OS to allow user to wait for I/O completion (polling
periodically to check busy/done)
ee
o Device-status table contains entry for each I/O device indicating its type,
address, and state

STORAGE HIERARCHY
ad
The basic unit of computer storage is the bit. A bit can contain one of two values, 0 and 1. All
other storage in a computer is based on collections of bits. Given enough bits, it is amazing how
many things a computer can represent: numbers, letters, images, movies, sounds, documents, and
programs, to name a few. A byte is 8 bits, and on most computers it is the smallest convenient
chunk of storage. For example, most computers don’t have an instruction to move a bit but do
.p

have one to move a byte. A less common term is word, which is a given computer architecture’s
native unit of data. A word is made up of one or more bytes. For example, a computer that has
64-bit registers and 64-bit memory addressing typically has 64-bit (8-byte) words. A computer
w

executes many operations in its native word size rather than a byte at a time.

Computer storage, along with most computer throughput, is generally measured and manipulated
in bytes and collections of bytes.
w

A kilobyte, or KB, is 1,024 bytes

a megabyte, or MB, is 1,0242 bytes

w

a gigabyte, or GB, is 1,0243 bytes

a terabyte, or TB, is 1,0244 bytes

a petabyte, or PB, is 1,0245 bytes

www.padeepz.net
Computer manufacturers often round off these numbers and say that a megabytewww.padeepz.net
is 1 million
bytes and a gigabyte is 1 billion bytes. Networking measurements are an exception to this
general rule; they are given in bits (because networks move data a bit at a time).

STORAGE STRUCTURE

 Main memory – only large storage media that the CPU can access directly

o Random access

t
o Typically volatile

ne
 Secondary storage – extension of main memory that provides large nonvolatile storage
capacity

 Hard disks – rigid metal or glass platters covered with magnetic recording material

.
o Disk surface is logically divided into tracks, which are subdivided into sectors

pz
o The disk controller determines the logical interaction between the device and the
computer

 Solid-state disks – faster than hard disks, nonvolatile

ee
o Various technologies

o Becoming more popular

 Storage systems organized in hierarchy

ad
o Speed

o Cost (per byte of storage)

o Volatility
.p

 Device Driver for each device controller to manage I/O

o Provides uniform interface between controller and kernel

w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
ad
.p

CACHING
w

 Important principle

 Performed at many levels in a computer

w

o in hardware,
w

o operating system,

o software

 Information in use copied from slower to faster storage temporarily

o Efficiency

 Faster storage (cache) checked first to determine if information is there

www.padeepz.net
 o If it is, information used directly from the cache (fast) www.padeepz.net

o If not, data copied to cache and used there

 Cache smaller than storage being cached

o Cache management important design problem

o Cache size and replacement policy

t
DIRECT MEMORY ACCESS STRUCTURE

ne
 Typically used for I/O devices that generate data in blocks, or generate data fast

 Device controller transfers blocks of data from buffer storage directly to main
memory without CPU intervention

.
 Only one interrupt is generated per block, rather than the one interrupt per byte

pz
HOW A MODERN COMPUTER SYSTEM WORKS

ee
ad
.p
w
w
w

TYPES OF SYSTEMS

 Most systems use a single general-purpose processor

o Most systems have special-purpose processors as well

 Multiprocessors systems growing in use and importance

www.padeepz.net
 o Also known as parallel systems, tightly-coupled systems www.padeepz.net

o Advantages include:

1. Increased throughput

2. Economy of scale

3. Increased reliability – graceful degradation or fault tolerance

t
o Two types:

ne
 Asymmetric Multiprocessing – each processor is assigned a specific task

 Symmetric Multiprocessing – each processor performs all tasks

SYMMETRIC MULTIPROCESSING ARCHITECTURE

.
pz
ee
ad
.p

DUAL CORE DESIGN

 Multicore
w

o Several cores on a single chip

o On chip communication is faster than between-chip

w

o Less power used

w

www.padeepz.net
 www.padeepz.net

t
ne
CLUSTERED SYSTEMS

.
pz
ee
ad

 Like multiprocessor systems, but multiple systems working together

.p

 Provides a high-availability service which survives failures

 Asymmetric clustering has one machine in hot-standby mode

w

 Symmetric clustering has multiple nodes running applications, monitoring each other

 Some clusters are for high-performance computing (HPC)

w

 Applications must be written to use parallelization

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
MEMORY LAYOUT FOR MULTIPROGRAMMED SYSTEMS

ee
ad
.p
w
w
w

OPERATING SYSTEM OPERATIONS

 Interrupt driven (hardware and software)

o Hardware interrupt by one of the devices

www.padeepz.net
 o Software interrupt (exception or trap): www.padeepz.net

 Software error (e.g., division by zero)

 Request for operating system service

 Other process problems include infinite loop, processes modifying each

other or the operating system

t
. ne
pz
 Dual-mode operation allows OS to protect itself and other system components

 User mode and kernel mode

ee
 Mode bit provided by hardware

 Provides ability to distinguish when system is running user code or kernel code
ad
 Some instructions designated as privileged, only executable in kernel mode

 System call changes mode to kernel, return from call resets it to user

 Timer to prevent infinite loop / process hogging resources

.p

 Timer is set to interrupt the computer after some time period

 Keep a counter that is decremented by the physical clock

w

 Operating system set the counter (privileged instruction)

 When counter zero generate an interrupt

w

 Set up before scheduling process to

 regain control, or
w

 terminate program that exceeds allotted time

PROCESS MANAGEMENT

 A process is a program in execution. It is a unit of work within the system. Program is a

passive entity, process is an active entity.

 Process needs resources to accomplish its task

www.padeepz.net
 o CPU, memory, I/O, files www.padeepz.net

o Initialization data

 Typically system has many processes, some user, some operating system running
concurrently on one or more CPUs

o Concurrency by multiplexing the CPUs among the processes / threads

ACTIVITIES

t
ne
 Creating and deleting both user and system processes

 Suspending and resuming processes

 Providing mechanisms for process synchronization

.
 Providing mechanisms for process communication

pz
 Providing mechanisms for deadlock handling

MEMORY MANAGEMENT

 ee
To execute a program all (or part) of the instructions must be in memory

 All (or part) of the data that is needed by the program must be in memory.

 Memory management determines what is in memory and when

ad
o Optimizing CPU utilization and computer response to users

 Memory management activities

o Keeping track of which parts of memory are currently being used and by whom
.p

o Deciding which processes (or parts thereof) and data to move into and out of
memory

o Allocating and deallocating memory space as needed

w

STORAGE MANAGEMENT

 OS provides uniform, logical view of information storage

w

o Different devices, same view

o Abstracts physical properties to logical storage unit - file

w

o Each medium is controlled by device (i.e., disk drive, tape drive)

 Varying properties include access speed, capacity, data-transfer rate,

access method (sequential or random)

 File-System management

www.padeepz.net
 o Files usually organized into directories www.padeepz.net

o Access control on most systems to determine who can access what

o OS activities include

 Creating and deleting files and directories

 Primitives to manipulate files and directories

t
 Mapping files onto secondary storage

ne
 Backup files onto stable (non-volatile) storage media

MASS STORAGE MANAGEMENT

 Usually disks used to store

.


pz
data that does not fit in main memory, or

 data that must be kept for a ―long‖ period of time

 Proper management is of central importance


ee
Entire speed of computer operation hinges on disk subsystem and its algorithms

Disk is slow, its I/O is often a bottleneck

 OS activities
ad
 Free-space management

 Storage allocation

 Disk scheduling
.p

 Multitasking environments must be careful to use most recent value, no matter where it is
stored in the storage hierarchy
w

 Multiprocessor environment must provide cache coherency in hardware such that all
CPUs have the most recent value in their cache


w

Distributed environment situation even more complex

 Several copies of a datum can exist

w

I/O SUBSYSTEM

 One purpose of OS is to hide peculiarities of hardware devices from the user

 I/O subsystem responsible for

o Memory management of I/O including buffering (storing data temporarily while it

is being transferred),
www.padeepz.net
 www.padeepz.net
o Caching (storing parts of data in faster storage for performance),

o General device-driver interface

o Drivers for specific hardware devices

 Protection – any mechanism for controlling access of processes or users to resources

defined by the OS

 Security – defense of the system against internal and external attacks

t
ne
o Huge range, including denial-of-service, worms, viruses, identity theft, theft of
service

 Systems generally first distinguish among users, to determine who can do what

o Access control for users and groups

.
pz
COMPUTING ENVIRONMENTS

TRADITIONAL

 Stand-alone general purpose machines


ee
But blurred as most systems interconnect with others (i.e., the Internet)

Portals provide web access to internal systems

 Network computers (thin clients) are like Web terminals

ad
 Mobile computers interconnect via wireless networks

 Networking becoming ubiquitous – even home systems use firewalls to protect home
computers from Internet attacks
.p

MOBILE

 Handheld smartphones, tablets, etc

w

 What is the functional difference between them and a ―traditional‖ laptop?

 Extra feature – more OS features (GPS, gyroscope)

w

 Allows new types of apps like augmented reality

 Use IEEE 802.11 wireless, or cellular data networks for connectivity

w

 Leaders are Apple iOS and Google Android

 Distributed computing

 Collection of separate, possibly heterogeneous, systems networked together

www.padeepz.net
 Network is a communications path, TCP/IP most common www.padeepz.net

 Local Area Network (LAN)

 Wide Area Network (WAN)

 Metropolitan Area Network (MAN)

 Personal Area Network (PAN)

t
 Network Operating System provides features between systems across network

ne
 Communication scheme allows systems to exchange messages

 Illusion of a single system

.
Client-Server Computing

pz
 Dumb terminals supplanted by smart PCs

 Many systems now servers, responding to requests generated by clients

 Compute-server system provides an interface to client to request services (i.e., database)

ee
 File-server system provides interface for clients to store and retrieve files
ad
.p
w

P2P does not distinguish clients and servers

o Instead all nodes are considered peers

w

o May each act as client, server or both

o Node must join P2P network

w

 Registers its service with central lookup service on network, or

 Broadcast request for service and respond to requests for service via
discovery protocol

o Examples include Napster and Gnutella, Voice over IP (VoIP) such as Skype

www.padeepz.net
 www.padeepz.net

t
ne
Virtualization

 Host OS, natively compiled for CPU

.
 VMM - virtual machine manager

pz
 Creates and runs virtual machines

 VMM runs guest OSes, also natively compiled for CPU

 ee
Applications run within these guest OSes

 Example: Parallels for OS X running Win and/or Linux and their apps

 Some VMM’es run within a host OS

ad
 But, some act as a specialized OS

 Example. VMware ESX: installed on hardware, runs when hardware boots, provides
services to apps, runs guest OSes


.p

Vast and growing industry

 Use cases

 Developing apps for multiple different OSes on 1 PC

w

 Very important for cloud computing


w

Executing and managing compute environments in data centers

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz

ee
Operating systems made available in source-code format rather than just binary
closed-source

 Counter to the copy protection and Digital Rights Management (DRM) movement
ad
 Started by Free Software Foundation (FSF), which has ―copyleft‖ GNU Public
License (GPL)

 Examples include GNU/Linux and BSD UNIX (including core of Mac OS X), and
.p

many more

 Can use VMM like VMware Player (Free on Windows), Virtualbox (open source and
free on many platforms - http://www.virtualbox.com)
w

 Use to run guest operating systems for exploration

OPERATING SYSTEM SERVICES

w

 Operating System Services

 User Operating System Interface

w

 System Calls

 Types of System Calls

 System Programs

 Operating System Design and Implementation

www.padeepz.net
  Operating System Structure www.padeepz.net

 Operating System Debugging

 Operating System Generation

 System Boot

 Operating systems provide an environment for execution of programs and services

t
(helpful functions) to programs and users

ne
 User services:

o User interface

 No UI, Command-Line (CLI), Graphics User Interface (GUI), Batch

.
o Program execution - Loading a program into memory and running it, end

pz
execution, either normally or abnormally (indicating error)

o I/O operations - A running program may require I/O, which may involve a file
or an I/O device
ee
o User services (Cont.):

o File-system manipulation - Programs need to read and write files and

directories, create and delete them, search them, list file Information, permission
management.
ad
o Communications – Processes may exchange information, on the same computer
or between computers over a network

 Communications may be via shared memory or through message passing

(packets moved by the OS)
.p

o Error detection – OS needs to be constantly aware of possible errors

 May occur in the CPU and memory hardware, in I/O devices, in user
w

program

 For each type of error, OS should take the appropriate action to ensure
w

correct and consistent computing

 Debugging facilities can greatly enhance the user’s and programmer’s

abilities to efficiently use the system
w

System services:

o For ensuring the efficient operation of the system itself via resource sharing

o Resource allocation - When multiple users or multiple jobs running

concurrently, resources must be allocated to each of them

www.padeepz.net
  www.padeepz.net
Many types of resources - CPU cycles, main memory, file storage, I/O
devices.

o Accounting - To keep track of which users use how much and what kinds of
computer resources

o Protection and security - The owners of information stored in a multiuser or

networked computer system may want to control use of that information,
concurrent processes should not interfere with each other

t


ne
Protection involves ensuring that all access to system resources is
controlled

 Security of the system from outsiders requires user authentication,

extends to defending external I/O devices from invalid access attempts

.
pz
ee
ad


.p

 CLI or command interpreter allows direct command entry

o Sometimes implemented in kernel, sometimes by systems program

w

o Sometimes multiple flavors implemented – shells

o Primarily fetches a command from user and executes it

w

o Sometimes commands built-in, sometimes just names of programs

 If the latter, adding new features doesn’t require shell modification

w

 Systems calls: programming interface to the services provided by the OS

 Typically written in a high-level language (C or C++)

 Mostly accessed by programs via a high-level Application Programming Interface

(API) rather than direct system call use

 Three most common APIs are

www.padeepz.net
 o Win32 API for Windows, www.padeepz.net

o POSIX API for POSIX-based systems

 including virtually all versions of UNIX, Linux, and Mac OS X,

o Java API for the Java virtual machine (JVM)

t
. ne
pz
ee
ad

 Typically, a number associated with each system call
.p

o System-call interface maintains a table indexed according to these numbers

 The system call interface invokes the intended system call in OS kernel and returns status
of the system call and any return values
w

 The caller need know nothing about how the system call is implemented

o Just needs to obey API and understand what OS will do as a result call
w

o Most details of OS interface hidden from programmer by API

 Managed by run-time support library (set of functions built into libraries

w

included with compiler)

www.padeepz.net
 www.padeepz.net

t
. ne
pz

 Three general methods used to pass parameters to the OS in system calls
ee
o Simplest: in registers

 In some cases, may be more parameters than registers

o Parameters stored in a block, or table, in memory, and address of block passed as

ad
a parameter in a register

 This approach taken by Linux and Solaris

o Parameters placed, or pushed, onto the stack by the program and popped off the
.p

stack by the operating system

o Block and stack methods do not limit the number or length of parameters being
passed
w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
TYPES OF SYSTEM CALLS

 Process control
ee
o create process, terminate process

o end, abort

o load, execute
ad
o get process attributes, set process attributes

o wait for time

o wait event, signal event

.p

o allocate and free memory

o Dump memory if error

w

o Debugger for determining bugs, single step execution

o Locks for managing access to shared data between processes

w

 File management

o create file, delete file

w

o open, close file

o read, write, reposition

o get and set file attributes

 Device management

www.padeepz.net
 o request device, release device www.padeepz.net

o read, write, reposition

o get device attributes, set device attributes

o logically attach or detach devices

 Information maintenance

t
o get time or date, set time or date

ne
o get system data, set system data

o get and set process, file, or device attributes

 Communications

.
o create, delete communication connection

pz
o send, receive messages if message passing model to host name or process name

 From client to server

ee
o Shared-memory model create and gain access to memory regions

o transfer status information

o attach and detach remote devices

ad
 Protection

o Control access to resources

o Get and set permissions

.p

o Allow and deny user access

w
w
w

www.padeepz.net
 www.padeepz.net

t
ne.
pz
ee

ad
.p
w
w
w


www.padeepz.net
 SYSTEM PROGRAMS www.padeepz.net

 System programs provide a convenient environment for program development and

execution.

 Most users’ view of the operation system is defined by system programs, not the actual
system calls

 They can be divided into:

t
o File manipulation

ne
 rm, ls, cp, mv, etc in Unix

o Status information sometimes stored in a File modification

o Programming language support

.
pz
o Program loading and execution

o Communications

o Background services
ee
o Application programs

 Provide a convenient environment for program development and execution

ad
o Some of them are simply user interfaces to system calls; others are considerably
more complex

o File management - Create, delete, copy, rename, print, dump, list, and generally
.p

manipulate files and directories

 Status information

o Some ask the system for info - date, time, amount of available memory, disk
w

space, number of users

o Others provide detailed performance, logging, and debugging information

w

o Typically, these programs format and print the output to the terminal or other
output devices
w

o Some systems implement a registry - used to store and retrieve configuration

information

 File modification

o Text editors to create and modify files

www.padeepz.net
 www.padeepz.net
o Special commands to search contents of files or perform transformations of the
text

 Programming-language support - Compilers, assemblers, debuggers and interpreters

sometimes provided

 Program loading and execution- Absolute loaders, relocatable loaders, linkage editors,
and overlay-loaders, debugging systems for higher-level and machine language

t
 Communications - Provide the mechanism for creating virtual connections among

ne
processes, users, and computer systems

o Allow users to send messages to one another’s screens, browse web pages, send
electronic-mail messages, log in remotely, transfer files from one machine to
another

.
 Background Services

pz
o Launch at boot time

 Some for system startup, then terminate

ee  Some from system boot to shutdown

o Provide facilities like disk checking, process scheduling, error logging, printing

o Run in user context not kernel context

ad
o Known as services, subsystems, daemons

 Application programs

o Don’t pertain to system

.p

o Run by users

o Not typically considered part of OS

w

o Launched by command line, mouse click, finger poke

 Design and Implementation of OS not ―solvable‖, but some approaches have proven
successful
w

 Internal structure of different Operating Systems can vary widely


w

Start the design by defining goals and specifications

 Affected by choice of hardware, type of system

 User goals and System goals

o User goals – operating system should be convenient to use, easy to learn, reliable,
safe, and fast

www.padeepz.net
 www.padeepz.net
o System goals – operating system should be easy to design, implement, and
maintain, as well as flexible, reliable, error-free, and efficient

 Important principle to separate

 Policy: What will be done?

Mechanism: How to do it?

 Mechanisms determine how to do something, policies decide what will be done

t
 The separation of policy from mechanism is a very important principle, it allows

ne
maximum flexibility if policy decisions are to be changed later (example – timer)

 Specifying and designing an OS is highly creative task of software engineering

 Much variation

.
o Early OSes in assembly language

pz
o Then system programming languages like Algol, PL/1

o Now C, C++


ee
Actually usually a mix of languages

o Lowest levels in assembly

o Main body in C
ad
o Systems programs in C, C++, scripting languages like PERL, Python, shell scripts

 More high-level language easier to port to other hardware

o But slower
.p

 General-purpose OS is a very large program

o How to implement and structure it?

w

o Can apply many ideas from software engineering

 Software engineering - a separate area in CS

w

 Studies design, development, and maintenance of software

 A common approach is to partition OS into modules/components

w

o Each modules is responsible for one (or several) aspect of the desired
functionality

o Each module has carefully defined interfaces

o Advantages:

 Traditional advantages of modular programming:

www.padeepz.net
  www.padeepz.net
Simplifying development and maintenance of computer programs,
etc

 Modules can be developed independently from each other

o Disadvantages:

 Efficiency can decrease (vs monolithic approach)

t
OPERATING SYSTEM STRUCTURES

ne
 In general, various ways are used to structure OSes

 Many OS’es don’t have well-defined structures

o Not one pure model: Hybrid systems

.
o Combine multiple approaches to address performance, security, usability needs

pz
 Simple structure – MS-DOS

 More complex structure - UNIX

 ee
Layered OSes

 Microkernel OSes

SIMPLE STRUCTURE
ad
 MS-DOS was created to provide the most functionality in the least space

 Not divided into modules

 MS-DOS has some structure

.p

o But its interfaces and levels of functionality are not well separated

 No dual mode existed for Intel 8088

w

o MS-DOS was developed for 8088

o Direct access to hardware is allowed

w

 System crashes possible

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz

 Traditional UNIX has limited structuring


ee
UNIX consists of 2 separable parts:

o Systems programs

o Kernel
ad
 UNIX Kernel

o Consists of everything that is

 below the system-call interface and

.p

 above the physical hardware

o Kernel provides
w

 File system, CPU scheduling, memory management, and other operating-

system functions

 This is a lot of functionality for just 1 layer

w

 Rather monolithic


w

But fast -- due to lack of overhead in communication inside kernel

www.padeepz.net
 www.padeepz.net

t
. ne
pz

LAYERED APPROACH
ee
 One way to make OS modular – layered approach

 The OS is divided into a number of layers (levels)

 Each layer is built on top of lower layers

ad
 The bottom layer (layer 0), is the hardware

 The highest (layer N) is the user interface


.p

Layers are selected such that each uses functions (operations) and services of only lower-
level layers

 Advantages:
w

o simplicity of construction and debugging

 Disadvantages:
w

o can be hard to decide how to split functionality into layers

o less efficient due to high overhead

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz

OPERATING SYSTEM DEBUGGING
ee
 Debugging is finding and fixing errors, or bugs

 OS generate log files containing error information

 Failure of an application can generate core dump file capturing memory of the process
ad
 Operating system failure can generate crash dump file containing kernel memory

 Beyond crashes, performance tuning can optimize system performance

.p

o Sometimes using trace listings of activities, recorded for analysis

o Profiling is periodic sampling of instruction pointer to look for statistical trends

 Kernighan’s Law: ―Debugging is twice as hard as writing the code in the first place.
w

Therefore, if you write the code as cleverly as possible, you are, by definition, not smart
enough to debug it.‖
w

PERFORMANCE TUNNING

 Improve performance by removing bottlenecks

w

 OS must provide means of computing and displaying measures of system behavior

 For example, ―top‖ program or Windows Task Manager

OS SYSGEN

 Operating systems are designed to run on any of a class of machines

www.padeepz.net
 Тhe system must be configured for each specific computer site www.padeepz.net

 The process of configuration is known as system generation SYSGEN

o How to format partitions

o Which hardware is present

o Etc

t
 Used to build system-specific compiled kernel or system-tuned

ne
 Can general more efficient code than one general kernel

SYSTEM BOOT

 How OS is loaded?

.
pz
 When power is initialized on system, execution starts at a predefined memory location

o Firmware ROM is used to hold initial bootstrap program (=bootstrap loader)

 Bootstrap loader
ee
o small piece of code

o locates the kernel, loads it into memory, and starts it

 Sometimes 2-step process is used instead

ad
o Simple bootstrap loader in ROM loads a more complex boot program from (a
fixed location on) disk

o This more complex loader loads the kernel

.p
w
w
w

www.padeepz.net
 www.padeepz.net

UNIT II

PROCESSES - PROCESS CONCEPT

 An operating system executes a variety of programs:

o Batch system – ―jobs”

t
o Time-shared systems – ―user programs” or ―tasks”

ne
 We will use the terms job and process almost interchangeably

 Process – is a program in execution (informal definition)

 Program is passive entity stored on disk (executable file), process is active

.
pz
o Program becomes process when executable file loaded into memory

 Execution of program started via GUI, command line entry of its name, etc

 One program can be several processes


ee
o Consider multiple users executing the same program

In memory, a process consists of multiple parts:

o Program code, also called text section

ad
o Current activity including

 program counter

 processor registers
.p

o Stack containing temporary data

 Function parameters, return addresses, local variables

w

o Data section containing global variables

o Heap containing memory dynamically allocated during run time

w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
DIAGRAM OF PROCESS STATE
ad
.p
w
w

 As a process executes, it changes state

w

o new: The process is being created

o ready: The process is waiting to be assigned to a processor

o running: Instructions are being executed

o waiting: The process is waiting for some event to occur

www.padeepz.net
 o terminated: The process has finished execution www.padeepz.net

PROCESS CONTROL BLOCK(PCB)

Each process is represented in OS by PCB

 PCB - info associated with the process

 Also called task control block

t
 Process state – running, waiting, etc

ne
 Program counter – location of instruction to next execute

 CPU registers – contents of all process-centric registers

 CPU scheduling information- priorities, scheduling queue pointers

.
pz
 Memory-management information – memory allocated to the process

 Accounting information – CPU used, clock time elapsed since start, time limits

 I/O status information – I/O devices allocated to process, list of open files
ee
ad
.p
w
w
w

CPU SWITCH FROM PROCESS TO PROCESS

www.padeepz.net
 www.padeepz.net

t
. ne
pz
THREADS
ee
 So far, process has a single thread of execution

o One task at a time

ad
 Consider having multiple program counters per process

o Multiple locations can execute at once

.p

o Multiple tasks at a time

o Multiple threads of control -> threads

 PCB must be extended to handle threads:

w

o Store thread details

o Multiple program counters

w

PROCESS REPRESENTATION IN LINUX

w

Represented by the C structure task_struct

pid t_pid; /* process identifier */

long state; /* state of the process */
unsigned int time_slice /* scheduling information */
struct task_struct *parent; /* this process’s parent */
struct list_head children; /* this process’s children */
www.padeepz.net
struct files_struct *files; /* list of open files */ www.padeepz.net
struct mm_struct *mm; /* address space of this process */

t
. ne
pz
PROCESS SCHEDULING

 Goal of multiprogramming:

o Maximize CPU use


ee
Goal of time sharing:

o Quickly switch processes onto CPU for time sharing

 Process scheduler – needed to meet these goals

ad
o Selects 1 process to be executed next on CPU

o Among available processes

 Maintains scheduling queues of processes

.p

o Job queue – set of all processes in the system

o Ready queue – set of all processes residing in main memory, ready and waiting
w

to execute

o Device queues – set of processes waiting for an I/O device

w

o Processes migrate among the various queues

READY QUEUE AND VARIOUS QUEUES

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
ad
REPRESENTATION OF PROCESS SCHEDULING
.p
w
w
w

 Queuing diagram
www.padeepz.net
 o a common representation of process scheduling www.padeepz.net

o represents queues, resources, flows

SCHEDULERS

 Scheduler – component that decides how processes are selected from these queues for
scheduling purposes

Long-term scheduler (or job scheduler)

t
ne
 On this slide - ―LTS‖ (LTS is not a common notation)

 In a batch system, more processes are submitted then can be executed in memory

 They are spooled to disk

.
 LTS selects which processes should be brought into the ready queue

pz
 LTS is invoked infrequently

 (seconds, minutes)  (may be slow, hence can use advanced algorithms)


ee
LTS controls the degree of multiprogramming

The number of processes in memory

 Processes can be described as either:

ad
I/O-bound process

 Spends more time doing I/O than computations, many short CPU bursts

CPU-bound process
.p

 Spends more time doing computations; few very long CPU bursts

 LTS strives for good process mix

w

 Short-term scheduler (or CPU scheduler)

o Selects 1 process to be executed next

w

 Among ready-to-execute processes

 From the ready queue

w

 Allocates CPU to this process

o Sometimes the only scheduler in a system

o Short-term scheduler is invoked frequently

 (milliseconds)  (must be fast)

www.padeepz.net
  Hence cannot use costly selection logic www.padeepz.net

 Medium-term scheduler can be added if degree of multiple programming needs to

decrease

o Used by time-sharing OSes, etc

 Too many programs poor performance users quit

t
Key idea:

ne
o Reduce the degree of multiprogramming by swapping

o Swapping removes a process from memory, stores on disk, brings back in from
disk to continue execution

.
pz
ee
ad

CONTEXT SWITCH
.p

 Context of a process represented in its PCB

 Context switch
w

o When CPU switches to another process, the system must:

 save the state of the old process, and

w

 load the saved state for the new process

 Context-switch time is overhead

w

o The system does no useful work while switching

o The more complex the OS and the PCB 

 the longer the context switch

 This overhead time is dependent on hardware support

www.padeepz.net
 o Some hardware provides multiple sets of registers per CPUwww.padeepz.net

 multiple contexts are loaded at once

 switch requires only changing pointer to the right set

OPERATIONS ON PROCESSES

 System must provide mechanisms for:

t
 process creation,

ne
 process termination,

PROCESS CREATION

 A (parent) process can create several (children) processes

.


pz
Children can, in turn, create other processes

 Hence, a tree of processes forms

 Generally, process identified and managed via a process identifier (pid)


ee
Resource sharing options (of process creation)

Parent and children share all resources

 Children share subset of parent’s resources

ad
 One usage is to prevent system overload by too many child processes

 Parent and child share no resources

 Execution options
.p

 Parent and children execute concurrently

 Parent waits until children terminate

w

 Address space options

o Child duplicate of parent

w

o Child has a program loaded into it

 UNIX examples
w

o fork() system call creates new process

 Child is a copy of parent’s address space

 except fork() returns 0 to child and nonzero to parent

www.padeepz.net
 o exec() system call used after a fork() to replace the process’www.padeepz.net
memory space with a
new program

t
ne
PROCESS TERMINATION

 Process executes last statement and then asks the operating system to delete it using the
exit() system call.

.
pz
o Returns status data from child to parent (via wait())

o Process’ resources are deallocated by operating system

 Parent may terminate the execution of children processes using the abort() system call.
ee
Some reasons for doing so:

o Child has exceeded allocated resources

o Task assigned to child is no longer required

ad
o The parent is exiting and the operating systems does not allow a child to continue
if its parent terminates

o Some OSes don’t allow child to exists if its parent has terminated

o cascading termination - if a process terminates, then all its children,

.p

grandchildren, etc must also be terminated.

o The termination is initiated by the operating system

w

 The parent process may wait for termination of a child process by using the wait()system
call.
w

o The call returns status information and the pid of the terminated process

o pid = wait(&status);
w

 If no parent waiting (did not invoke wait()) process is a zombie

o All its resources are deallocated, but exit status is kept

 If parent terminated without invoking wait , process is an orphan

o UNIX: assigns init process as the parent

o Init calls wait periodically

www.padeepz.net
 CO OPERATING PROCESSES www.padeepz.net

 Processes within a system may be independent or cooperating

o When processes execute they produce some computational results

o Independent process cannot affect (or be affected) by such results of another

process

o Cooperating process can affect (or be affected) by such results of another process

t
ne
 Advantages of process cooperation

o Information sharing

o Computation speed-up

.
o Modularity

pz
o Convenience

INTERPROCESS COMMUNICATION

 For fast exchange of information, cooperating processes need some interprocess

ee
communication (IPC) mechanisms

 Two models of IPC

o Shared memory
ad
o Message passing
.p
w
w
w

 An area of memory is shared among the processes that wish to communicate

www.padeepz.net
  The communication is under the control of the users processes, www.padeepz.net
not the OS.

 Major issue is to provide mechanism that will allow the user processes to synchronize
their actions when they access shared memory.

 Producer-consumer problem – a common paradigm for cooperating processes

o Used to exemplify one common generic way/scenario of cooperation among

processes

t
o We will use it to exemplify IPC

ne
o Very important!

 Producer process

o produces some information

.
pz
o incrementally

 Consumer process

ee o consumes this information

o as it becomes available

 Challenge:

o Producer and consumer should run concurrently and efficiently

ad
o Producer and consumer must be synchronized

 Consumer cannot consume an item before it is produced

 Shared-memory solution to producer-consumer

.p

o Uses a buffer in shared memory to exchange information

 unbounded-buffer: assumes no practical limit on the buffer size

w

 bounded-buffer assumes a fixed buffer size

 Shared data
w

#define BUFFER_SIZE 10

typedef struct {
w

...

} item;

item buffer[BUFFER_SIZE];

int in = 0;
www.padeepz.net
int out = 0; www.padeepz.net

item next_produced;

while (true) {

next_produced = ProduceItem();

t
while (((in + 1) % BUFFER_SIZE) == out)

ne
; /* do nothing, no space in buffer */

//wait for consumer to get items and //free up some space

.
/* enough space in buffer */

pz
buffer[in] = next_produced; //put item into buffer

in = (in + 1) % BUFFER_SIZE;

}
ee
MESSAGE PASSING

 Mechanism for processes to communicate and to synchronize their actions

ad
 Message system – processes communicate with each other without resorting to shared
variables

 IPC facility provides two operations:

.p

o send(message)

o receive(message)

 The message size is either fixed or variable

w

o If processes P and Q wish to communicate, they need to:

w

o Establish a communication link between them

o Exchange messages via send/receive

w

o Implementation issues:

o How are links established?

o Can a link be associated with more than two processes?

o How many links can there be between every pair of communicating processes?

o What is the capacity (buffer size) of a link?

www.padeepz.net
 o Is the size of a message that the link can accommodate fixedwww.padeepz.net
or variable?

o Is a link unidirectional or bi-directional?

Logical implementation of communication link

o Direct or indirect

o Synchronous or asynchronous

t
o Automatic or explicit buffering

ne
DIRECT COMMUNICATION

 Processes must name each other explicitly:

o send (P, message) – send a message to process P

.
o receive(Q, message) – receive a message from process Q

pz
o Properties of a direct communication link

o Links are established automatically

ee
o A link is associated with exactly one pair of communicating processes

o Between each pair there exists exactly one link

o The link may be unidirectional, but is usually bi-directional

ad
INDIRECT COMMUNICATION

 Messages are directed and received from mailboxes (also referred to as ports)

 Each mailbox has a unique id

.p

 Processes can communicate only if they share a mailbox

 Properties of an indirect communication link

w

 Link established only if processes share a common mailbox

 A link may be associated with many processes

w

 Each pair of processes may share several communication links

 Link may be unidirectional or bi-directional

w

 Operations

o create a new mailbox (port)

o send and receive messages through mailbox

o destroy a mailbox
www.padeepz.net
 o Primitives are defined as: www.padeepz.net

send(A, message) – send a message to mailbox A

receive(A, message) – receive a message from mailbox A

SYNCHRONIZATION

 Message passing may be either

t
o Blocking, or

ne
o Non-blocking

 Blocking is considered synchronous

o Blocking send -- the sender is blocked until the message is received

.
o Blocking receive -- the receiver is blocked until a message is available

pz
 Non-blocking is considered asynchronous

o Non-blocking send -- the sender sends the message and continues

ee
o Non-blocking receive -- the receiver receives:

 A valid message, or

 Null message
ad
 Different combinations possible

o If both send and receive are blocking – called a rendezvous

 Producer-consumer is trivial via rendezvous

.p

message next_produced;
w

while (true) {
ProduceItem(&next_produced);
w

send(next_produced);
w

BUFFERING

 Queue of messages is attached to the link.

 Implemented in one of three ways

1. Zero capacity
www.padeepz.net
– no messages are queued on a link. www.padeepz.net

- Sender must wait for receiver (rendezvous)

2.Bounded capacity

– finite length of n messages

- Sender must wait if link full

3.Unbounded capacity

t
ne
– infinite length
- Sender never waits

THE CRITICAL SECTION PROBLEM

 Consider system of n processes {P0, P1, … ,Pn-1}

.
pz
 Each process has critical section segment of code

o Process may be changing common variables, updating table, writing file, etc

o When one process in critical section, no other is allowed to be in its critical


eesection

Critical section problem is to design protocol to solve this

 Each process
ad
o must ask permission to enter critical section in entry section,

o may follow critical section with exit section,

o then remainder section

.p
w
w
w

• Each process Pi has its own critical section

• Accesses to shared data only in CS

• Lock – will consider later

www.padeepz.net
 • In shared data, accessible by all www.padeepz.net

• Often 1 lock for all n processes

• Guards access to all critical sections

• Ensures mutual exclusivity

t
 Assume that each process executes at a nonzero speed

ne
 No assumption concerning relative speed of the n processes

1. Mutual Exclusion

l Only one process can be in the critical section at a time –

.
 otherwise what critical section?

pz
2. Progress

l Intuition: No process is forced to wait for an available resource –

l
ee  otherwise very wasteful.

Formal: If no process is executing in its critical section and there exist some
processes that wish to enter their critical section, then the selection of the
processes that will enter the critical section next cannot be postponed indefinitely
ad
3. Bounded Waiting

l Intuition: No process can wait forever for a resource –

 otherwise an easy solution: no one gets in

.p

l Formal: A bound must exist on the number of times that other processes are
allowed to enter their critical sections after a process has made a request to enter
its critical section and before that request is granted
w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
 Previous solutions are complicated and generally inaccessible to application
ad
programmers

 OS designers build software tools to solve critical section problem

 Simplest is mutex lock

.p

 Protect a critical section by:

o first acquire() a lock

w

o then release() the lock

o Boolean variable indicating if lock is available or not

w

 Calls to acquire() and release() must be atomic

o Usually implemented via hardware atomic instructions

w

 But this solution requires busy waiting

o This lock therefore called a spinlock

www.padeepz.net
 www.padeepz.net

t
. ne
pz
SEMAPHORE

n
ee
Synchronization tool that provides more sophisticated ways (than Mutex locks) for
process to synchronize their activities.
ad
n Semaphore S – integer variable

n Can only be accessed via two indivisible (atomic) operations

l wait() and signal()

.p

l Originally called P() and V()-- important to remember, used often

 P() (wait) – from Dutch proberen: ―to test‖

w

 V() (signal) – from Dutch verhogen: ―to increment‖

n Definition of the wait() operation

w

wait(S) {

while (S <= 0)
w

; // Notice, busy waits, spending CPU cycles, not (always) good

S--;

n Definition of the signal() operation

www.padeepz.net
signal(S) { www.padeepz.net

S++;

n Counting semaphore – integer value can range over an unrestricted domain

l Can control access to resource that has a finite number of instances

t
l Initialized to the number of resources available

ne
n Binary semaphore – integer value can range only between 0 and 1

l Same as a mutex lock

n Can solve various synchronization problems

.
n Consider P1 and P2 that require S1 to happen before S2

pz
Create a semaphore ―synch‖ initialized to 0

P1:

S1;
ee
signal(synch);

P2:
ad
wait(synch);

S2;

n Can implement a counting semaphore S via binary semaphores

.p

n Data Structures

int S = n; // semaphore count

w

mutex S_guard = 1; // initial value 1 (FREE)

mutex delay = 0; // for delaying other processes

w

 Operations wait() and signal() must be executed atomically

 Thus, the implementation becomes the critical section problem

w

o wait and signal code are placed in the critical section

 Hence, can now have busy waiting in critical section implementation

o If critical section rarely occupied

 Little busy waiting

www.padeepz.net
 o But, applications may spend lots of time in critical sections www.padeepz.net

 Hence, the busy-waiting approach is not a good solution

 A waiting queue is associated with each semaphore

o It stores the processes waiting on the semaphore

 typedef struct{

t
int value;

ne
struct process *list; // waiting queue

} semaphore;

 Two operations:

.
o block – place the process invoking the operation on the appropriate waiting queue

pz
o wakeup – remove one of processes in the waiting queue and place it in the ready
queue

wait(semaphore *S) {
ee
S->value--;

if (S->value < 0) {
add this process to S->list;
ad
block(); //suspends self, sleeps, avoids CPU cycles

}
.p

signal(semaphore *S) {

S->value++;
w

if (S->value <= 0) {
remove a process P from S->list;
w

wakeup(P);

}
w

 S->value can be negative

 |S->value| is then the number of processes waiting on the semaphore

 It is never negative in classical definition

www.padeepz.net
 DEADLOCK www.padeepz.net

 Synchronization problems

 Deadlock – two or more processes are waiting indefinitely for an event that can be
caused by only one of the waiting processes

 Let S and Q be two semaphores initialized to 1

t
P0 P1

ne
wait(S); wait(Q);

wait(Q); wait(S);

... ...

.
signal(S); signal(Q);

pz
signal(Q); signal(S);

 Starvation (= indefinite blocking)

o Related to deadlocks (but different)

ee
o Occurs when a process waits indefinitely in the semaphore queue

o For example, assume a LIFO semaphore queue


ad
Processed pushed first into it might not get a chance to execute

 Priority Inversion

o A situation when a higher-priority process needs to wait for a lower-priority

process that holds a lock
.p

Classical problems used to test newly-proposed synchronization schemes

o Bounded-Buffer Problem
w

o Readers and Writers Problem

o Dining-Philosophers Problem
w

MONITORS

monitor monitor-name
w

// shared variable declarations

function P1 (…) { … }

…
www.padeepz.net
 function Pn (…) {……} www.padeepz.net

initialization_code (…) { … }

o Recall that: an Abstract data type (ADT) -- encapsulates data

o internal variables only accessible by code within the procedure

t
o A monitor type – is an ADT that provides a convenient and effective mechanism

ne
for process synchronization

o For several concurrent processes

o Encapsulation

.
 Local variables accessed only via local functions

pz
 Local functions access only local vars and params

o Only one process at a time may be active within the monitor

o A lock guards shared data

ee
o Hence, the programmer does not need to code this constraint
ad
.p
w
w
w

www.padeepz.net
 www.padeepz.net
o Monitors not powerful enough to model some synchr. schemes

o Hence, additional synchr. mechanisms are added:

Provided by the condition construct

condition x, y;

o Two operations are allowed on a condition variable x:

t
o x.wait()

ne
 A process that invokes the operation is suspended (sleeps)

 until x.signal()is called

 Lets other processes enter the monitor

.


pz
releases the lock to shared data, atomically with sleep

o x.signal()

 Resumes one of processes (if any) that invoked x.wait()

ee  If no x.wait() was called, then x.signal()has no effect x
ad
.p
w
w
w

www.padeepz.net
Issues with monitors: assume www.padeepz.net

o Process P invokes x.signal(), and

o Process Q is suspended in x.wait()

Who proceeds next?

 Both Q and P cannot execute in parallel

t
 Because they are within a monitor

ne
Options include

o Signal and wait – P waits until Q either leaves the monitor or it waits for another
condition

.
o Signal and continue – Q waits until P either leaves the monitor or it waits for

pz
another condition

o Both have pros and cons – language implementer can decide

CPU SCHEDULING
ee
o Maximum CPU utilization obtained with multiprogramming

o waiting for I/O is wasteful

o 1 thread will utilize only 1 core

ad
o CPU–I/O Burst Cycle

o Process execution consists of:

 a cycle of CPU execution

.p

 and I/O wait

o CPU burst followed by I/O burst

w

o CPU burst distribution is of main concern

w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
ad
SCHEDULING LEVELS

High-Level Scheduling


.p

See Long-term scheduler or Job Scheduling from Chapter 3

 Selects jobs allowed to compete for CPU and other system resources.

Intermediate-Level Scheduling
w

 See Medium-Term Scheduling from Chapter 3


w

Selects which jobs to temporarily suspend/resume to smooth fluctuations in

system load.

Low-Level (CPU) Scheduling or Dispatching

w

 Selects the ready process that will be assigned the CPU.

 Ready Queue contains PCBs of processes.

 Short-term scheduler

 Selects 1 process from the ready queue

www.padeepz.net
  then allocates the CPU to it www.padeepz.net

 Queue may be ordered in various ways

 CPU scheduling decisions may take place when a process:

 Switches from running to waiting state

 Switches from running to ready state

t
 Switches from waiting to ready

ne
 Terminates

 Scheduling under 1 and 4 is called nonpreemptive (=cooperative)

.
All other scheduling is called preemptive

pz
 Process can be interrupted and must release the CPU

 Special care should be taken to prevent problems that can arise

ee  Access to shared data – race condition can happen, if not handled

Dispatcher

o a module that gives control of the CPU to the process selected by the
short-term scheduler; this involves:
ad
 switching context

 switching to user mode

 jumping to the proper location in the user program to restart

.p

that program

Dispatch latency

o Time it takes for the dispatcher to stop one process and start another
w

running

o This time should be as small as possible

w

 How do we decide which scheduling algorithm is good?

 Many criteria for judging this has been suggested

w

o Which characteristics considered can change significantly which algo

is considered the best

 CPU utilization – keep the CPU as busy as possible

 Throughput – # of processes that complete their execution per time unit

www.padeepz.net
  www.padeepz.net
Turnaround time – amount of time to execute a particular process

 Waiting time – amount of time a process has been waiting in the ready queue

 Response time – amount of time it takes to stat responding

o Used for interactive systems

o Time from when a request was submitted until the first response is

t
produced

ne
 Max CPU utilization

 Max throughput

 Min turnaround time

.
 Min waiting time

pz
 Min response time

Process Burst Time

ee P1

P2
24

P3 3
ad
Suppose that the processes arrive in the order: P1 , P2 , P3
The Gantt Chart for the schedule is:

P1 P2 P3
.p

0 24 27 30

Waiting time for P1 = 0; P2 = 24; P3 = 27

Average waiting time: (0 + 24 + 27)/3 = 17

w

FCFS
w

Suppose that the processes arrive in the order:

P2 , P3 , P1
w

 The Gantt chart for the schedule is:

 Waiting time for P1 = 6; P2 = 0; P3 = 3

 Average waiting time: (6 + 0 + 3)/3 = 3

 Much better than previous case

 Hence, average waiting time of FCFS not minimal

www.padeepz.net
  And it may vary substantially www.padeepz.net

 FCFS is nonpreemptive

 Not a good idea for timesharing systems

 FCFS suffers from the convoy effect,

P2 P3 P1

t
ne
0 3 6 30

 Convoy effect – when several short processes wait for long a process to get off the CPU

 Assume

.
o 1 long CPU-bound process

pz
o Many short I/O-bound processes

 Execution:
eeo The long one occupies CPU

 The short ones wait for it: no I/O is done at this stage

 No overlap of I/O with CPU utilizations

o The long one does its first I/O

ad
 Releases CPU

 Short ones are scheduled, but do I/O, release CPU quickly

.p

o The long one occupies CPU again, etc

 Hence low CPU and device utilization

w

SJF

 Associate with each process the length of its next CPU burst
w

o SJF uses these lengths to schedule the process with the shortest time

 Notice, the burst is used by SJF,

w

o not the process end-to-end running time

 implied by word ―job‖ in SJF

o Hence, it should be called ``Shorted-Next-CPU-Burst‖

o However, ―job‖ is used for historic reasons

 Two versions of SJF: preemptive and nonpreemptive

www.padeepz.net
 o Assume www.padeepz.net

 A new process Pnew arrives while the current one Pcur is still executing

 The burst of Pnew is less than what is left of Pcur

o Nonpreemptive SJF – will let Pcur finish

o Preemptive SJF – wil preempt Pcur and let Pnew execute

t
 This is also called shortest-remaining-time-first scheduling

ne
 Advantage:

o SJF is optimal in terms of the average waiting time

 Challenge of SJF:

.
o Hinges on knowing the length of the next CPU burst

pz
 But how can we know it?

 Solutions: ask user or estimate it

ee
o In a batch system and long-term scheduler

 Could ask the user for the job time limit

 The user is motivated to accurately estimate it

ad
 Lower value means faster response

 Too low a value will cause time-limit violation and job

rescheduling
.p

o In a short-term scheduling

 Use estimation

ProcessArrival Time Burst Time

w

P1 0.0 6
w

P2 2.0 8

P3 4.0 7
w

P4 5.0 3

SJF scheduling chart

P4 P1 P3 P2
0 3 9 16 24
www.padeepz.net
 Average waiting time = (3 + 16 + 9 + 0) / 4 = 7 www.padeepz.net

PRIORITY

 A priority number (integer) is associated with each process

 The CPU is allocated to the process with the highest priority (smallest integer  highest
priority)

o Preemptive

t
ne
o Nonpreemptive

o SJF is priority scheduling where priority is the inverse of predicted next CPU
burst time

 Problem  Starvation – low priority processes may never execute

.
pz
 Solution  Aging – as time progresses increase the priority of the process

ee
ad
.p
w
w

Average waiting time = 8.2 msec

ROUND ROBIN
w

 Each process gets a small unit of CPU time

o Time quantum q

o Usually 10-100 milliseconds

 After this time has elapsed:

o the process is preempted and

www.padeepz.net
 o added to the end of the ready queue www.padeepz.net

 The process might run for ≤ q time

o For example, when it does I/O

 If

o n processes in the ready queue, and

t
o the time quantum is q

ne
 then

o ―Each process gets 1/n of the CPU time‖

 Incorrect statement from the textbook

.
o in chunks of ≤ q time units at once

pz
o Each process waits ≤ (n-1)q time units

ee
ad
.p

Process Burst Time

w

P1 24

P2 3
w

P3 3
w

The Gantt chart is:

P1 P2 P3 P1 P1 P1 P1 P1
0 4 7 10 14 18 22 26 30

www.padeepz.net
DEADLOCK www.padeepz.net

Deadlock – two or more processes are waiting indefinitely for an event that can be
caused by only one of the waiting processes

Example: Let S and Q be two semaphores initialized to 1

P0 P1

wait(S); wait(Q);

t
ne
wait(Q); wait(S);

... ...

signal(S); signal(Q);

.
signal(Q); signal(S);

pz
 System consists of resources

 Resource types R1, R2, . . ., Rm

 CPU cycles, files, memory space, I/O devices, etc

ee
 Each resource type Ri has Wi instances.

o Type: CPU


ad
2 instances - CPU1, CPU2

o Type: Printer

 3 instances - printer1, printer2, printer3

.p

 Each process utilizes a resource as follows:

o Request resource


w

A process can request a resource of a given type

 E.g., ―I request any printer‖

w

 System will then assign a instance of that resource to the process

 E.g., some printer will be assigned to it

w

 If cannot be granted immediately, the process must wait until it can get it

o Use resource

 Operate on the resource, e.g. print on the printer

o Release resource

 Mutexes and Semaphores

www.padeepz.net
 o Special case: www.padeepz.net

 Each mutex or semaphor is treated as a separate resource type

 Because a process would want to get not just ―any‖ lock among a group of
locks, but a specific lock that guards a specific shared data type

 e.g., lock that guards a specific queue

t
Mutual exclusion: only one process at a time can use a resource

ne
 Hold and wait: a process holding at least one resource is waiting to acquire additional
resources held by other processes

 No preemption: a resource can be released only voluntarily by the process holding it,
after that process has completed its task

.
 Circular wait: there exists a set {P0, P1, …, Pn} of waiting processes such that

pz
o P0 is waiting for a resource that is held by P1, and so on:

o P0  P1  P2  …  Pn–1 Pn  P0


ee
Notice: ―Circular wait‖ implies ―Hold and Wait‖

o Why then not test for only the ―Circular wait‖?

o Because, computationally, ―Hold and wait‖ can be tested much more efficiently
than ―Circular wait‖
ad
o Some algorithms we consider only need to check H&W

 Deadlocks can occur via system calls, locking, etc.

.p

 Trivial example

o Mutexes A, B – unlocked initially

o Process P1: wait(A); wait(B);

w

o Process P2: wait(B); wait(A);

w
w

www.padeepz.net
 UNIT III www.padeepz.net

MAIN MEMORY

 Program must be brought (from disk) into memory and placed within a process for it
to be run

 Main memory and registers are only storage CPU can access directly

t
Memory unit only sees a stream of addresses + read requests, or address + data and
write requests

ne
 Register access in one CPU clock (or less)

 Main memory can take many cycles, causing a stall

.
Cache sits between main memory and CPU registers

pz
 Protection of memory required to ensure correct operation

 A pair of base and limit registers define the logical address space

 CPU must check every memory access generated in user mode to be sure it is
ee
between base and limit for that user
ad
.p
w
w
w

www.padeepz.net
 www.padeepz.net

HARDWARE PROTECTION

t
. ne
pz
ee
ADDRESS BINDING

 Programs on disk, ready to be brought into memory to execute form an input queue

o Without support, must be loaded into address 0000

ad
 Inconvenient to have first user process physical address always at 0000

o How can it not be?

 Further, addresses represented in different ways at different stages of a program’s life

.p

o Source code addresses usually symbolic

o Compiled code addresses bind to relocatable addresses

w

 i.e. ―14 bytes from beginning of this module‖

o Linker or loader will bind relocatable addresses to absolute addresses

w

 i.e. 74014

o Each binding maps one address space to another

w

 Address binding of instructions and data to memory addresses can happen at three
different stages

o Compile time: If memory location known a priori, absolute code can be

generated; must recompile code if starting location changes

www.padeepz.net
 www.padeepz.net
o Load time: Must generate relocatable code if memory location is not known at
compile time

o Execution time: Binding delayed until run time if the process can be moved
during its execution from one memory segment to another

 Need hardware support for address maps (e.g., base and limit registers)

MULTISTEP PROCESSING

t
. ne
pz
ee
ad
.p
w

LOGICAL VS PHYSICAL ADDRESS

 The concept of a logical address space that is bound to a separate physical address space
w

is central to proper memory management

o Logical address – generated by the CPU; also referred to as virtual address

w

o Physical address – address seen by the memory unit

 Logical and physical addresses are the same in compile-time and load-time address-
binding schemes; logical (virtual) and physical addresses differ in execution-time
address-binding scheme

 Logical address space is the set of all logical addresses generated by a program

www.padeepz.net
  www.padeepz.net
Physical address space is the set of all physical addresses generated by a program

MMU

 Hardware device that at run time maps virtual to physical address

 Many methods possible, covered in the rest of this chapter

 To start, consider simple scheme where the value in the relocation register is added to

t
every address generated by a user process at the time it is sent to memory

ne
o Base register now called relocation register

o MS-DOS on Intel 80x86 used 4 relocation registers

 The user program deals with logical addresses; it never sees the real physical addresses

.
o Execution-time binding occurs when reference is made to location in memory

pz
o Logical address bound to physical addresses

DYNAMIC ALLOCATION

 ee
Routine is not loaded until it is called

 Better memory-space utilization; unused routine is never loaded

 All routines kept on disk in relocatable load format

ad
 Useful when large amounts of code are needed to handle infrequently occurring cases

 No special support from the operating system is required

o Implemented through program design

.p

o OS can help by providing libraries to implement dynamic loading

w
w
w

DYNAMIC LINKING
www.padeepz.net
  www.padeepz.net
Static linking – system libraries and program code combined by the loader into the
binary program image

 Dynamic linking –linking postponed until execution time

 Small piece of code, stub, used to locate the appropriate memory-resident library routine

 Stub replaces itself with the address of the routine, and executes the routine

t
Operating system checks if routine is in processes’ memory address

ne
o If not in address space, add to address space

 Dynamic linking is particularly useful for libraries

 System also known as shared libraries

.
 Consider applicability to patching system libraries

pz
o Versioning may be needed

SWAPPING


ee
A process can be swapped temporarily out of memory to a backing store, and then
brought back into memory for continued execution

o Total physical memory space of processes can exceed physical memory

 Backing store – fast disk large enough to accommodate copies of all memory images for
ad
all users; must provide direct access to these memory images

 Roll out, roll in – swapping variant used for priority-based scheduling algorithms; lower-
priority process is swapped out so higher-priority process can be loaded and executed
.p

 Major part of swap time is transfer time; total transfer time is directly proportional to the
amount of memory swapped

 System maintains a ready queue of ready-to-run processes which have memory images
w

on disk

 Does the swapped out process need to swap back in to same physical addresses?
w

 Depends on address binding method

o Plus consider pending I/O to / from process memory space

w

 Modified versions of swapping are found on many systems (i.e., UNIX, Linux, and
Windows)

o Swapping normally disabled

o Started if more than threshold amount of memory allocated

o Disabled again once memory demand reduced below threshold

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
CONTIGUOUS ALLOCATION

 Main memory must support both OS and user processes

ad
 Limited resource, must allocate efficiently

 Contiguous allocation is one early method

 Main memory usually into two partitions:

.p

o Resident operating system, usually held in low memory with interrupt vector

o User processes then held in high memory

w

o Each process contained in single contiguous section of memory

 Relocation registers used to protect user processes from each other, and from changing
w

operating-system code and data

o Base register contains value of smallest physical address

w

o Limit register contains range of logical addresses – each logical address must be
less than the limit register

o MMU maps logical address dynamically

o Can then allow actions such as kernel code being transient and kernel changing
size

www.padeepz.net
 www.padeepz.net

t
. ne
pz
FRAGMENTATION

 External Fragmentation – total memory space exists to satisfy a request, but it is not
contiguous


ee
Internal Fragmentation – allocated memory may be slightly larger than requested
memory; this size difference is memory internal to a partition, but not being used

 First fit analysis reveals that given N blocks allocated, 0.5 N blocks lost to fragmentation

o 1/3 may be unusable -> 50-percent rule

ad
 Reduce external fragmentation by compaction

o Shuffle memory contents to place all free memory together in one large block
.p

o Compaction is possible only if relocation is dynamic, and is done at execution

time

o I/O problem
w

 Latch job in memory while it is involved in I/O

 Do I/O only into OS buffers

w

 Now consider that backing store has same fragmentation problems

SEGMENTATION
w

Memory-management scheme that supports user view of memory

A program is a collection of segments

A segment is a logical unit such as:

main program

www.padeepz.net
 procedure www.padeepz.net

function

method

object

local variables, global variables

t
common block

ne
stack

symbol table

arrays

.
pz
ee
ad
.p


w

Logical address consists of a two tuple:

 <segment-number, offset>,
w

 Segment table – maps two-dimensional physical addresses; each table entry has:

o base – contains the starting physical address where the segments reside in
memory
w

o limit – specifies the length of the segment

o Segment-table base register (STBR) points to the segment table’s location in

memory

 Segment-table length register (STLR) indicates number of segments used by a

program;
www.padeepz.net
  segment number s is legal if s < STLR www.padeepz.net

 Protection

o With each entry in segment table associate:

 validation bit = 0  illegal segment

 read/write/execute privileges

t
 Protection bits associated with segments; code sharing occurs at segment level

ne
 Since segments vary in length, memory allocation is a dynamic storage-allocation
problem

 A segmentation example is shown in the following diagram

.
pz
ee
ad
.p
w
w

PAGING

 Physical address space of a process can be noncontiguous; process is allocated physical

memory whenever the latter is available
w

o Avoids external fragmentation

o Avoids problem of varying sized memory chunks

 Divide physical memory into fixed-sized blocks called frames

o Size is power of 2, between 512 bytes and 16 Mbytes

www.padeepz.net
  Divide logical memory into blocks of same size called pages www.padeepz.net

 Keep track of all free frames

 To run a program of size N pages, need to find N free frames and load program

 Set up a page table to translate logical to physical addresses

 Backing store likewise split into pages

t
 Still have Internal fragmentation

ne
 Address generated by CPU is divided into:

o Page number (p) – used as an index into a page table which contains base
address of each page in physical memory

.
o Page offset (d) – combined with base address to define the physical memory

pz
address that is sent to the memory unit

o For given logical address space 2m and page size 2 n

ee
page number page offset
p d
ad
m -n n
PAGING STRUCTURE
.p
w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
 Calculating internal fragmentation
ad
o Page size = 2,048 bytes

o Process size = 72,766 bytes

o 35 pages + 1,086 bytes

.p

o Internal fragmentation of 2,048 - 1,086 = 962 bytes

o Worst case fragmentation = 1 frame – 1 byte

w

o On average fragmentation = 1 / 2 frame size

o So small frame sizes desirable?

w

o But each page table entry takes memory to track

o Page sizes growing over time

w

 Solaris supports two page sizes – 8 KB and 4 MB

 Process view and physical memory now very different

 By implementation process can only access its own memory

www.padeepz.net
 www.padeepz.net

t
. ne
pz

ee
Page table is kept in main memory

 Page-table base register (PTBR) points to the page table

ad
 Page-table length register (PTLR) indicates size of the page table

 In this scheme every data/instruction access requires two memory accesses

o One for the page table and one for the data / instruction
.p

 The two memory access problem can be solved by the use of a special fast-lookup
hardware cache called associative memory or translation look-aside buffers (TLBs)
w

 Some TLBs store address-space identifiers (ASIDs) in each TLB entry – uniquely
identifies each process to provide address-space protection for that process
w

o Otherwise need to flush at every context switch

 TLBs typically small (64 to 1,024 entries)

w

 On a TLB miss, value is loaded into the TLB for faster access next time

o Replacement policies must be considered

o Some entries can be wired down for permanent fast access

ASSOCIATIVE MEMORY

 Associative memory – parallel search

www.padeepz.net
  Address translation (p, d) www.padeepz.net

o If p is in associative register, get frame # out

o Otherwise get frame # from page table in memory

Page # Frame #

t
. ne
pz
PAGING WITH TLB

ee
ad
.p
w
w
w

 Associative Lookup =  time unit

o Can be < 10% of memory access time

 Hit ratio = 

o Hit ratio – percentage of times that a page number is found in the associative
registers; ratio related to number of associative registers
www.padeepz.net
  www.padeepz.net
Consider  = 80%,  = 20ns for TLB search, 100ns for memory access

 Effective Access Time (EAT)

 EAT = (1 + )  + (2 + )(1 – )

 =2+–

 Consider  = 80%,  = 20ns for TLB search, 100ns for memory access

t
o EAT = 0.80 x 100 + 0.20 x 200 = 120ns

ne
 Consider more realistic hit ratio ->  = 99%,  = 20ns for TLB search, 100ns for
memory access

o EAT = 0.99 x 100 + 0.01 x 200 = 101ns

.
MEMORY PROTECTION

pz
 Memory protection implemented by associating protection bit with each frame to indicate
if read-only or read-write access is allowed

o Can also add more bits to indicate page execute-only, and so on

ee
 Valid-invalid bit attached to each entry in the page table:

o ―valid‖ indicates that the associated page is in the process’ logical address space,
and is thus a legal page
ad
o ―invalid‖ indicates that the page is not in the process’ logical address space

o Or use page-table length register (PTLR)

 Any violations result in a trap to the kernel

.p

VALID OR INVALID BIT IN PAGE TABLE

w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
STRUCTURE OF THE PAGE TABLE
ad
 Memory structures for paging can get huge using straight-forward methods

o Consider a 32-bit logical address space as on modern computers

o Page size of 4 KB (212)

.p

o Page table would have 1 million entries (232 / 212)

o If each entry is 4 bytes -> 4 MB of physical address space / memory for page
table alone
w

 That amount of memory used to cost a lot

 Don’t want to allocate that contiguously in main memory

w

 Hierarchical Paging


w

Hashed Page Tables

 Inverted Page Tables

TWO LEVEL PAGE SCHEME

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
 A logical address (on 32-bit machine with 1K page size) is divided into:
ad
o a page number consisting of 22 bits

o a page offset consisting of 10 bits

o Since the page table is paged, the page number is further divided into:
.p

o a 12-bit page number

o a 10-bit page offset

w

o Thus, a logical address is as follows:

w
w

 where p1 is an index into the outer page table, and p2 is the displacement within the page
of the inner page table

 Known as forward-mapped page table

www.padeepz.net
 ADDRESS TRANSLATION SCHEME www.padeepz.net

t
. ne
pz
64 BIT ARCHITECTURE


ee
Even two-level paging scheme not sufficient

If page size is 4 KB (212)

o Then page table has 252 entries

ad
o If two level scheme, inner page tables could be 210 4-byte entries

o Address would look like

o Outer page table has 242 entries or 244 bytes

.p

o One solution is to add a 2nd outer page table

o But in the following example the 2nd outer page table is still 234 bytes in size
w

 And possibly 4 memory access to get to one physical memory location

w
w

www.padeepz.net
 www.padeepz.net

t
ne
INTEL 32 BIT ARCHITECTURE

.
pz
ee
ad
.p
w
w

VIRTUAL MEMORY
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
VIRTUAL ADDRESS SPACE

 Usually design logical address space for stack to start at Max logical address and grow
ad
―down‖ while heap grows ―up‖

o Maximizes address space use

o Unused address space between the two is hole

.p

 No physical memory needed until heap or stack grows to a given new

page

 Enables sparse address spaces with holes left for growth, dynamically linked libraries,
w

etc

 System libraries shared via mapping into virtual address space

w

 Shared memory by mapping pages read-write into virtual address space

 Pages can be shared during fork(), speeding process creation

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
SHARED LIBRARY USING VIRTUAL MEMORY
ad
.p
w
w
w

DEMAND PAGING
www.padeepz.net
 Could bring entire process into memory at load time www.padeepz.net

 Or bring a page into memory only when it is needed

o Less I/O needed, no unnecessary I/O

o Less memory needed

o Faster response

t
o More users

ne
 Similar to paging system with swapping (diagram on right)

 Page is needed  reference to it

o invalid reference  abort

.
pz
o not-in-memory  bring to memory

 Lazy swapper – never swaps a page into memory unless page will be needed

o Swapper that deals with pages is a pager

ee
ad
.p
w
w
w

 With swapping, pager guesses which pages will be used before swapping out again

 Instead, pager brings in only those pages into memory

 How to determine that set of pages?

o Need new MMU functionality to implement demand paging

www.padeepz.net
  If pages needed are already memory resident www.padeepz.net

o No difference from non demand-paging

 If page needed and not memory resident

o Need to detect and load the page into memory from storage

 Without changing program behavior

t
 Without programmer needing to change code

ne
VALID OR INVALID BIT

 With each page table entry a valid–invalid bit is associated

(v  in-memory – memory resident, i  not-in-memory)

.
 Initially valid–invalid bit is set to i on all entries

pz
 Example of a page table snapshot:

ee
ad
.p
w

During MMU address translation, if valid–invalid bit in page table entry is i  page fault
w

If there is a reference to a page, first reference to that page will trap to operating system:
w

PAGE FAULT

1. Operating system looks at another table to decide:

l Invalid reference  abort

l Just not in memory

2. Find free frame

www.padeepz.net
 3. Swap page into frame via scheduled disk operation www.padeepz.net

4. Reset tables to indicate page now in memory

Set validation bit = v

5. Restart the instruction that caused the page fault

6. Extreme case – start process with no pages in memory

l OS sets instruction pointer to first instruction of process, non-memory-resident ->

t
page fault

ne
l And for every other process pages on first access

l Pure demand paging

7. Actually, a given instruction could access multiple pages -> multiple page faults

.
pz
l Consider fetch and decode of instruction which adds 2 numbers from memory and
stores result back to memory

l Pain decreased because of locality of reference

ee
8. Hardware support needed for demand paging

l Page table with valid / invalid bit

l Secondary memory (swap device with swap space)

ad
l Instruction restart

INSTRUCTION RESTART

Consider an instruction that could access several different locations

.p

block move
auto increment/decrement location

Restart the whole operation?

w

 What if source and destination overlap?

w
w

www.padeepz.net
 www.padeepz.net

PERFORMANCE

n Stages in Demand Paging (worse case)

1. Trap to the operating system

t
2. Save the user registers and process state

ne
3. Determine that the interrupt was a page fault

4. Check that the page reference was legal and determine the location of the page on the
disk

.
5. Issue a read from the disk to a free frame:

pz
1. Wait in a queue for this device until the read request is serviced

2. Wait for the device seek and/or latency time

3. Begin the transfer of the page to a free frame

ee
6. While waiting, allocate the CPU to some other user

7. Receive an interrupt from the disk I/O subsystem (I/O completed)

8. Save the registers and process state for the other user
ad
9. Determine that the interrupt was from the disk

10. Correct the page table and other tables to show page is now in memory
.p

11. Wait for the CPU to be allocated to this process again

12. Restore the user registers, process state, and new page table, and then resume the
interrupted instruction
w

PAGE REPLACEMENT

 Prevent over-allocation of memory by modifying page-fault service routine to

w

include page replacement

 Use modify (dirty) bit to reduce overhead of page transfers – only modified pages
are written to disk
w

 Page replacement completes separation between logical memory and physical

memory – large virtual memory can be provided on a smaller physical memory

NEED

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
BASIC PAGE REPLACEMENT

1. Find the location of the desired page on disk

ad
2. Find a free frame:
- If there is a free frame, use it
- If there is no free frame, use a page replacement algorithm to select a victim frame
- Write victim frame to disk if dirty
.p

3. Bring the desired page into the (newly) free frame; update the page and frame tables

4. Continue the process by restarting the instruction that caused the trap
w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
ALGORITHMS
ad
 Frame-allocation algorithm determines

o How many frames to give each process

o Which frames to replace

.p

 Page-replacement algorithm

o Want lowest page-fault rate on both first access and re-access


w

Evaluate algorithm by running it on a particular string of memory references (reference

string) and computing the number of page faults on that string

o String is just page numbers, not full addresses

w

o Repeated access to the same page does not cause a page fault

o Results depend on number of frames available

w

 In all our examples, the reference string of referenced page numbers is

 7,0,1,2,0,3,0,4,2,3,0,3,0,3,2,1,2,0,1,7,0,1

FIFO

www.padeepz.net
 www.padeepz.net

t
ne
OPTIMAL ALGORITHM

.
pz
LRU
ee
ad
.p

 Keep a pool of free frames, always

w

o Then frame available when needed, not found at fault time

o Read page into free frame and select victim to evict and add to free pool
w

o When convenient, evict victim


w

Possibly, keep list of modified pages

o When backing store otherwise idle, write pages there and set to non-dirty

 Possibly, keep free frame contents intact and note what is in them

o If referenced again before reused, no need to load contents again from disk

o Generally useful to reduce penalty if wrong victim frame selected

www.padeepz.net
  Each process needs minimum number of frames www.padeepz.net

 Example: IBM 370 – 6 pages to handle SS MOVE instruction:

o instruction is 6 bytes, might span 2 pages

o 2 pages to handle from

o 2 pages to handle to

t
 Maximum of course is total frames in the system

ne
 Two major allocation schemes

o fixed allocation

o priority allocation

.


pz
Many variations

THRASHING

 If a process does not have ―enough‖ pages, the page-fault rate is very high
ee
o Page fault to get page

o Replace existing frame

o But quickly need replaced frame back

ad
o This leads to:

 Low CPU utilization

 Operating system thinking that it needs to increase the degree of

.p

multiprogramming

 Another process added to the system

 Thrashing  a process is busy swapping pages in and out

w

 Why does demand paging work?

Locality model
w

 Process migrates from one locality to another


w

Localities may overlap

 Why does thrashing occur?

 size of locality > total memory size

 Limit effects by using local or priority page replacement

www.padeepz.net
 www.padeepz.net

t
ne.
pz
ee
ad
.p
w
w
w

www.padeepz.net
 UNIT IV www.padeepz.net

FILE

 Contiguous logical address space

 Types:

o Data

t
 numeric

ne
 character

 binary

o Program

.
 Contents defined by file’s creator

pz
o Many types

 Consider text file, source file, executable file


ee
FILE ATTRIBUTES

Name – only information kept in human-readable form

 Identifier – unique tag (number) identifies file within file system

ad
 Type – needed for systems that support different types

 Location – pointer to file location on device

 Size – current file size

.p

 Protection – controls who can do reading, writing, executing

 Time, date, and user identification – data for protection, security, and usage
w

monitoring

 Information about files are kept in the directory structure, which is maintained on the
disk
w

 Many variations, including extended file attributes such as file checksum


w

Information kept in the directory structure

OPERATIONS

 File is an abstract data type

 Create

 Write – at write pointer location

www.padeepz.net
  Read – at read pointer location www.padeepz.net

 Reposition within file - seek

 Delete

 Truncate

 Open(Fi) – search the directory structure on disk for entry Fi, and move the content

t
of entry to memory

ne
 Close (Fi) – move the content of entry Fi in memory to directory structure on disk

 Several pieces of data are needed to manage open files:

o Open-file table: tracks open files

.
o File pointer: pointer to last read/write location, per process that has the file

pz
open

o File-open count: counter of number of times a file is open – to allow removal

of data from open-file table when last processes closes it
ee o Disk location of the file: cache of data access information

o Access rights: per-process access mode information

OPEN FILE LOCKING

ad
 Provided by some operating systems and file systems

o Similar to reader-writer locks

o Shared lock similar to reader lock – several processes can acquire concurrently
.p

o Exclusive lock similar to writer lock

 Mediates access to a file

w

 Mandatory or advisory:

o Mandatory – access is denied depending on locks held and requested

w

o Advisory – processes can find status of locks and decide what to do

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
ad
STRUCTURE

 None - sequence of words, bytes

 Simple record structure

.p

o Lines

o Fixed length
w

o Variable length

 Complex Structures
w

o Formatted document

o Relocatable load file

w

 Can simulate last two with first method by inserting appropriate control characters

 Who decides:

o Operating system

o Program

SEQUENTIAL FILE ACCESS

www.padeepz.net
 www.padeepz.net

t
ne
ACCESS METHODS

Sequential Access

read next

.
pz
write next

reset

no read after last write

ee (rewrite)

Direct Access – file is fixed length logical records

read n
ad
write n

position to n

read next
.p

write next

rewrite n
w

n = relative block number

Relative block numbers allow OS to decide where file should be placed

w
w

www.padeepz.net
  Can be built on top of base methods www.padeepz.net

 General involve creation of an index for the file

 Keep index in memory for fast determination of location of data to be operated on

(consider UPC code plus record of data about that item)

 If too large, index (in memory) of the index (on disk)

t
IBM indexed sequential-access method (ISAM)

ne
o Small master index, points to disk blocks of secondary index

o File kept sorted on a defined key

o All done by the OS

.
 VMS operating system provides index and relative files as another example

pz
ee
ad
.p
w
w

DIRECTORY STRUCTURE
w

A collection of nodes containing information about all files

Both the directory structure and the files reside on disk

DISK STRUCTURE

 Disk can be subdivided into partitions

www.padeepz.net
  Disks or partitions can be RAID protected against failure www.padeepz.net

 Disk or partition can be used raw – without a file system, or formatted with a file system

 Partitions also known as minidisks, slices

 Entity containing file system known as a volume

 Each volume containing file system also tracks that file system’s info in device directory

t
or volume table of contents

ne
 As well as general-purpose file systems there are many special-purpose file systems,
frequently all within the same operating system or computer

FILE SYSTEM ORGANIZATION

.
pz
ee
ad
.p

TYPES

 We mostly talk of general-purpose file systems

w

 But systems frequently have may file systems, some general- and some special- purpose

 Consider Solaris has

w

o tmpfs – memory-based volatile FS for fast, temporary I/O

o objfs – interface into kernel memory to get kernel symbols for debugging
w

o ctfs – contract file system for managing daemons

o lofs – loopback file system allows one FS to be accessed in place of another

o procfs – kernel interface to process structures

o ufs, zfs – general purpose file systems

www.padeepz.net
DIRECTORY OPERATIONS www.padeepz.net

 Search for a file

 Create a file

 Delete a file

 List a directory

t
 Rename a file

ne
 Traverse the file system

DIRECTORY ORGANIZATION

 Efficiency – locating a file quickly

.
pz
 Naming – convenient to users

o Two users can have same name for different files

o The same file can have several different names


ee
Grouping – logical grouping of files by properties, (e.g., all Java programs, all games, …)

SINGLE LEVEL DIRECTORY

ad
.p

 A single directory for all users

w

 Naming problem

 Grouping problem
w

TWO LEVEL DIRECTORY

 Path name
w

 Can have the same file name for different user

 Efficient searching

 No grouping capability

www.padeepz.net
 www.padeepz.net

t
ne
TREE STRUCTURED DIRECTORIES

.
pz
ee
ad
.p

 Efficient searching
w

 Grouping Capability


w

Current directory (working directory)

o cd /spell/mail/prog
w

o type list

ACYCLIC GRAPH DIRECTORIES

www.padeepz.net
 www.padeepz.net

t
. ne
pz

ee
Two different names (aliasing)

 If dict deletes list  dangling pointer

o Solutions:
ad
o Backpointers, so we can delete all pointers
Variable size records a problem

o Backpointers using a daisy chain organization

.p

o Entry-hold-count solution

 New directory entry type

w

o Link – another name (pointer) to an existing file

o Resolve the link – follow pointer to locate the file

w

GENERAL GRAPH DIRECTORY

w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
FILE SYSTEM MOUNTING
ee
ad
.p
w
w

 A file system must be mounted before it can be accessed

w

 A unmounted file system is mounted at a mount point

MOUNT POINT

www.padeepz.net
 www.padeepz.net

t
. ne
pz
FILE SHARING


ee
Sharing of files on multi-user systems is desirable

Sharing may be done through a protection scheme

 On distributed systems, files may be shared across a network

ad
 Network File System (NFS) is a common distributed file-sharing method

 If multi-user system

o User IDs identify users, allowing permissions and protections to be per-user

.p

Group IDs allow users to be in groups, permitting group access rights

o Owner of a file / directory

o Group of a file / directory

w

 Uses networking to allow file system access between systems

o Manually via programs like FTP

w

o Automatically, seamlessly using distributed file systems

w

o Semi automatically via the world wide web

 Client-server model allows clients to mount remote file systems from servers

o Server can serve multiple clients

o Client and user-on-client identification is insecure or complicated

o NFS is standard UNIX client-server file sharing protocol

www.padeepz.net
 o CIFS is standard Windows protocol www.padeepz.net

o Standard operating system file calls are translated into remote calls

 Distributed Information Systems (distributed naming services) such as LDAP, DNS,

NIS, Active Directory implement unified access to information needed for remote
computing

 All file systems have failure modes

t
o For example corruption of directory structures or other non-user data, called

ne
metadata

 Remote file systems add new failure modes, due to network failure, server failure

 Recovery from failure can involve state information about status of each remote request

.
 Stateless protocols such as NFS v3 include all information in each request, allowing easy

pz
recovery but less security

PROTECTION

 File owner/creator should be able to control:

ee
o what can be done

o by whom

 Types of access
ad
o Read

o Write
.p

o Execute

o Append

o Delete
w

o List


w

Mode of access: read, write, execute

 Three classes of users on Unix / Linux

w

RWX

 a) owner access 7  111

RWX

 b) group access 6  110

RWX
www.padeepz.net
  c) public access 1  001 www.padeepz.net

 Ask manager to create a group (unique name), say G, and add some users to the group.

 For a particular file (say game) or subdirectory, define an appropriate access.

t
ne
FILE SYSTEM STRUCTURE

.
 File structure

pz
o Logical storage unit

o Collection of related information

 File system resides on secondary storage (disks)

ee
o Provided user interface to storage, mapping logical to physical

o Provides efficient and convenient access to disk by allowing data to be stored,

located retrieved easily
ad
 Disk provides in-place rewrite and random access

o I/O transfers performed in blocks of sectors (usually 512 bytes)

 File control block – storage structure consisting of information about a file

.p

 Device driver controls the physical device

 File system organized into layers

w

LAYERED FILE SYSTEM

w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
FILE SYSTEM LAYERS
ad
 Device drivers manage I/O devices at the I/O control layer

o Given commands like ―read drive1, cylinder 72, track 2, sector 10, into memory
location 1060‖ outputs low-level hardware specific commands to hardware
.p

controller

 Basic file system given command like ―retrieve block 123‖ translates to device driver

 Also manages memory buffers and caches (allocation, freeing, replacement)

w

o Buffers hold data in transit

o Caches hold frequently used data

w

 File organization module understands files, logical address, and physical blocks
w

o Translates logical block # to physical block #

o Manages free space, disk allocation

 Logical file system manages metadata information

o Translates file name into file number, file handle, location by maintaining file
control blocks (inodes in UNIX)
www.padeepz.net
 o Directory management www.padeepz.net

o Protection

 Layering useful for reducing complexity and redundancy, but adds overhead and can
decrease performanceTranslates file name into file number, file handle, location by
maintaining file control blocks (inodes in UNIX)

o Logical layers can be implemented by any coding method according to OS

t
designer

ne
FILE SYSTEM IMPLEMENTATION

 We have system calls at the API level, but how do we implement their functions?

o On-disk and in-memory structures

.
 Boot control block contains info needed by system to boot OS from that volume

pz
o Needed if volume contains OS, usually first block of volume

 Volume control block (superblock, master file table) contains volume details


ee
o Total # of blocks, # of free blocks, block size, free block pointers or array

Directory structure organizes the files

o Names and inode numbers, master file table

ad
 Per-file File Control Block (FCB) contains many details about the file

o inode number, permissions, size, dates

o NFTS stores into in master file table using relational DB structures

.p
w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
VIRTUAL FILE SYSTEMS
ad
 Virtual File Systems (VFS) on Unix provide an object-oriented way of implementing
file systems

 VFS allows the same system call interface (the API) to be used for different types of file
systems
.p

o Separates file-system generic operations from implementation details

o Implementation can be one of many file systems types, or network file system
w

 Implements vnodes which hold inodes or network file details

o Then dispatches operation to appropriate file system implementation routines

w
w

www.padeepz.net
VFS IMPLEMENTATION www.padeepz.net

 For example, Linux has four object types:

o inode, file, superblock, dentry

 VFS defines set of operations on the objects that must be implemented

o Every object has a pointer to a function table

t
 Function table has addresses of routines to implement that function on that

ne
object

 For example:

• int open(. . .)—Open a file

.
• int close(. . .)—Close an already-open file

pz
• ssize t read(. . .)—Read from a file

• ssize t write(. . .)—Write to a file

• int mmap(. . .)—Memory-map a file

ee
DIRECTORY IMPLEMENTATION

• Linear list of file names with pointer to the data blocks

o Simple to program
ad
o Time-consuming to execute

 Linear search time

.p

 Could keep ordered alphabetically via linked list or use B+ tree

• Hash Table – linear list with hash data structure

o Decreases directory search time

w

o Collisions – situations where two file names hash to the same location

o Only good if entries are fixed size, or use chained-overflow method

w

ALLOCATION METHOD CONTIGUOUS

w

• An allocation method refers to how disk blocks are allocated for files:

• Contiguous allocation – each file occupies set of contiguous blocks

o Best performance in most cases

o Simple – only starting location (block #) and length (number of blocks) are
required
www.padeepz.net
 www.padeepz.net
o Problems include finding space for file, knowing file size, external fragmentation,
need for compaction off-line (downtime) or on-line

t
. ne
pz
•
ee
Linked allocation – each file a linked list of blocks

o File ends at nil pointer

o No external fragmentation
ad
o Each block contains pointer to next block

o No compaction, external fragmentation

o Free space management system called when new block needed

.p

o Improve efficiency by clustering blocks into groups but increases internal

fragmentation

o Reliability can be a problem

w

o Locating a block can take many I/Os and disk seeks

w
w

www.padeepz.net
 www.padeepz.net

t
ne.
pz
ee
FILE ALLOCATION TABLE
ad
.p
w
w
w

www.padeepz.net
• Indexed allocation www.padeepz.net

• Each file has its own index block(s) of pointers to its data blocks

• Logical view

t
. ne
pz
• index table
ee
ad
.p
w
w
w

• Need index table

• Random access

• Dynamic access without external fragmentation, but have overhead of index block
www.padeepz.net
 • www.padeepz.net
Mapping from logical to physical in a file of maximum size of 256K bytes and block
size of 512 bytes. We need only 1 block for index table

FREE SPACE MANAGEMENT

• File system maintains free-space list to track available blocks/clusters

o (Using term ―block‖ for simplicity)

• Bit vector or bit map (n blocks)

t
. ne
pz
ee
Block number calculation

(number of bits per word) *(number of 0-value words) +offset of first 1 bit

• Bit map requires extra space

ad
o Example:

 block size = 4KB = 212 bytes

 disk size = 240 bytes (1 terabyte)

.p

 n = 240/212 = 228 bits (or 32MB)

 if clusters of 4 blocks -> 8MB of memory

w

• Easy to get contiguous files

Linked list (free list)

w

• Cannot get contiguous space easily

• No waste of space
w

• No need to traverse the entire list (if # free blocks recorded)

www.padeepz.net
 www.padeepz.net

t
. ne
pz
• Grouping

o Modify linked list to store address of next n-1 free blocks in first free block, plus

•
ee
a pointer to next block that contains free-block-pointers (like this one)

Counting

o Because space is frequently contiguously used and freed, with contiguous-

allocation allocation, extents, or clustering
ad
 Keep address of first free block and count of following free blocks

 Free space list then has entries containing addresses and counts

• Space Maps
.p

o Used in ZFS

o Consider meta-data I/O on very large file systems

w

 Full data structures like bit maps couldn’t fit in memory -> thousands of
I/Os
w

o Divides device space into metaslab units and manages metaslabs

 Given volume can contain hundreds of metaslabs

w

o Each metaslab has associated space map

 Uses counting algorithm

o But records to log file rather than file system

 Log of all block activity, in time order, in counting format

www.padeepz.net
 www.padeepz.net
o Metaslab activity -> load space map into memory in balanced-tree structure,
indexed by offset

 Replay log into that structure

 Combine contiguous free blocks into single entry

• Efficiency dependent on:

o Disk allocation and directory algorithms

t
ne
o Types of data kept in file’s directory entry

o Pre-allocation or as-needed allocation of metadata structures

o Fixed-size or varying-size data structures

.
PAGE CACHE

pz
• A page cache caches pages rather than disk blocks using virtual memory techniques
and addresses

• Memory-mapped I/O uses a page cache

•
ee
Routine I/O through the file system uses the buffer (disk) cache

This leads to the following figure

ad
.p
w
w
w

www.padeepz.net
 www.padeepz.net

IO SYSTEMS

 Incredible variety of I/O devices

o Storage

o Transmission

t
o Human-interface

ne
 Common concepts – signals from I/O devices interface with computer

o Port – connection point for device

o Bus - daisy chain or shared direct access

.


pz
PCI bus common in PCs and servers, PCI Express (PCIe)

 expansion bus connects relatively slow devices

o Controller (host adapter) – electronics that operate port, bus, device

ee 


Sometimes integrated

Sometimes separate circuit board (host adapter)

 Contains processor, microcode, private memory, bus controller, etc

ad
 Some talk to per-device controller with bus controller, microcode,
memory, etc
.p
w
w
w

www.padeepz.net
 www.padeepz.net

t
. ne
pz
ee
 I/O instructions control devices
ad
 Devices usually have registers where device driver places commands, addresses, and data
to write, or read data from registers after command execution

o Data-in register, data-out register, status register, control register

o Typically 1-4 bytes, or FIFO buffer

.p

 Devices have addresses, used by

o Direct I/O instructions

w

o Memory-mapped I/O

 Device data and command registers mapped to processor address space

w

 Especially for large address spaces (graphics)

w

www.padeepz.net
 www.padeepz.net

t
ne.
pz
ee
ad
.p
w
w
w

www.padeepz.net
 www.padeepz.net

UNIT V

LINUX SYSTEM – BASIC CONCEPTS

Linodes run Linux. Linux is an operating system, just like Windows and Mac OS X. As an
operating system, Linux manages your Linode’s hardware and provides services your other
software needs to run.

t
Linux is a very hands-on operating system. If running Windows is like driving an automatic,

ne
then running Linux is like driving a stick. It can take some work, but once you know your way
around Linux, you’ll be using the command line and installing packages like a pro. This article
aims to ease you into the world of Linux.

This guide is intended to be very beginner-friendly. It takes a Linux 101 approach to

.
explanations for basic concepts. There are a few how-to sections as well, which are intended to
get you on your feet with your Linode. At times we’ll link off to a different guide that has more

pz
details on a particular topic.

Everything on a Linux system is case-sensitive. That means that photo.jpg, photo.JPG, and
Photo.jpg are all different files. Usernames and passwords are also case-sensitive.
ee
Linux, like Mac OS X, is based on the Unix operating system. A research team at AT&T’s Bell
Labs developed Unix in the late 1960s and early 1970s with a focus on creating an operating
system that would be accessible and secure for multiple users.

Corporations started licensing Unix in the 1980s and 1990s. By the late 1980s, there was interest
ad
in building a free operating system that would be similar to Unix, but that could be tinkered with
and redistributed. In 1991, Linus Torvalds released the Linux kernel as free, open-source
software. Open source means that the code is fully visible, and can be modified and redistributed.

Strictly speaking, Linux is the kernel, not the entire operating system. The kernel provides an
interface between your Linode’s hardware and the input/output requests from applications. The
.p

rest of the operating system usually includes many GNU libraries, utilities, and other software,
from the Free Software Foundation. The operating system as a whole is known as GNU/Linux.
w

A Little Bit About Servers

Your Linode is a type of server. What’s a server? A server is a type of computer that provides
services over a network, or connected group of computers. When people think about servers,
w

they’re usually thinking of a computer that is:

 Always (or almost always) on


w

Connected to the Internet

 Contains programs and files for websites and/or other Internet content

Since a server is a type of computer, there are a lot of similarities between a Linode and your
home computer. Some important similarities include:

 The physical machine: Your Linode is hosted on a physical machine. It’s sitting in one of
our data centers.
www.padeepz.net
  The operating system: As we mentioned in the introduction, Linodes www.padeepz.net
use the Linux
operating system. It’s just another type of operating system like Windows or Mac OS X.
 Applications: Just like you can install applications on your home computer or
smartphone, you can install applications on your Linode. These applications help your
Linode do things like host a website. WordPress is a popular website application, for
example. Applications are also known as software and programs.
 Files and directories: In the end, whether it’s an application or a photo, everything on
your Linode is a file. You can create new files, edit and delete old ones, and navigate
through directories just like you would on your home computer. In Linux, folders are

t
called directories.
 Internet access: Your Linode is connected to the Internet. That’s how you connect to it to

ne
get everything set up, and how your users connect to it to view your website or download
your app.

.
SYSTEM ADMINISTRATION

pz
Linux System Administration Basics
This presents a collection of common issues and useful tips for Linux system administration.
Whether you’re new to system administration or have been maintaining systems for some time,
ee
we hope these tips are helpful regardless of your background or choice in Linux distributions.

Basic Configuration
These tips cover some of the basic steps and issues encountered during the beginning of system
ad
configuration. We provide a general getting started guide for your convenience if you’re new to
Linode and basic Linux system administration. Additionally, you may find some of our
Introduction to Linux Concepts guide useful.

Set the Hostname

.p

Please follow our instructions for setting your hostname. Issue the following commands to make
sure it is set properly:
w

1 hostname
2 hostname -f

The first command should show your short hostname, and the second should show your fully
w

qualified domain name (FQDN).

Set the Timezone

w

When setting the timezone of your server, it may be best to set it to the timezone of the bulk of
your users. If you’re unsure which timezone would be best, consider using universal coordinated
time or UTC (i.e. Greenwich Mean Time).

By default, Linode base installs are set to Eastern Standard Time. The following process will set
the timezone manually, though many operating systems provide a more elegant method for
changing timezones. To change the time zone manually, you must find the proper zone file in
/usr/share/zoneinfo/ and link that file to /etc/localtime. See the example below for
www.padeepz.net
 www.padeepz.net
common possibilities. Please note that all contents following the double hashes (eg. ##) are
comments and should not be copied into your terminal.

ln -sf /usr/share/zoneinfo/UTC /etc/localtime ## for Universal Coordinated

Time
1
2
ln -sf /usr/share/zoneinfo/EST /etc/localtime ## for Eastern Standard Time
3
4
ln -sf /usr/share/zoneinfo/US/Central /etc/localtime ## for American Central
5
time (including DST)
6

t
7
ln -sf /usr/share/zoneinfo/US/Eastern /etc/localtime ## for American Eastern

ne
(including DST)

To change the time zone in Debian and Ubuntu systems, issue the following command and
answer the questions as prompted by the utility:

.
1 dpkg-reconfigure tzdata

pz
In Arch Linux, set the timezone in the /etc/rc.conf file by configuring the TIMEZONE= setting
in the ―Localization‖ section. This line will resemble the following:

/etc/rc.conf
ee
TIMEZONE=‖America/New_York‖

Note that the string specified in TIMEZONE refers to the ―zoneinfo‖ file located in or below the
/usr/share/zoneinfo/ directory.
ad
Use the /etc/hosts File

The /etc/hosts file provides a list of IP addresses with corresponding hostnames. This allows
you to specify hostnames for an IP address once on the local machine, and then have multiple
applications connect to external resources via their hostnames. The system of host files predates
.p

DNS, and hosts files are always checked before DNS is queried. As a result, /etc/hosts can be
useful for maintaining small ―internal‖ networks, for development purposes, and for managing
clusters.
w

Some applications require that the machine properly identify itself in the /etc/hosts file. As a
result, we recommend configuring the /etc/hosts file shortly after deployment. Here is an
example file:
w

/etc/hosts

127.0.0.1 localhost.localdomain localhost 12.34.56.78 squire.example.com squire

w

You can specify a number of hostnames on each line separated by spaces. Every line must begin
with one and only one IP address. In this case, replace 12.34.56.78 with your machine’s IP
address. Let us consider a few additional /etc/hosts entries:

/etc/hosts

74.125.67.100 test.com 192.168.1.1 stick.example.com

www.padeepz.net
 www.padeepz.net
In this example, all requests for the test.com‖ hostname or domain will resolve to the IP address
74.125.67.100, which bypasses the DNS records for test.com and returns an alternate website.

The second entry tells the system to look to 192.168.1.1 for the domain stick.example.com.
These kinds of host entries are useful for using ―private‖ or ―back channel‖ networks to access
other servers in a cluster without needing to access the public network.

Network Diagnostics

t
The following tips address the basic usage and functionality of a number of tools that you can

ne
use to assess and diagnose network problems. If you suspect connectivity issues, including
output of the relevant commands in your support ticket can help our staff diagnose your issue.
This is particularly helpful in cases where networking issues are intermittent.

Using the Ping Command

.
The ping command tests the connection between the local machine and a remote address or

pz
machine. The following command ―pings‖ google.com and 74.125.67.100:

1 ping google.com ping 74.125.67.100

These commands send a bit of data (i.e. an ICMP packet) to the remote host, and wait for a
ee
response. If the system is able to make a connection, for every packet it will report on the ―round
trip time.‖ Here is the output of four pings of google.com:

64 bytes from yx-in-f100.1e100.net (74.125.45.100): icmp_seq=1 ttl=50

time=33.8 ms
ad
1 64 bytes from yx-in-f100.1e100.net (74.125.45.100): icmp_seq=2 ttl=50
2 time=53.2 ms
3 64 bytes from yx-in-f100.1e100.net (74.125.45.100): icmp_seq=3 ttl=50
4 time=35.9 ms
64 bytes from yx-in-f100.1e100.net (74.125.45.100): icmp_seq=4 ttl=50
time=37.5 ms
.p

In this case yx-in-f100.1e100.net is the reverse DNS for this IP address. The time field
specifies in milliseconds that the round trip takes for an individual packet. When you’ve gathered
the amount of information you need, send Control+C to interrupt the process. At this juncture,
you’ll be presented with some statistics. This will resemble:
w

1 --- google.com ping statistics ---

2 4 packets transmitted, 4 received, 0% packet loss, time 3007ms
w

3 rtt min/avg/max/mdev = 33.890/40.175/53.280/7.679 ms

There are several important data points to notice. They are:

w

 Packet Loss, or the discrepancy between the number of packets sent and the number of packets
that return successfully.
 Round Trip Time statistics on the final line report important information about all the ping
responses. For this ping we see that the fastest packet round trip took 33.89 milliseconds. The
longest packet took 53.28 milliseconds. The average round trip took 40.175 milliseconds. A
single standard deviation unit for these four packets is 7.67 milliseconds.

www.padeepz.net
Use the ping tool to contact a server and ensure that you are able to make awww.padeepz.net
connection.
Furthermore, ping is useful as an informal diagnostic tool to measure point-to-point network
latency, and as a network connection testing tool.

Using the traceroute Command

The traceroute command expands on the functionality of the ping command. traceroute
provides a report on the path that the packets take to get from the local machine to the remote
machine. Route information is useful when troubleshooting a networking issue: if there is packet

t
loss in one of the first few ―hops‖ the problem is often related to the user’s local area network

ne
(LAN) or Internet service provider (ISP). By contrast, if there is packet loss near the end of the
route, the problem may be caused by an issue with the server’s connection.

Here is an example of output from a traceroute command:

1 traceroute to google.com (74.125.53.100), 30 hops max, 40 byte packets

.
pz
1 207.192.75.2 (207.192.75.2) 0.414 ms 0.428 ms 0.509 ms 2 vlan804.tbr2.mmu.nac.net
(209.123.10.13) 0.287 ms 0.324 ms 0.397 ms 3 0.e1-1.tbr2.tl9.nac.net (209.123.10.78) 1.331 ms
1.402 ms 1.477 ms 4 core1-0-2-0.lga.net.google.com (198.32.160.130) 1.514 ms 1.497 ms 1.519
ms 5 209.85.255.68 (209.85.255.68) 1.702 ms 72.14.238.232 (72.14.238.232) 1.731 ms 21.031
ms 6 209.85.251.233 (209.85.251.233) 26.111 ms 216.239.46.14 (216.239.46.14) 23.582 ms
ee
23.468 ms 7 216.239.43.80 (216.239.43.80) 123.668 ms 209.85.249.19 (209.85.249.19) 47.228
ms 47.250 ms 8 209.85.241.211 (209.85.241.211) 76.733 ms 216.239.43.80 (216.239.43.80)
73.582 ms 73.570 ms 9 209.85.250.144 (209.85.250.144) 86.025 ms 86.151 ms 86.136 ms 10
64.233.174.131 (64.233.174.131) 80.877 ms 216.239.48.34 (216.239.48.34) 76.212 ms
64.233.174.131 (64.233.174.131) 80.884 ms 11 216.239.48.32 (216.239.48.32) 81.267 ms
81.198 ms 81.186 ms 12 216.239.48.137 (216.239.48.137) 77.478 ms pw-in-f100.1e100.net
ad
(74.125.53.100) 79.009 ms 216.239.48.137 (216.239.48.137) 77.437 ms

Often the hostnames and IP addresses on either side of a failed jump are useful in determining
who operates the machine where the routing error occurs. Failed jumps are designated by line
with three asterisks (e.g. * * *).
.p

Furthermore, including traceroute information in tickets to Linode support is sometimes useful

when trying to diagnose network issues. You may also want to forward traceroute information
to your Internet Service Provider (ISP) if you suspect that the connectivity issue is with your
w

ISP’s network. Recording traceroute information is particularly useful if you are experiencing
an intermittent issue.
w

Using the mtr Command

The ―mtr‖ command, like the traceroute tool, provides information about the route that Internet
traffic takes between the local system and a remote host. However, mtr provides additional
w

information about the round trip time for the packet. In a way, you can think of mtr as a
combination of traceroute and ping.

Here is the example output of an mtr command:

1 HOST: squire.example.com Loss% Snt Last Avg Best Wrst StDev

1. 256.129.75.4 0.0% 10 0.4 0.4 0.3 0.6 0.1

www.padeepz.net
 2. vlan804.tbr2.mmu.nac.net 0.0% 10 0.3 0.4 0.3 0.7 0.1 www.padeepz.net
3. 0.e1-1.tbr2.tl9.nac.net 0.0% 10 4.3 4.4 1.3 11.4 4.1
4. core1-0-2-0.lga.net.google.c 0.0% 10 64.9 11.7 1.5 64.9 21.2
5. 209.85.255.68 0.0% 10 1.7 4.5 1.7 29.3 8.7
6. 209.85.251.9 0.0% 10 23.1 35.9 22.6 95.2 27.6
7. 72.14.239.127 0.0% 10 24.2 24.8 23.7 26.1 1.0
8. 209.85.255.190 0.0% 10 27.0 27.3 23.9 37.9 4.2
9. gw-in-f100.1e100.net 0.0% 10 24.1 24.4 24.0 26.5 0.7

t
Used without the --report flag, mtr tracks the speed of the connection in real time until you
exit the program. Additionally, be aware that mtr will pause for a few moments before

ne
generating output. For more information regarding mtr consider our guide to diagnosing network
issues with mtr.

System Diagnostics

.
If you’re having an issue with your Linode that is neither related to networking, nor another

pz
easily diagnosable application issue, it is worthwhile to rule out ―hardware‖ and operating
system level issues. Use the following tools to better diagnose and resolve these kinds of issues.

If you determine that you have a problem with memory usage, please reference our document
regarding resolving memory usage issues. Use the following tools and approaches to determine
ee
the specific cause of your troubles.

Check Current Memory Usage

If you need to see how much memory your system is using at the current moment issue the
ad
following command:

1 free -m

On a moderately utilized Linode 1GB, this command will generate output that resembles the
following:
.p

1 total used free shared buffers cached

Mem: 1002 956 46 0 171 357 -/+ buffers/cache: 427 575 Swap: 127 39 88
w

This output takes a little bit of careful reading to interpret correctly. Out of a total 1002
megabytes of memory (RAM), the system is using 956 megabytes, and has 46 megabytes free.
w

However, the system also has 427 megabytes of ―stale‖ data buffered and stored in cache. The
operating system will ―drop‖ the caches when and if it needs the space, but retains the cache if
there is no other need for the space. It is totally normal for a Linux system to leave old data in
RAM until the space is needed, and you should not be alarmed if only a small amount of
w

memory is actually ―free.‖

In the above example, there are 575 megabytes of memory that are actually free. This means 575
megabytes are available to your system when you start an additional process or a running
application needs more memory.

Monitor IO Usage with vmstat

www.padeepz.net
The vmstat tool provides information about memory, swap utilization, IOwww.padeepz.net
wait, and system
activity. It is particularly useful for diagnosing I/O-related issues.

If you think you’re having an I/O issue then run the following command:

1 vmstat 1 20

This runs a vmstat every second, twenty times. We find this gives a pretty good sample of the
current state of the system. The output generated resembles the following:

t
ne
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu--
1 --
2 r b swpd free buff cache si so bi bo in cs us sy id
3 wa
4 0 0 4 32652 47888 110824 0 0 0 2 15 15 0 0 100 0
5 0 0 4 32644 47888 110896 0 0 0 4 106 123 0 0 100 0
6 0 0 4 32644 47888 110912 0 0 0 0 70 112 0 0 100 0

.
7 0 0 4 32644 47888 110912 0 0 0 0 92 121 0 0 100 0
8 0 0 4 32644 47888 110912 0 0 0 36 97 136 0 0 100 0

pz
9 0 0 4 32644 47888 110912 0 0 0 0 96 119 0 0 100 0
10 0 0 4 32892 47888 110912 0 0 0 4 96 125 0 0 100 0
11 0 0 4 32892 47888 110912 0 0 0 0 70 105 0 0 100 0
12 0 0 4 32892 47888 110912 0 0 0 0 97 119 0 0 100 0
13 0 0 4 32892 47888 110912 0 0 0 32 95 135 0 0 100 0
14 0 0 ee 4 33016 47888 110912 0 0 0 0 75 107 0 0 100 0
15 0 0 4 33512 47888 110912 0 0 0 24 113 134 0 0 100 0
16 0 0 4 33512 47888 110912 0 0 0 0 175 244 0 0 100 0
17 0 0 4 33512 47888 110912 0 0 0 0 92 148 0 0 100 0
18 0 0 4 33512 47888 110912 0 0 0 0 114 162 0 0 100 0
19 0 0 4 33512 47888 110912 0 0 0 36 100 157 0 0 100 0
20 0 0 4 33388 47888 110912 0 0 0 0 116 166 0 0 100 0
21 0 0 4 33388 47888 110912 0 0 0 0 97 157 0 0 100 0
ad
22 0 0 4 33388 47888 110912 0 0 0 0 89 144 0 0 100 0
0 0 4 33380 47888 110912 0 0 0 0 181 185 0 0 99 0

The memory and swap columns provide the same kind of information provided by the ―free -m‖
command, albeit in a slightly more difficult to comprehend format. The most salient information
.p

produced by this command is the wa column, which is the final column in most implementations.
This field displays the amount of time the CPU spends waiting for IO operations to complete.

If this number is consistently and considerably higher than 0, you might consider taking
w

measures to address your IO usage. However, if the vmstat output resembles the above, you can
be sure in the knowledge that you’re not experiencing an IO-related issues.

If you are experiencing an intermittent issue, you will need to run vmstat when you experience
w

the issue in order to properly diagnose or rule out an IO issue. vmsat output can sometimes help
support diagnose problems.
w

Monitor Processes, Memory, and CPU Usage with htop

If you want a more organized and real-time view of the current state of your system, we
recommend a tool called htop. This is not installed by default on most systems. To install htop,
issue one of the following commands, depending on which distribution you use:

1 apt-get install htop

2 yum install htop
3 pacman -S htop
www.padeepz.net
4 emerge sys-process/htop www.padeepz.net

Now, at a command line, issue the following command:

1 htop

You can quit at any time by pressing the F10 or Q keys. There are a couple of htop behaviors that
may not be initially intuitive. Take note of the following:

t
 The memory utilization graph displays used memory, buffered memory, and cached memory.

ne
The numbers displayed at the end of this graph reflect the total amount of memory available
and the total amount memory on the system as reported by the kernel.
 The default configuration of htop presents all application threads as independent processes,
which is non-intuitive. You can disable this by selecting the “setup” option with F2, then
“Display Options,” and then toggling the “Hide userland threads” option.
 You can toggle a “Tree” view with the F5 key that usefully displays the processes in a hierarchy

.
and shows which processes were spawned by which other processes. This is helpful in

pz
diagnosing a problem when you’re having trouble figuring out what processes are what.

File System Management

Historically, web developers and editors have used the FTP protocol to transfer and manage files
ee
on a remote system. FTP, however, is very insecure and inefficient for managing the files on a
system when you have SSH access.

If you’re new to administering systems and the Linux world, you might consider our ―Tools &
Reference‖ section and articles including: ―installing and using WinSCP‖ using rsync to
synchronize files and ―using SSH and the terminal.‖
ad
As always, if you are giving other users access to upload files to your server, it would be wise to
consider the security implications of all additional access that you grant to third parties seriously.

How to Upload files to a Remote Server

.p

If you’re used to using an FTP client, OpenSSH (which is included and active with all of the
Linode provided installation templates) allows you to use an FTP-like interface over the SSH
protocol. Known as ―SFTP,‖ many clients support this protocol, including: ―WinSCP‖ for
w

Windows, ―Cyberduck‖ for Mac OS X, and ―Filezilla‖ for Linux, OS X, and Windows desktops.

If you are accustomed to FTP, SFTP is great option. Do note that by default, whatever access a
w

user has to a file system at the command line, they will also have over SFTP. Consider file
permissions very carefully.

Conversely, you can use Unix utilities including scp and rsync to securely transfer files to your
w

Linode. On local machine, a command to copy team-info.tar.gz would look like:

1 scp team-info.tar.gz squire@lollipop.example.com:/home/squire/backups/

The command, scp, is followed by the name of the file on the local file system to be transferred.
Next is the username and hostname of the remote machine, separated by an ―at‖ sign (e.g. @).
Following the hostname, there is a colon (e.g. :) and the path on the remote server where the file
should be uploaded to. Taken another way, this command would be:
www.padeepz.net
 scp [/path/to/local/file] [remote-username]@[remote- www.padeepz.net
1
hostname]:[/path/to/remote/file]

This command is available by default on OS X and Linux machines. You can use it to copy files
to a Linode, as well as between remote servers. If you have SSH keys deployed, you can use the
scp command without entering a password for every transfer.

The syntax of scp follows the form scp [source] [destination]. You can copy files from a
remote host to the local machine by reversing the order of the paths in the above example.

t
ne
How to Protect Files on a Remote Server

Because Linode servers are network accessible and often have a number of distinct users,
maintaining the security of files is often an important concern. We recommend you familiarize
yourself with our basic security guide. Furthermore, our documentation of access control with
user accounts and permissions may provide additional insight.

.
pz
Additionally, we suggest the following best practices for maintaining security:

 Only give users the permission to do what they need to. This includes application specific users.
 Only run services on public interfaces that you are actively using. One common source of
security vulnerabilities are in daemons that are left running and unused. This includes database


ee
servers, HTTP development servers, and FTP servers.
Use SSH connections whenever possible to secure and encrypt the transfer of sensitive
information.

Understanding and Using Sym Linking

ad
―Symbolic Linking,‖ colloquially ―sym linking,‖ allows you to create an object in your
filesystem that points to another object on your filesystem. This is useful when you need to
provide users and applications access to specific files and directories without reorganizing your
folders. This way you can provide restricted users access to your web-accessible directories
without moving your DocumentRoot into their home directories.
.p

To create a symbolic link, issue a command in the following format:

1 ln -s /home/squire/config-git/etc-hosts /etc/hosts
w

This creates a link of the file etc-hosts at the location of the system’s /etc/hosts file. More
generically. this command would read:
w

1 ln -s [/path/to/target/file] [/path/to/location/of/sym/link]

Note the following features of the link command:

w

 The final term, the location of the link, is optional. If you opt to omit the link destination, a link
will be created in the current directory with the same name as the file you’re linking to.
 When specifying the location of the link, ensure that path does not have a final trailing slash.
You can create a sym link that targets a directory, but sym links cannot terminate with slashes.
 You may remove a symbolic link without affecting the target file.
 You can use relative or absolute paths when creating a link.

www.padeepz.net
How to Manage and Manipulate Files on a Linux System www.padeepz.net

If you’re new to using Linux and manipulating files on the terminal interface we encourage you
to consider our using the terminal document. This tip provides an overview of basic file
management operations.

To copy files, issue the following command:

cp /home/squire/todo.txt /home/squire/archive/todo.01.txt

t
ne
This copies todo.txt to an archive folder, and adds a number to the file name. If you want to
recursively copy all of the files and subdirectories in a directory to another directory, use the -R
option. This command looks like:

1 cp -R /home/squire/archive/ /srv/backup/squire.01/

.
If you need to move a file or directory, use the following command:

pz
1 mv /home/squire/archive/ /srv/backup/squire.02/

You can also use the mv command to rename a file.

ee
To delete a file, issue a command in the following form:

1 rm scratch.txt

This will delete the scratch.txt file from the current directory.
ad
For more information about file system navigation and manipulation, please consider our
documentation of file system navigation in the using the terminal document.

Package Management
.p

Contemporary Linux systems use package management tools to facilitate the installation and
maintenance of all software on your system. For more in-depth coverage of this topic, please
reference our package management guide.
w

While package management provides a number of powerful features, it is easy to obviate the
benefits of package management. If you install software manually without package management
tools, it becomes very difficult to keep your system up to date and to manage complex
w

dependencies. For these reasons, we recommend installing all software through package
management tools unless other means are absolutely necessary. The following tips outline a
couple of basic package management tasks.
w

How to Know What Packages are Installed on Your System

Because packages are so easy to install, and often pull in a number of dependencies, it can be
easy to lose track of what software is installed on your system. The following commands provide
a list of installed packages on your system.

For Debian and Ubuntu systems, issue the following command:

www.padeepz.net
1 dpkg -l www.padeepz.net

The following example presents the first few lines of the output of this command on a production
Debian Lenny system.

||/ Name Version Description

+++-============================-============================-
1 ===============================
2 ii adduser 3.110 add and remove users and groups
3 ii apache2-mpm-itk 2.2.6-02-1+lenny2 multiuser MPM for Apache

t
4 2.2
5 ii apache2-utils 2.2.9-10+lenny4 utility programs for

ne
6 webservers
7 ii apache2.2-common 2.2.9-10+lenny4 Apache HTTP Server
8 common files
9 ii apt 0.7.20.2+lenny1 Advanced front-end for dpkg
ii apt-utils 0.7.20.2+lenny1 APT utility programs
ii bash 3.2-4 The GNU Bourne Again SHell

.
pz
For CentOS and Fedora systems, issue the following command:

1 yum list installed

The following example presents a few relevant lines of the output of this command:
ee
1 MAKEDEV.i386 3.23-1.2 installed
2 SysVinit.i386 2.86-15.el5 installed

CentOS and Fedora systems provide the name of the package (e.g. SysVinit), the architecture it
was compiled for (e.g. i386), and the version of the build installed on the system (e.g. 2.86-
ad
15.el5).

For Arch Linux systems, issue the following command:

1 pacman -Q
.p

This command provides a total list of all packages installed on the system. Arch also allows you
to filter these results to display only packages that were explicitly installed (with the -Qe option)
or that were installed as dependencies (with the -Qd option). The above command is thus the
w

union of the output of the following commands:

1 pacman -Qe
2 pacman -Qd
w

The following is a sample of the format of this output:

w

1 perl-www-mechanize 1.60-
2 perl-yaml 0.70-1
3 pkgconfig 0.23-1
4 procmail 3.22-2
5 python 2.6.4-1
6 rsync 3.0.6-1

For Gentoo Linux systems, issue the following command:

www.padeepz.net
1 emerge -evp --deep world www.padeepz.net

The following is a sample of the format of this output:

These are the packages that would be merged, in order:

1
2
Calculating dependencies... done!
3
[ebuild R ] sys-libs/ncurses-5.6-r2 USE="unicode -debug -doc -gpm -
4
minimal -nocxx -profile -trace" 0 kB
5
[ebuild R ] virtual/libintl-0 0 kB
6

t
[ebuild R ] sys-libs/zlib-1.2.3-r1 0 Kb

ne
Because there are often a large number of packages installed on any given system, the output of
these commands is often quite large. As a result, it is often useful to use tools like grep and less
to make these results more useful. For example, the command :

1 dpkg -l | grep "python"

.
pz
will return a list of all packages with the word python in their name or description. Similarly, the
following command:

1 dpkg -l | less
ee
will return the same list as the plain ―dpkg -l ; however, the results will appear in the less
pager, which allows you to search and scroll more easily.

You can append | grep "[string]" to these commands to filter package list results, or | less
to display the results in a pager, regardless of distribution.
ad
How to Discover Package Names and Information

Sometimes the name of a package doesn’t correspond to the name that you may associate with a
given piece of software. As a result, most package management tools make provide an interface
to search the package database. These search tools may be helpful if you’re looking for a specific
.p

piece of software but don’t know what it’s called.

For Debian and Ubuntu systems, issue the following command:

w

1 apt-cache search [package-name]

This will search the local package database for a given term and generate a list with brief
w

descriptions. An excerpt of the output for apt-cache search python follows:

txt2regex - A Regular Expression "wizard", all written with bash2 builtins

1
vim-nox - Vi IMproved - enhanced vi editor
w

2
vim-python - Vi IMproved - enhanced vi editor (transitional package)
3
vtk-examples - C++, Tcl and Python example programs/scripts for VTK
4
zope-plone3 - content management system based on zope and cmf
5
zorp - An advanced protocol analyzing firewall
6
groovy - Agile dynamic language for the Java Virtual Machine
7
python-django - A high-level Python Web framework
8
python-pygresql-dbg - PostgreSQL module for Python (debug extension)
9
python-samba - Python bindings that allow access to various aspects of
10
Samba

www.padeepz.net
Note that apt-cache search queries the full records for all of the packages,www.padeepz.net
and not simply the
titles and the descriptions displayed here, hence the inclusion of vim-nox and groovy which both
mention python in their descriptions. To see the full record on a package issue the following
command:

1 apt-cache show [package-name]

This provides information regarding the maintainer, the dependencies, the size, the homepage of
the upstream project, and a description of the software. This command can be used to provide

t
additional information about a package from the command line.

ne
For CentOS and Fedora systems, issue the following command:

1 yum search [package-name]

This generates a list of all packages available in the package database that match the given term.

.
See the following excerpt for an example of the output of yum search wget:

pz
Loaded plugins: fastestmirror
1
Loading mirror speeds from cached hostfile
2
* addons: centos.secsup.org
3
* base: centos.secsup.org
4
* extras: centos.secsup.org
ee
5
* updates: styx.biochem.wfubmc.edu
6
================================ Matched: wget
7
=================================
8
wget.i386 : A utility for retrieving files using the HTTP or FTP protocols.

You can use the package management tools to discover more information about a specific
ad
package. Use the following command to get a full record from the package database:

1 yum info [package-name]

This output presents more in-depth information concerning the package, its dependencies,
.p

origins, and purpose.

For Arch Linux systems, issue the following command:

w

1 pacman -Ss [package-name]

This will perform a search of the local copy of the package database. Here is an excerpt of results
for a search for ―python:
w

1 extra/twisted 8.2.0-1
2 Asynchronous networking framework written in Python.
w

3 community/emacs-python-mode 5.1.0-1
4 Python mode for Emacs

The terms ―extra‖ and ―community‖ refer to which repository the software is located in. This
level of specificity is unnecessary when specifying packages to install or display more
information about. To request more information about a specific package issue a command in the
following form:

1 pacman -Si [package-name]

www.padeepz.net
Running pacman with the -Si option generates the package’s record from www.padeepz.net
the database. This
record includes information about dependencies, optional dependencies, package size, and a brief
description.

For Gentoo Linux systems, issue one of the following commands:

1 emerge --search [package-name]

2 emerge --searchdoc [package-name]

t
The first command only searches the database for package names. The second command
searches through the database for package names and descriptions. These commands will allow

ne
you to search your local package tree (i.e. portage) for the specific package name or term. The
output of either command is similar to the excerpt presented bellow.

1 Searching...
2 [ Results for search key : wget ]

.
3 [ Applications found : 4 ]
4

pz
5 * app-emacs/emacs-wget
6 Latest version available: 0.5.0
7 Latest version installed: [ Not Installed ]
8 Size of files: 36 kB
9 Homepage: http://pop-club.hp.infoseek.co.jp/emacs/emacs-wget/
10 Description:
ee Wget interface for Emacs
11 License: GPL-2

Because the output provided by the emerge --search command is rather verbose, there is no
―show more information‖ tool, unlike other distributions’ tools. The emerge --search
command accepts input in the form of a regular expression if you need to narrow results even
further.
ad
Since there are often a large number of results for package searches, these commands output a
great quantity of text. As a result it is often useful to use tools like grep and less to make these
results easier to scroll. For example, the command :
.p

1 apt-cache search python | grep "xml"

will return the subset of the list of packages which matched for the search term ―python,‖ and
that mention xml in their name or short description. Similarly, the following command:
w

1 apt-cache search python | less

w

will return the same list as the plain apt-cache search python but the results will appear in
the less pager. This allows you to search and scroll more conveniently.

You can append | grep "[string]" to any of these commands to filter package search results,
w

or | less to display the results in the less pager, regardless of distribution.

Text Manipulation
Among Linux and UNIX-like systems, nearly all system configuration information is stored and
manipulated in plain text form. These tips provide some basic information regarding the
manipulation of text files on your system.
www.padeepz.net
How to Search for a String in Files with grep www.padeepz.net

The grep tool allows you to search a stream of text, such as a file or the output of a command,
for a term or pattern matching a regular expression.

To use the grep tool, issue a command in the following format:

1 grep "^Subject:.*HELP.*" /home/squire/mbox

t
This will search the mail spool for subject lines (i.e. begins with the word ―Subject:‖), beginning

ne
with any number of characters, containing the word ―help‖ in upper case, and followed by any
number of additional characters. grep would then print these results on the terminal.

grep provides a number of additional options that, if specified, force the program to output the
context for each match (e.g. with -C 2 for two lines of context). With -n, grep outputs the line
number of the match. With -H, grep prints the file name for each match, which is useful when

.
you ―grep‖ a group of files or ―grep‖ recursively through a file system (e.g. with -r). Consider

pz
the output of grep --help for more options.

To grep a group of files, you can specify the file with a wildcard, as in the following example:

1 grep -i "morris" ~/org/*.txt

ee
This will find and match against every occurrence of the word ―morris,‖ while ignoring case
(because of the option for -i). grep will search all files in the ~/org/ directory with a .txt
extension.
ad
You can use grep to filter the results of another command that sends output to standard out (e.g.
stdout). This is accomplished by ―piping‖ the output of one command ―into grep.‖ For
instance:

1 ls /home/squire/data | grep "1257"

.p

In this example, we assume that the /home/squire/data directory contains a large number of
files that have a UNIX time stamp in their file name. The above command will filter the output
to only display those tiles that have the four digits ―1257‖ in their file name. Note, in these cases
w

grep only filters the output of ls and does not look into file contents. For more information
regarding grep consider the full documentation of the grep command.

How to Search and Replace Across a Group of Files

w

While the grep tool is quite powerful for filtering text on the basis of regular expressions, if you
need to edit a file or otherwise manipulate the text you may use the sed tool. sed, or the Stream
w

EDitor, allows you search for a regular expression pattern and replace it with another string.

sed is extremely powerful, and we recommend that you back up your files and test your sed
commands thoroughly before running them, particularly if you’re new to using sed. Here is a
very simple sed one-liner, intended to illustrate its syntax.

1 sed -i `s/^good/BAD/` morning-star.txt

www.padeepz.net
This replaces occurrences of the word ―good‖ occurring at the beginning ofwww.padeepz.net
a line (as noted by
the ^) with the string ―BAD‖ in the file morning-star.txt. The option -i tells sed to perform
the replacements ―in place.‖ sed can make backups of the files it edits if you specify a suffix
after the -i option, as in -iBAK. In the above command this option would save the original file as
morning-star.txt.BAK before making changes.

We can surmise that the format of a sed statement is:

1 's/[regex]/[replacement]/'

t
ne
To match literal slashes (e.g. /), you must escape them with a backslash (e.g. \). As a result, to
match a / character you would use \/ in the sed expression. If you are searching for a string that
has a number of slashes, you can replace the slashes which another character. For instance:

1 's|r/e/g/e/x|regex|'

.
This would strip the slashes from the string r/e/g/e/x so that this string would be regex after

pz
running the sed command on the file that contains the string.

The following example, from our migrating a server to your Linode document, searches and
replaces one IP address with another. In this case 98.76.54.32 is replaced with 12.34.56.78:
ee
1 sed -i 's/98\.76\.54\.32/12\.34\.56\.78/'

In the above example, period characters are escaped as \.. In regular expressions the full-stop
(period) character matches to any character.
ad
Once again, sed is a very powerful and useful tool; however, if you are unfamiliar with it, we
strongly recommend testing your search and replace patterns before making any edit of
consequence. For more information about sed consider the full documentation of text
manipulation with sed.
.p

How to Edit Text Interactively

In many Linode Library documents, you may be instructed to edit the contents of a file. To do
this, you need to use a text editor. Most of the distribution templates that Linode provides come
w

with an implementation of the vi/vim text editor and the nano text editor. These are small,
lightweight, and very powerful text editors that allow you manipulate the text of a file from the
terminal environment.
w

There are other options for text editors, notably emacs and ―zile.‖ Feel free to install these
programs using your operating system’s package manager. Make sure you search your package
database so that you can install a version compiled without GUI components (i.e. X11).
w

To open a file, simply issue a command beginning with the name of the editor you wish to run
followed by the name of the file you wish to edit. Here are a number of example commands that
open the /etc/hosts file:

1 nano /etc/hosts
2 vi /etc/hosts
3 emacs /etc/hosts
4 zile /etc/hosts
www.padeepz.net
When you’ve made edits to a file, you can save and exit the editor to returnwww.padeepz.net
to the prompt. This
procedure varies between different editors. In emacs and zile, the key sequence is the same:
depress control and type x and s to save. This operation is typically notated ―C-x C-s‖ and then
―C-x C-c‖ to close the editor. In nano, press Control-O (notated \^O) and confirm the file name
to write the file, and type \^X to exit from the program.

Since vi and vim are modal editors, their operation is a bit more complex. After opening a file in
vi, you can enter ―insert‖ mode by pressing the ―i‖ key; this will let you edit text in the
conventional manner. To save the file, you must exit into ―normal‖ mode by pressing the escape

t
key (Control-[ also sends escape), and then type :wq to write the file and quit the program.

ne
This provides only the most basic outline of how to use these text editors, and there are
numerous external resources which will provide a more thorough introduction for more advanced
use of this software.

Web Servers and HTTP Issues

.
pz
Linodes do not come with any particular web server installed by default. You have the choice
and power to install and configure your web server as you see fit. This allows you to deploy a
configuration in a way that makes sense for your application and desired use case. The Linode
Library contains a number of documents regarding the installation and maintenance of various
web servers.
ee
The following tips cover a number of basic web serving tasks and functions, as well as some
guidance for users new to the world of web servers.

How to Serve Websites

ad
Web servers work by listening on a TCP port, typically port 80 for ―http‖ and port 443 for
―https.‖ When a visitor makes a request for content, the servers respond by delivering the
resource requested. Typically resources are specified with a URL that contains the protocol,
http or https; a colon and two slashes, ://; hostname or domain, www.example.com or
squire.example.com; and the path to a file, /images/avatar.jpg, or index.html. Thus a full
.p

URL would resemble: http://www.example.com/images/avatar.jpg.

In order to provide these resources to connected users, your Linode needs to be running a web
server. There are multiple different HTTP servers and countless configurations to provide
w

support for various web development frameworks. The three most popular general use web
servers are the Apache HTTP server, Lighttpd server (―Lighty‖), and nginx server (―Engine X‖).
Each server has its strengths and weaknesses, and your choice depends largely on your
w

experience and the nature of your needs.

Once you’ve chosen a web server, you need to decide what (if any) scripting support you need to
install. Scripting support allows you to run dynamic content with your web server and program
w

server side scripts in languages such as Python, PHP, Ruby, and Perl.

If you need a full web application stack, we encourage you to consider one of our more full-
featured LAMP stack guides. If you need support for a specific web development framework,
consult our tutorials for installing and using specific web development frameworks.

How to Choose a Web Server

www.padeepz.net
In most situations, end users are completely oblivious to which web serverwww.padeepz.net
you use. As a result,
choosing a web server is often a highly personal decision based on the comfort of the
administrator and the requirements of the deployment in question. This can be a challenge for the
new systems administrator. This tip attempts to offer some guidance by providing some
background and information which might be helpful during this process.

The Apache HTTP Server is considered by many to be the de facto standard web server. It is the
most widely deployed open source web server, its configuration interface has been stable for
many years, and its modular architecture allows it to function in many different types of

t
deployments. Apache forms the foundation of the LAMP stack, and contains superb support for
integrating dynamic server-side applications into the web server.

ne
By contrast, web servers like Lighttpd and nginx are highly optimized for serving static content
in an efficient manner. If you have a deployment where server resources are limited and are
facing a great deal of demand, consider one of these servers. They are very functional and run
very well with minimal systems resources. Lighttpd and nginx can be more complex to set up

.
than Apache, and can be difficult to configure with regards to integration with dynamic content

pz
interpreters. Furthermore, as these servers are more directed at niche use cases, there are more
situations and applications which remain undocumented.

Finally the Cherokee web server provides a general purpose web server with an easy to configure
interface. Cherokee might be a good option for some basic deployments.
ee
Remember that the choice of web servers is often contextually determined. Specific choices
depend on factors like: the type of content you want to serve, the demand for that content, and
your comfort with that software as an administrator.

Reading Apache Logs

ad
Often, when there is something wrong with an Apache web sever configuration or site, it is
difficult to determine what the cause of the error is from the behavior of the web server. There
are a number of common issues with which you might begin your troubleshooting efforts.
However, when more complex issues arise it is important to review the Apache error logs.
.p

By default, error logs are located in the /var/log/apache2/error.log file. You can track or ―tail‖
this log with the following command:
w

1 tail -F /var/log/apache2/error.log

In the default virtual host configurations suggested in our Apache installation and LAMP guides,
we suggest the following error logging setup:
w

Apache Virtual Host Configuration

w

ErrorLog /srv/www/example.com/logs/error.log CustomLog

/srv/www/example.com/logs/access.log combined

Where bucknell.net represents the name of your virtual host, and the location of relevant files.
These configuration directives make Apache create two log files that contain logging information
specific to that virtual host. This allows you to easily troubleshoot errors on specific virtual
hosts. To track or tail the error log, issue the following command:

www.padeepz.net
1 tail -F /srv/www/example.com/logs/error.log www.padeepz.net

This will allow you to see new error messages as they appear. Often problems can be diagnosed
by using specific parts of an error message from an Apache log as a term in Web search (e.g.
Google.) Common errors to look for include:

 Missing files, or errors in file names.

 Permissions errors.
 Configuration errors.

t
 Dynamic code execution or interpretation errors.

ne
DNS Servers and Domain Names
The Domain Name System, or DNS, is the service that the Internet uses to associate the hard to
remember and manage IP addresses with more human-usable domain names. These tips address

.
several specific DNS related tasks. To learn more about DNS consider our overview of the
domain name system. If you are familiar with DNS and just need to figure out how to set up your

pz
DNS server, consider our documentation of the Linode DNS manager.

Using CNAMEs to Redirect DNS Queries

CNAME DNS records make it possible to redirect requests for one hostname or domain to
ee
another hostname or domain. This is useful in situations where you want to direct requests for
one domain to another, but don’t want to set up the web-server to handle requests.

CNAMEs are only valid when pointing from one domain to another. If you need to redirect a full
URL, you will need to set up a web server and configure redirection and/or virtual hosting on the
ad
server level. CNAMEs will allow you to redirect subdomains, such as team.example.com, to
other subdomains or domains, such as jack.example.org. CNAMEs must point a valid a
domain that has a valid A Record, or to another CNAME.

Although limited in their capabilities, CNAMEs can be quite useful in some situations. In
particular, if you need to change the hostname of a machine, CNAMEs are quite useful. To learn
.p

how to set up CNAME records with the Linode Manager, consult our documentation of the
Linode DNS Manager.

How to Set Up Subdomains

w

When reading domain names, we commonly refer to parts before the main or first-level domain
as ―sub-domains.‖ For example, in the domain team.example.com, team is a sub-domain for the
w

root domain example.com.

If you want to create and host a sub-domain, consider the following process:
w

First we need to create an A Record

in the DNS zone for the domain. This is easily accomplished when using the Linode DNS
Manager. As always, you may host the DNS for your domain with any provider you choose.

In order for your server to respond to requests for this domain, you must set up a server to
respond to these requests. For web servers like Apache this requires configuring a new virtual

www.padeepz.net
 www.padeepz.net
host. For XMPP Servers you must configure an additional host to receive the requests for this
host. For more information, consult the documentation for the specific server you wish to deploy.

Once configured, subdomains function identically to first-level domains on your server in almost
all respects. If you need to, you can set up HTTP redirection for the new sub domain.

SMTP Servers and Email Issues

We provide a number of guides that cover email-related issues. The following tips attempt to

t
further demystify email management.

ne
Choosing an Email Solution

There are two major components of the email stack that are typically required for basic email
functionality. The most important part of the tool chain is the SMTP server or ―Mail Transfer
Agent.‖ The MTA, as it is often called, sends mail from one server to another. The second

.
crucial part of an email system is a server that permits users to access and download that mail

pz
from the server to their own machine. Typically these server use a protocol such as POP3 or
IMAP to provide remote access to the mailbox.

There are additional components in the email server tool chain. These components may or may
not be optional depending on the requirements of your deployment. They include filtering and
ee
delivery tools like procmail, anti-virus filters like ClamAV, mailing list managers like MailMan,
and spam filters like SpamAssassin. These components function independently of which MTA
and remote mailbox accessing server you chose to deploy.

The most prevalent SMTP servers or MTAs in the UNIX-like world are Postfix, Exim, and
ad
Sendmail. Sendmail has the longest history and many systems administrators have extensive
experience with it. Postfix is robust and modern, and is compatible with many different
deployment types. Exim is the default MTA in Debian systems, and many consider it to be easier
to use for basic tasks. For remote mailbox access, servers like Courier and Dovecot are widely
deployed to provide remote access to mailboxes.
.p

If you are in need of an integrated and easy to install email solution we encourage you to
consider the Citadel groupware server. Citadel provides an integrated ―turnkey‖ solution that
includes an SMTP server, remote mailbox access, real time collaboration tools including XMPP,
and a shared calendar interface. Along similar lines, we also provide documentation for the
w

installation of the Zimbra groupware server.

If, by contrast, you want a more simple and modular email stack, we urge you to consider one of
w

our guides built around the Postfix SMTP server.

Finally, it’s possible to outsource email service to a third party provider, such as Google Apps or
FastMail.fm. This allows you to send and receive mail from your domain, without hosting email
w

services on your Linode. Consult our documentation for setting up Google Apps for your
domain.

Sending Email From Your Server

In many cases, administrators have no need for a complete email stack like those documented in
our email guides. However, applications running on that server still need to be able to send mail
for notifications and other routine purposes.
www.padeepz.net
The configuration of applications to send notifications and alerts is beyondwww.padeepz.net
the scope of this tip,
most applications rely on a simple ―sendmail‖ interface. Nevertheless, the modern MTAs Postfix
provides a sendmail-compatible interfaces located at /usr/sbin/sendmail.

You can install postfix on Debian and Ubuntu systems with the following command:

1 apt-get install postfix

On CentOS and Fedora systems you can install postfix by issuing the following command:

t
ne
1 yum install postfix

Once Postfix is installed, your applications should be able to access the sendmail interface,
located at /usr/sbin/sendmail. Most applications running on your Linode should be able to
send mail normally with this configuration.

.
If you simply want to use your server to send email through an external SMTP server, you may

pz
want to consider a more simple tool like msmtp. Since msmtp is packaged in most distributions
you can install using the command appropriate to your distribution:

1 apt-get install msmtp

2 yum install msmtp
ee
3 pacman -S msmtp

Use the command type msmtp, to find the location of msmtp on your system. Typically the
program is located at /usr/bin/msmtp. You can specify authentication credentials with
command line arguments or by declaring SMTP credentials in a configuration file. Here is an
example .msmtprc file.
ad
.msmtprc example

account default host smtp.example.com from squire@example.com auth on user squire password
s3cr37 tls on tls_certcheck off port 587
.p

The .msmptrc file needs to be set to mode 600, and owned by the user account that will be
sending mail. If the configuration file is located at /srv/smtp/msmtprc, you can call mstmp
with the following command:
w

1 /usr/bin/msmtp --file=/srv/smtp/msmtprc
w

VIRTUALIZATION

Virtualization projects are the focus of many IT professionals who are trying to consolidate
w

servers or data centers, decrease costs and launch successful ―green‖ conservation initiatives.
Virtualizing IT resources can be thought of as squeezing an enterprise’s computer processing
power, memory, network bandwidth and storage capacity onto the smallest number of hardware
platforms possible and then apportioning those resources to operating systems and applications
on a time-sharing basis. This approach aims to make the most efficient possible use of IT
resources. It differs from historical computing and networking models, which have typically
involved inextricably binding a given software
www.padeepz.net
 www.padeepz.net
application or service to a specific operating system (OS), which, in turn, has been developed to
run on a particular hardware platform. By contrast, virtualization decouples these components,
making them available from a common resource pool. In this respect, virtualization prevents IT
departments from having to worry about the particular
hardware or software platforms installed as they deploy additional services. The decoupling and
optimization of these components is possible whether you are virtualizing servers, desktops,
applications, storage devices or networks. To virtualize some or all of a computing
infrastructure’s resources, IT departments require special virtualization software, firmware or a
third- party service that makes use of virtualization software or firmware. This

t
software/firmware component, called the hypervisor or the virtualizattion layer, performs the
mapping between virtual and physical resouces. It is what enables the various resources to be

ne
decoupled,then aggregated and dispensed, irrespective of the underlying hardware and, in some
cases, the software OS. In effect, the hypervisor takes over hardware management from the OS.
In addition to the hypervisor virtualization technology, the organization overseeing the
virtualization project requires a virtualization management tool – which might be procured from
the same or a different supplier – to set up and manage virtual devices and policies.

.
Why Virtualize?

pz
One key reason why IT organizations are considering virtualization of some or all of their
computing infrastructures is that the technology helps them to derive the biggest bang out of
their computing buck.

SETTING UP XEN
ee
Xen is a type 1, bare-metal virtual machine monitor (or hypervisor), which provides the ability to
run one or more operating system instances on the same physical machine. Xen, like other types
of virtualization, is useful for many use cases such as server consolidation and isolation of
production and development environments (Eg.: corporate and personal environments on the
same system).
ad
As of Ubuntu 11.10 (Oneiric), the default kernel included in Ubuntu can be used directly with
the Xen hypervisor as the management (or control) domain (dom0 or "Domain0" in Xen
terminology).
.p

Our example uses LVM for virtual disks and network bridging for virtual network cards. It also
assumes Xen 4.1 (the version available in 12.04). It assumes a familiarity with general
virtualization issues, as well as with the specific Xen terminology. Please see the Xen wiki (see
http://wiki.xen.org/wiki/Xen_Overview) for more information.
w

During Installation of Ubuntu

 Note: For VMware Workstation test users:

w

On VM settings enable "Virtualize Intel VT-x/EPT or AMD-V/RVI"

During the install of Ubuntu for the Partitioning method choose "Guided - use the entire disk and
setup LVM". Then, when prompted to enter "Amount of volume group to use for guided
w

partitioning" enter a value large enough for the Xen dom0 system, leaving the rest for virtual
disks. Enter a value smaller than the size of your installation drive. For example 100 GB should
be large enough for a minimal Xen dom0 system. Keep in mind that in our model stay inside that
guest (dom0) all installation media for guest OSs and other useful files, so that guest must have
enough space on it.

www.padeepz.net
 www.padeepz.net
After Installation of Ubuntu

Install GUI
sudo apt-get update
sudo apt-get install ubuntu-desktop

To skip the login screen completely, boot into the console and then start the GUI

t
sudo gedit /etc/default/grub

ne
Change line GRUB_CMDLINE_LINUX_DEFAULT="quiet splash" to
GRUB_CMDLINE_LINUX_DEFAULT="text".

Save and exit.

.
sudo update-grub

pz
Reboot and you should come up directly in tty1.

Login, and then "startx" to boot into the default desktop.

ee
To logoff "Unity" (default Ubuntu desktop) from command line type "gnome-session-quit".

Install Windows Remote Desktop

We understand that this is the best approach to have remote access to "Ubuntu Server 12.04
LTS" into a cross-plataform environment. The "XRDP" is an implementation of the "Remote
ad
Desktop" standards from Microsoft and works on the same way as for Windows. Allows remote
desktop access via native Windows client machines (or "RDESKTOP" on Ubuntu), does not
require loading the "Ubuntu Server 12.04 LTS" GUI (Graphical User Interface) on boot and
allows multiple simultaneous sessions. To use "RDESKTOP" on Ubuntu 12.04 LTS with
"TSCLIENT" see my post on http://superuser.com/questions/420291/ubuntu-12-04-how-to-get-
.p

tsclient-back.

With "XRDP" you can easily use Microsoft RDP to connect to Ubuntu without any
configuration. All you need to do is install the "xrdp" package, then open Remote Desktop
w

Connection from Windows and connect. That's it, nothing to configure.

Without wasting anymore of your time, let’s get going.

w

 sudo apt-get install xrdp

Next, open Windows Remote Desktop Connection (RDP) and type Ubuntu Server hostname or
w

IP address.

Off Course You Can Use the Great SSH

As you may already know, SSH is a secure communication protocol that lets you remotely
access networked computers. It is known as a replacement for Telnet which is very unsecure.
While Telnet sends traffic in plain text, SSH on the other hand uses a secure protocol to
communicate.
www.padeepz.net
Run the commands below to install SSH Server. www.padeepz.net

 sudo apt-get install openssh-server

To log in on remote machine type on your terminal

 ssh <remote_user>@<ip_or_name>

Installing Xen (XCP - Xen Cloud Platform)

t
ne
XCP (Xen Cloud Platform) is the open source version similar to Citrix Xen Server that uses the
Xen Hypervisor. XCP uses XAPI or XenAPI to manage Xen hosts. XCP is based on CentOS 5.5.

Project Kronos is an initiative to port the XAPI tool stack to Debian and Ubuntu. It is a
management stack implemented in OCaml that configures and controls Xen hosts, attached
storage, networking and virtual machine life cycle. It exposes a HTTP API and provides a

.
command line interface (xe) for resource management.

pz
XenCenter is Windows desktop application by Citrix that is distributed with XenServer for
managing servers running XenServer (the equivalent of linux is OpenXenManager). It uses
XAPI for talking to Xen resource pools. Since we are setting up XAPI, we can use XenCenter to
manage the server.
ee
Why use XCP-XAPI on Debian/Ubuntu when XCP appliance exists?

 Manage dom0 using a configuration management framework (Puppet, Chef...)

 Apply security updates to dom0 root file system
 Run Xen version 4.1.
ad
 Ubuntu Server 12.04 is a LTS release that is supported for 5 years

Installing and configuring Xen Hypervisor

 Install the Xen Hypervisor

.p

o sudo apt-get install xen-hypervisor

Setup GRUB to boot the Xen Hypervisor

w

o sudo sed -i 's/GRUB_DEFAULT=.*\+/GRUB_DEFAULT="Xen 4.1-

amd64"/' /etc/default/grub

Disable apparmor at boot

w

o sudo sed -i
's/GRUB_CMDLINE_LINUX=.*\+/GRUB_CMDLINE_LINUX="apparmor=0"/'
/etc/default/grub
w

Restrict "dom0" to 1GB of memory and 1 VCPU (example)

sudo gedit /etc/default/grub

After GRUB_CMDLINE_LINUX="apparmor=0" add the line

GRUB_CMDLINE_XEN="dom0_mem=1G,max:1G dom0_max_vcpus=1"

www.padeepz.net
 Update Grub with the config changes we just made www.padeepz.net

o sudo update-grub

Reboot the server so that Xen boots on the server

o sudo reboot

Once the server is back online ensure that Xen is running

t
o cat /proc/xen/capabilities should display "control_d"

ne
o Note: To stop or start xcp-xapi
sudo /etc/init.d/xcp-xapi stop (or start)

Installing and configuring XAPI (XenAPI)

.
o Install XCP-XAPI
 sudo apt-get install xcp-xapi

pz
-> Choose "bridge" when prompted for network backend

o Setup the default toolstack

 sudo gedit /etc/default/xen
ee -> Set "TOOLSTACK=xapi"

o Disable xend from starting at boot

 sudo sed -i -e 's/xend_start$/#xend_start/' -e
's/xend_stop$/#xend_stop/' /etc/init.d/xend
ad
Note: Only xend the deamon needs to be disabled from starting, "/etc/init.d/xend"
handles other things like modules and xenfs. Do not disable it from the runlevel.
Disable service xendomains

 sudo update-rc.d xendomains disable

.p

Fix for "qemu" which emulates the console does not have the keymaps in the correct
location
w

 sudo mkdir /usr/share/qemu; sudo ln -s /usr/share/qemu-

linaro/keymaps /usr/share/qemu/keymaps
w

Network configuration

 This section describes how to set up Linux bridging in Xen. It assumes eth0 is
both your primary interface to dom0 and the interface you want your VMs to
w

use. It also assumes that you will use manually IP configuration.

 sudo apt-get install bridge-utils

Note: If not already installed

 As you are working with a desktop install, disable "Network Manager"

 sudo stop network-manager
sudo gedit /etc/NetworkManager/NetworkManager.conf
www.padeepz.net
  www.padeepz.net
Now make sure the "managed" line contains "managed=false"
Save and exit

Since Network Manager is not managing your network interfaces

anymore, you must manually enter that information. Below we explain
how.

 Setup bridge networking

 sudo gedit /etc/network/interfaces

t
 Create a bond called xenbr0. The file should look like this for a
static network configuration:

ne
 # This file describes the network
interfaces available on your system
 # and how to activate them. For more
information, see interfaces(5).

 # The loopback network interface

.
 auto lo

pz
 iface lo inet loopback


 # Xen network interface for "dom0"
 auto xenbr0
 iface xenbr0 inet static
ee 



# IP address
address 192.168.1.111
# Subnet mask
 netmask 255.255.255.0
 # Default Gateway
 gateway 192.168.1.1
ad
 # DNS Server
 dns-nameservers 192.168.1.1

 bridge_ports eth0
 iface eth0 inet manual

.p


 # The primary network interface
 # auto eth0
 # iface eth0 inet dhcp
w

Configure xcp to use "bridge" networking instead of "openswitch"

 sudo gedit /etc/xcp/network.conf

w

-> Replace ―openswitch‖ with ―bridge‖

Note: For VMware Workstation test users:

w

- Configure with "bridge" for network adapter

and run these comands on host:

sudo chmod ugo+rwx /dev/vmnet0

sudo chown <username> /dev/vmnet0
sudo chown :<usergroup> /dev/vmnet0

Eg.:
sudo chmod ugo+rwx
/dev/vmnet0
www.padeepz.net
 www.padeepz.net
sudo chown eduardo
/dev/vmnet0
sudo chown :eduardo
/dev/vmnet0

To test whether the network is working run the command

 sudo /etc/init.d/networking restart

All set! Ready to reboot and let xcp-xapi toolstack take over

t
ne
 sudo reboot

On restart confirm that xcp is working

 sudo xe vm-list
 This should list the control domain

.
 "

pz
 uuid (RO) : dbcf74d2-ee50-edd5-d44d-
b81fc8ba1777
 name-label (RW): Control domain on host:
ubuntu-xenserver-1
 power-state (RO): running
 "
ee -> If your output looks similar, "xapi" is running on the server, if
you get ―Connection refused‖ then xapi is not setup correctly!

Configure Storage Repository for Use With XAPI

ad
 NFS servers are a common form of shared filesystem infrastructure, and
can be used as a storage repository substrate for virtual disks. As NFS
storage repositories are shared, the virtual disks stored in them allow VMs
to be started on any server in a resource pool and to be migrated between
them using XenMotion. When you configure an NFS storage repository,
.p

you simply provide the hostname or IP address of the NFS server and the
path to a directory that will be used to contain the storage repository (if
this resource is on another machine, it is not our case). The NFS server
must be configured to export the specified path to all servers in the pool.
w

 To show your volume group (VG)

 sudo pvs

Create a LV with X GBs

w

 sudo lvcreate -L <X>GB -n

<StorageRepositoryName> /dev/<VG>
 Eg1.: sudo lvcreate -L 25GB -n
w

StorageRepository /dev/ubuntus1204
Eg2.: sudo lvcreate -l 100%FREE -n
StorageRepository /dev/ubuntus1204

Register the logical volume for use with XAPI

 sudo xe sr-create type=ext shared=true name-

label=<StorageRepositoryName> device-
config:device=/dev/<VG>/<StorageRepositoryName>
www.padeepz.net
 www.padeepz.net
Eg.: sudo xe sr-create type=ext shared=true
name-label=StorageRepository device-
config:device=/dev/ubuntus1204/StorageRepositor
y

This should display the Storage Repository

 sudo xe sr-list name-

label=<StorageRepositoryName>
Eg.: sudo xe sr-list name-

t
label=StorageRepository
 "

ne
 uuid ( RO): 37bc5263-c9fc-8876-d24c-
d5927f1bbed2
 name-label ( RW): StorageRepository
 name-description ( RW):
 host ( RO): ubuntus1204
 type ( RO): ext

.
 content-type ( RO):
 "

pz
Configure a ISO Repository for Use With XAPI

An ISO Repository contains ISOs (disk images) with operational systems to perform the installations.
ee
Then the following example makes a storage repository called ISOs

sudo xe sr-create name-label=<LocalISORepositoryName> type=iso shared=true

device-config:location=<FolderPath> device-config:legacy_mode=true content-
type=iso
Eg.:
ad
sudo mkdir -p /var/opt/xen/LocalISORepository/
sudo xe sr-create name-label=LocalISORepository type=iso shared=true device-
config:location=/var/opt/xen/LocalISORepository/ device-
config:legacy_mode=true content-type=iso

This should display the ISO Repository

.p

sudo xe sr-list name-label=<LocalISORepositoryName>

Eg.: sudo xe sr-list name-label=LocalISORepository
"
uuid ( RO): 26edb27b-72fc-af56-ad2f-4d15a8d8e3f7
w

name-label ( RW): LocalISORepository

name-description ( RW):
host ( RO): ubuntus1204
type ( RO): iso
w

content-type ( RO): iso

"
w

www.padeepz.net
 www.padeepz.net
VMWARE ON LINUX

VMware Workstation 5.5 Host System Requirements

What do you need to get the most out of VMware Workstation 5? Take the following list of
requirements as a starting point. Like physical computers, the virtual machines running under
VMware Workstation generally perform better if they have faster processors and more memory.

t
PC Hardware

ne
 Standard x86-compatible or x86-64-compatible personal computer

 400 MHz or faster CPU minimum (500 MHz recommended)

Compatible processors include

.
pz
 Intel�: Celeron�, Pentium� II, Pentium III, Pentium 4, Pentium M (including computers
with Centrino™ mobile technology), Xeon™ (including "Prestonia"), EM64T

 AMD™: Athlon™, Athlon MP, Athlon XP, Athlon 64, Duron™, Opteron™, Turion™ 64

 Experimental support for AMD Sempron™

ee
For additional information, including notes on processors that are not compatible, see the
VMware knowledge base at
www.vmware.com/support/kb/enduser/std_adp.php?p_faqid=967.
ad
 Multiprocessor systems supported

 AMD Opteron, AMD Athlon 64, AMD Turion 64, AMD Sempron, Intel EM64T; support for
64-bit guest operating systems is available only on the following specific versions of these
processors:
.p

 AMD Athlon 64, revision D or later

 AMD Opteron, revision E or later

w

 AMD Turion 64, revision E or later

 AMD Sempron, 64-bit-capable revision D or later (experimental support)

w

 Intel EM64T VT-capable processors (experimental support)

Memory
w

 128 MB minimum (256 MB recommended)

You must have enough memory to run the host operating system, plus the memory required for
each guest operating system and for applications on the host and guest. See your guest operating
system and application documentation for their memory requirements.

www.padeepz.net
Display www.padeepz.net

 16-bit or 32-bit display adapter recommended

Disk Drives

Guest operating systems can reside on physical disk partitions or in virtual disk files.

Hard Disk

t
 IDE and SCSI hard drives supported, up to 950GB capacity

ne
 At least 1GB free disk space recommended for each guest operating system and the
application software used with it; if you use a default setup, the actual disk space needs are
approximately the same as those for installing and running the guest operating system and

.
applications on a physical computer.

pz
 For Installation — 80MB (Linux) or 250MB (Windows) free disk space required for basic
installation. You can delete the installer afterwards to reclaim disk space.

Optical CD-ROM/DVD-ROM Drive

ee
 IDE and SCSI optical drives supported

 CD-ROM and DVD-ROM drives supported

 ISO disk image files supported

ad
Local Area Networking (Optional)

 Any Ethernet controller supported by the host operating system

 Non-Ethernet networks supported using built-in network address translation (NAT) or using a
.p

combination of host-only networking plus routing software on the host operating system

Host Operating System

w

VMware Workstation is available for both Windows and Linux host operating systems.

Windows Host Operating Systems (32-Bit)

w

Workstation supports the following Windows 32-bit host operating systems.

 Windows Server 2003 Standard Edition, SP1

w

Windows Server 2003 Web Edition, SP1

Windows Server 2003 Small Business Edition, SP1
Windows Server 2003 Enterprise Edition, SP1
Windows Server 2003 R2
(Listed versions are also supported with no service pack.)

 Windows XP Home Edition, SP1, SP2

www.padeepz.net
Windows XP Professional, SP1, SP2 www.padeepz.net
(Listed versions are also supported with no service pack.)

 Windows 2000 Server SP3, SP4

Windows 2000 Professional, SP3, SP4
Windows 2000 Advanced Server, SP3, SP4

Windows Host Operating Systems (64-Bit)

t
 Windows Server 2003 x64 Edition SP1

ne
Windows Server 2003 x64 Edition R2

 Windows XP Professional x64 Edition

Internet Explorer 4.0 or higher is required for the Windows online help system.

.
Linux Host Operating Systems (32-Bit)

pz
Supported distributions and kernels are listed below. VMware Workstation may not run on
systems that do not meet these requirements.

Note: As newer Linux kernels and distributions are released, VMware modifies and tests its
ee
products for stability and reliability on those host platforms. We make every effort to add support
for new kernels and distributions in a timely manner, but until a kernel or distribution is added to
the list below, its use with our products is not supported. Look for newer prebuilt modules in the
download area of our Web site. Go to www.vmware.com/download/.
ad
 Mandriva Corporate Server 4
Mandriva Linux 2007 (experimental support)
Mandriva Linux 2006
Mandrake Linux 10.1
Mandrake Linux 9.0 — stock 2.4.19
.p

 Red Hat Enterprise Linux 5.0, AS, ES, WS (experimental support)

Red Hat Enterprise Linux 4.0, AS, ES, WS, updates 1, 2, 3, 4
Red Hat Enterprise Linux 3.0, AS, ES, WS, updates 1, 2, 3, 4, 5, 6, 7, 8
w

Red Hat Enterprise Linux 2.1 — stock 2.4.9-e3

Red Hat Linux 9.0 — stock 2.4.20-8, upgrade 2.4.20-20.9

w

Red Hat Linux 8.0 — stock 2.4.18

Red Hat Linux 7.3 — stock 2.4.18
Red Hat Linux 7.2 — stock 2.4.7-10, upgrade 2.4.9-7, upgrade 2.4.9-13, upgrade 2.4.9-21,
w

upgrade 2.4.9-31
Red Hat Linux 7.1 — stock 2.4.2-2, upgrade 2.4.3-12
Red Hat Linux 7.0 — stock 2.2.16-22, upgrade 2.2.17-14

 SUSE Linux Enterprise Server 10

SUSE Linux Enterprise Server 9, 9 SP1, 9 SP2, 9 SP3 — stock 2.6.5-797
(Listed versions are also supported with no service pack.)
SUSE Linux Enterprise Server 8, stock 2.4.19
www.padeepz.net
 www.padeepz.net
SUSE Linux 10.1
SUSE Linux 10 — stock 2.6.13
SUSE Linux 9.3 — stock 9.3-2.6.11.4
SUSE Linux 9.2, SP1 — stock 9.2-2.6.8-24.11 (Listed versions are also supported with no
service pack.)
SUSE Linux 9.1 — stock 2.6.4-52
SUSE Linux 9.0 — stock 2.4.21-99

t
SUSE Linux 8.2 — stock 2.4.20

ne
 Ubuntu Linux 6.10 (experimental support)
Ubuntu Linux 6.06
Ubuntu Linux 5.10
Ubuntu Linux 5.04

.
Platforms not listed above are not supported. A Web browser is required for the Help system.

pz
Linux Host Operating Systems (64-Bit)

Supported distributions and kernels are listed below. VMware Workstation may not run on
systems that do not meet these requirements.
ee
Note: As newer Linux kernels and distributions are released, VMware modifies and tests its
products for stability and reliability on those host platforms. We make every effort to add support
for new kernels and distributions in a timely manner, but until a kernel or distribution is added to
the list below, its use with our products is not supported. Look for newer prebuilt modules in the
ad
download area of our Web site. Go to www.vmware.com/download/.

 Mandriva Corporate Server 4

Mandriva Linux 2007 (experimental support)
Mandriva Linux 2006
.p

 Red Hat Enterprise Linux 5.0, AS, ES, WS (experimental support)

Red Hat Enterprise Linux 4.0, AS, ES, WS, updates 1, 2, 3, 4
Red Hat Enterprise Linux 3.0, AS, ES, WS stock 2.4.21, updates 2.4.21-15, 6, 7, 8
w

SUSE Linux Enterprise Server 10

SUSE Linux Enterprise Server 9, SP1, SP2, SP3 — stock 2.6.5-797
w

(Listed versions are also supported with no service pack.)

SUSE Linux 10.1

SUSE Linux 10 — stock 2.6.13
w

SUSE Linux 9.3 — stock 9.3-2.6.11.4

SUSE Linux 9.2, SP1 — stock 9.2-2.6.8-24.11 (Listed versions are also supported with no
service pack.)
SUSE Linux 9.1 — stock 2.6.4-52

 Ubuntu Linux 6.10 (experimental support)

Ubuntu Linux 6.06
www.padeepz.net
Ubuntu Linux 5.10 www.padeepz.net
Ubuntu Linux 5.04

Platforms not listed above are not supported. A Web browser is required for the Help system.

t
. ne
pz
ee
ad
.p
w
w
w

www.padeepz.net