CSC220 Lab 1 2010

You can work in pairs on this lab if you wish. Otherwise work individually.

Login

• Login to a Linux box around you using your CSC220a account.
• edit the file .bashrc that should already be in your directory:

 emacs -nw .bashrc

• add the following two lines at the end of the file:

PS1='[\h]\n[\t] \w\$: '
PS2='> '

• Save the file.
• By default, the shell into which you start is the tcsh. Switch to bash by typing the following command at the prompt:

 bash

• Verify that your prompt has changed and looks something like this:

[beowulf]
[09:52:24] ~$:

• You are now ready to explore bash commands...

Path-Related Questions

Question 1

What is the path of your account? In other words, what subdirectories does one have to traverse to reach your account.

Question 2

What other subdirectories are at the same height as yours in the directory tree?

You will notice that all user accounts are in a major directory called Users. The old standard for users on a linux system is to have the users in a directory called home in the root directory.

Question 3

You will notice that some user accounts still exist in home. Which are they?

Questions about File-Searching

With Linux, the name of the printers supported are listed in a file called /etc/printcap. Look at its contents.

Question 1

Figure out a way to get only the entries from this file that represent printers that are in Ford Hall.

Filtering Log Files

Open a browser window and load up the following URL: http://maven.smith.edu/~hadoop/log.txt

Notice that it is a long log of the output of a research program I have been running recently. It is very long and contains a lot of information: some useful, some not.

You are going to get a copy of this file into your account. Instead of copying and pasting the text into a file, you are going to use a useful utility called wget. Wget is a Linux utility that allows you to grab Web pages from Web sites, without using a browser.

Try it:

 wget http://maven.smith.edu/~hadoop/log.txt

Check that the file is in your directory

Question 1

How many lines of text does the file contain?

Question 2

The second question is to list only the lines that list the real execution time. These lines look like this:

real	23m6.777s

Go ahead, list the real execution times. What is the shortest time recorded? The longest?

Question 3

List not only the real execution times, but also the lines of the form:

processing noTasks = 17240  maxNoTasks = 8,  splitSize = 33554432L

The output should look something like this:

processing noTasks = 862  maxNoTasks = 8  splitSize = 33554432L
real	22m11.284s
processing noTasks = 862  maxNoTasks = 16  splitSize = 33554432L
real	0m13.113s
processing noTasks = 1724  maxNoTasks = 8  splitSize = 33554432L
real	0m10.891s
processing noTasks = 1724  maxNoTasks = 16  splitSize = 33554432L
real	0m40.891s
processing noTasks = 80  maxNoTasks = 8  splitSize = 33554432L
real	2m54.601s

Pipes

Question 1

How many users have accounts on grendel (or beowulf)? (Hints: use ls and wc and a pipe)

Question 2

Who are the three users who have modified their account most recently? (Hints: remember the -ltr switches for ls, and pipe the output to the appropriate command that will give you only three lines)

Question 3

Same question, but the three who haven't modified (or touched) their account for the longest time?

Question 4

Who are the users who have modified their account today? (Hints: today is a unique character pattern... "Sep 7" )

More challenging pipes

• Use emacs to create the following program in your 220a account: stdouterr.py
• Make your program executable

chmod +x stdouterr.py

• run your program

./stdouterr.py

observe the long output. If you observe closely the listing, you will discover that some lines contain error codes, of the form

Error 404: blue screen alert!

Question 1

run the program and filter its output so that you see only the lines containing error messages

Question 2 (tricky)

How many lines are output by the program?

Question 3 (trickier)

run the program and use commands that will display only the number of Error messages