Speaking UNIX, Part 3: Do everything right from the command line

It's a wired, wired, wired world

Discover three essential UNIX® utilities that deliver the entire Internet to your command line.

Martin Streicher (martin.streicher@gmail.com), Chief Technology Officer, McClatchy Interactive

Photo of Martin StreicherMartin Streicher is a freelance Ruby on Rails developer and the former Editor-in-Chief of Linux Magazine. Martin holds a Masters of Science degree in computer science from Purdue University and has programmed UNIX-like systems since 1986. He collects art and toys. You can reach Martin at martin.streicher@gmail.com.



05 September 2006

Also available in Chinese Russian

The UNIX® command line is a WYTIWYG interface -- that is, What you type is what you get. UNIX provides hundreds, if not thousands, of commands with which you can manipulate a large variety of resources available in the kernel and user space. Need to monitor CPU usage? Try top or ps. Need to remove all files ending in .bak? Try rm *.bak. Want help with a new command? Run man.

But what do you do when the resources you need reside on a remote system on your wide-area network (WAN) and on the global Internet? To quote The Hitchhiker's Guide to the Galaxy, "don't panic!". The UNIX command line readily downloads and uploads files, connects to remote computers, and interrogates the state of far-flung servers and networks. Grab your towel: It's time for a trip to extra-solar systems.

Work locally, transfer globally

In Part 1 and Part 2 of this series, you learned how much you can accomplish with the UNIX command line. With just a few keystrokes -- including a pipe (|) or redirection -- you can create an impromptu data processing machine more powerful than the sum of its parts.

While some of the resources you use daily are likely to be local -- that is, resident on your own workstation -- a significant and growing number of assets -- files, e-mail messages, and tools -- are likely to be stored at a distance (say, on a machine connected to your WAN or to the Internet). Web browsers provide almost universal access to such resources, with one caveat: Point-and-click can quickly become tedious, even onerous, especially if you must retrieve more than one handful of items. Moreover, if you want to script -- essentially, capture and replay -- repetitive or error-prone tasks, a windowing browser is a difficult ally.

Much like ls, cp, mail, uptime, du manage, and query local resources, UNIX has a suite of command-line tools to access remote resources, too. This article introduces you to a few of those tools, including a useful technique that both facilitates access to remote systems and protects your authentication credentials. Specifically, you'll learn wget, curl, and Secure shell (ssh). The wget and curl tools transfer files; using ssh, you can securely log in to remote systems and transfer files quickly and easily.

The trouble with Telnet (and others)

If any of your systems run rsh (or its variants -- rcp, rexec, rlogin, or Irdist) or telnet, disable and remove them and the accompanying daemons immediately. In addition, if you don't permit anonymous File Transfer Protocol (FTP), disable the FTP software, too.

Exploits with rsh and FTP

For more information and demonstrations of the exploits possible with rsh and FTP, see "Security Vulnerability between FTP and Berkeley Rsh/Rlogin Protocols" and "Stupid, Stupid Protocols: Telnet, FTP, rsh/rcp/rlogin."

Although rsh and telnet are longtime UNIX stalwarts, attackers can leverage either utility to (easily) compromise your system. You or your system administrator should halt and remove this software wherever it's found running and replace the features of those packages with ssh.

For privileged FTP access, use sftp. Replace rdist with the more advanced rsync. Or, if you must provide anonymous FTP (or downloads over HTTP), be sure to use firewall hardware and software to isolate all publicly accessible computers from sensitive internal servers.

But first, let's discuss the pesky problems that passwords present.

"You don't need no stinkin' passwords!"

Access to most computers and services is typically protected. In some cases, authenticating your identity (and hence your privilege to access the system) might require a complex challenge-response exchange, a Secure Sockets Layer (SSL) certificate, or even a biometric scan. Typically, however, a password suffices to gain access. Much like your personal identification number (PIN), your password is your secret; if you choose your password well, it's likely difficult for others to guess it at random. The combination of your name and a strong password provide sufficient corroboration.

Of course, strong passwords can be difficult to remember, and the strain only worsens as you collect and memorize yet another eight-character key (of, say, numbers, punctuation, and mixed case). Typing a password over and over again can be downright annoying -- worse, it presents a significant obstacle to hands-off automation.

Recognizing this encumbrance, many command-line utilities allow you to provide your username and password as command-line arguments. For example, you can log in to an FTP site without intervention by using a command such as:

ftp ftp://joe:passwd@www.example.com

However, using such a facility can reveal your credentials to other users sharing your computer. (Try ps -Aeww, for example, to see the complete command line and environment of every process on the system.)

To provide the same convenience as command-line options without the inherent risks, many programs can read your credentials from a special file called a .netrc (pronounced net-r-c) file, which typically resides in ~/.netrc. Your .netrc file must be owner read-write mode (mode 0600 or -rw-------) only, and each entry in the file must adhere to this simple syntax:

machine ftp.example.com login zaphod password I()Trillian!

machine www.magazine.com login abner password MmG8y*tr

default login anonymous password zaphod@heartofgold.com

The first two lines provide the machine keyword and the computer's domain name, the login keyword and your login name on the computer, and the password keyword followed by the password associated with your login. The credentials on the last line provide a default for any system not specifically named. The default line must be the last line in your .netrc file. (For the full extent of .netrc file configuration options, type man 5 netrc to see the .netrc man page.)

Obviously, if any file has any identity data in it, secure it with user read-write only mode (mode 600) or user read-only mode (mode 400) to prevent you from accidentally overwriting or removing it. You might also want to protect your home directory with mode 700.

Now, whenever you launch a .netrc-enabled application, including those applications I discuss next, the appropriate login name and password are passed automatically to the desired service, with nary a single peck at the keyboard. You can typically disable this auto-login feature with the -n option.

The process of transference

Along with HTTP and HTTP over SSL (HTTPS) for Web pages, FTP is one of the most often used Internet application protocols. Through FTP, a client can connect to a server, acquire a list of directories and files, and either download a file (that is, request a file from the server) or upload a file (that is, send the file a server to persist). URLs of the forms ftp://ftp.example.com/path/to/anotherfile.zip and ftp://user:password@ftp.example.com/path/to/file.zip imply, using the FTP protocol, connect to ftp.example.com and download the file /path/to/anotherfile.zip. The latter URL simply adds credentials for login.

On most desktop computers, such URLs launch the browser or the default FTP application to download the specified file. However, you can use the same URLs with the wget command-line utility -- a robust utility for downloading files over HTTP, HTTPS, and FTP. It supports .netrc files and is entirely non-interactive, making it ideal for automation. If your system does not have wget, you can download its source code from the GNU Software Foundation. It builds readily on all UNIX variants with just a few simple commands, and you can place the utility in your personal bin directory or in a central directory.

Assuming that you have a .netrc file in place, let's look at some examples of what wget can do. (In the examples below, the line number is provided for reference; you don't need to type the numbers.) Listing 1 shows how to use wget to download files without leaving the comfort of the command line.

Listing 1. Using wget to download files at the command line
1 $ wget http://ftp.gnu.org/pub/gnu/wget/wget-1.10.2.tar.gz
--16:02:29--  http://ftp.gnu.org/pub/gnu/wget/wget-1.10.2.tar.gz
           => `wget-1.10.2.tar.gz'
Resolving ftp.gnu.org... 199.232.41.7
Connecting to ftp.gnu.org[199.232.41.7]:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1,213,056 [application/x-tar]

100%[=====================>] 1,213,056    531.22K/s             

16:02:37 (529.57 KB/s) - `wget-1.10.2.tar.gz' saved [1213056/1213056]

2 $ wget -q ftp://mirror.linux.duke.edu/pub/centos/4.3/os\
  /i386/RELEASE-NOTES-en.html

3 $ cat url_list.txt
http://www.wikipedia.com
http://valdez.barebones.com/pub/freeware/TextWrangler_2.1.3.dmg

4 $ wget -i -nv url_list.txt
16:06:00 URL:http://www.wikipedia.org/ [33606] -> "index.html" [1]
16:06:41 URL:http://valdez.barebones.com/pub/freeware/
  TextWrangler_2.1.3.dmg [9488296/9488296] -> 
  "TextWrangler_2.1.3.dmg" [1]
FINISHED --16:06:41--
Downloaded: 9,521,902 bytes in 2 files

5 $ ls
RELEASE-NOTES-en.html   index.html              wget-1.10.2.tar.gz
TextWrangler_2.1.3.dmg  url_list.txt

Command 1 downloads the most recent wget source code from its project home page over HTTP. By default, wget apprises you of progress. You can disable all messages with the -q (for quiet mode) option. Command 2 retrieves a version of the CentOS release notes over FTP, albeit very quietly.

Keeping URLs intact

Here's a tip: Many HTTP URLs contain characters that are also special to your shell. For example, many URLs contain a question mark (?), which separates the host name and path from a list of arguments. However, the shell interprets the question mark as a wildcard.

To bypass interpretation by your shell, simply put the URL in single quotation marks. To avoid strange and long filenames, use wget -o to name the output file. Here's an example:

$ wget -o sharkey\
'http://www.example.com/\
download.cgi?proj=science&file=sharkey'

If you have a long list of URLs to download, you need not place each one on the command line. Instead, you can create (or rather, generate) a list of URLs to download. Command 3 shows url_list.txt, a simple text catalog containing two URLs; Command 4 downloads the two URLs. Use the -i option when you provide a list. The -nv option -- an acronym for not verbose -- provides more concise messages.

Unless you provide a file name for the download file (using the -o option), wget creates a new, local file with the same name as the remote file, omitting the entire leading URL. Command 5 shows the four files downloaded in commands 1 through 3.

The wget utility has many options and features. It can spider an FTP or Web site and download a complete hierarchy of files. You can also set a quota for automatic downloads, provide cookies, and continue a previous download that was interrupted. Read the wget man page to learn about the tool's many tricks.

Going up

The wget utility is invaluable for hands-off downloads, but it can't upload files. Nor can it interoperate with secure FTP, telnet, and a host of other (older and less-used) Internet protocols. For those kinds of transfers, you must turn to the veritable Swiss Army knife of networking: curl.

The curl command-line utility can get and put data, so it's ideal for transferring local files to remote servers. Better yet, the underpinning of curl -- the libcurl library -- has a rich application programming interface (API) that allows you to interrogate all the features of curl directly into your own applications. The C, C++, PHP, and Perl programming languages are just four of the many languages that can leverage libcurl. If your system lacks curl and libcurl, you can download the source code from the libcurl home page.

Because curl can copy local files to remote servers, it's ideal for small backups. For example, Listing 2 shows a shell script that copies a directory full of database dumps to a remote FTP server for safekeeping.

Listing 2. Using curl to store database dumps remotely
foreach db (mydns mysql cms tv radio)
  /usr/bin/mysqldump --ppassword --add-drop-table -Q --complete-insert $db > $db.sql 
end

find dbs -mtime -1 -type f -name '*.sql' -print | foreach file (`xargs`)
  curl -n -T $file ftp://ftp1.archive.example.com
end

The curl -n command forces curl to read your .netrc file. The -T option tells curl to upload the named file(s) to the given URL. If you omit the target file name, curl simply reuses the name of the file being uploaded.

As you might guess, curl has even more options than wget. It's worthwhile to read the curl man page and keep it in mind. The curl project also maintains a list of uses, including instructions on how to use the HTTP POST and PUT commands, how to provide login credentials, how to use SSL certificates, and how to debug your curl requests. A quick tip: Try curl -v --trace-ascii ... to generate tracing information.

Six addresses of separation

Modern computing depends largely on countless, spindly interconnections among machines of all shapes, sizes, and services. Indeed, even in a small computing environment, one computer might be dedicated to e-mail, another to serving Web pages, and others to performing yet more specialized tasks. In this environment -- typically connected by a local area network (LAN), WAN, or Virtual Private Network (VPN) -- it's quite common and necessary to log in to several computers per day. Systems administrators bounce from one computer to another each and every hour, but it's common for developers and other users to log in to require remote access for a critical application.

The X Window System and current desktop software make remote access fairly transparent: A window is a window, and the underlying application could be running on any computer. But again, the command line has a special place, even in a mouse-centric world. For example, how can you run the same command on multiple computers painlessly? Or, more simply, how do you launch an xterm window on the remote system?

Remote system access is the purview of ssh and its derivatives, scp and sftp. ssh is the secure version of rsh, while scp and sftp are secure replacements for rcp and FTP, respectively. Why is it secure? ssh and its variants provide stronger authentication mechanisms and encrypt all traffic using your choice of several ciphers. Even if someone sniffed your network, ssh traffic would look like so much gobblygook.

The simplest use of ssh is ssh hostname. This command connects to hostname and presents you with a login and password prompt. Provide the right credentials, and you're in:

(www.joe.com) $ ssh web.example.com
Login: arthur
Password: ******
( web.example.com) $

If you simply want to run one command on a remote system, you don't need to log in. Just provide the command as an argument to ssh. For example, the command shown in Listing 3 runs hostname -a -v on the remote computer.

Listing 3. Use ssh to run a command on a remote system
(www.joe.com) $ ssh db.linux-mag.com hostname -a -v
Login: vogon
Password: ******
db
gethostname()=`db.linux-mag.com'
Resolving `db.linux-mag.com' ...
Result: h_name=`db.linux-mag.com'
Result: h_aliases=`db'
Result: h_addr_list=`64.34.170.230'

ssh opened a connection to db.linux-mag.com and passed the hostname -a -v arguments to the remote computer, which ran the command and returned the output to the local computer.

ssh also provides a convenient way to copy files and entire directories from one computer to another. scp is almost as easy to use as cp. Here's an example:

(www.joe.com) $ scp -p -r ~/myproject web.example.com:

This command copies the ~/myproject directory to web.example.com. If you omit a destination path name, files are copied to your home directory. The -p option preserves the date and time stamps on all the files, while -r enables recursive mode, where scp descends and copies all subdirectories, as well.

By the way, the previous scp command is the equivalent of running:

(www.joe.com) $ tar czf - ~/myproject | ssh www.example.com tar xvzf - 
Login: deepthought
Password: ******

Yes, you can pipe the output of a local command to a remote command (or vice versa).

Chances are, you're already tired of all those password prompts. Again, the repeated prompts simply slow down work and prevent automation. You might also be tired of typing long host names over and over again. Luckily, ssh supports public or private key authentication and system aliases.

Let's set up a public or private key pair using the DSA encryption scheme. To do so, you must generate the key pair, copy the public key to the remote system, add it to the list of known keys, and verify that everything works, as shown in Listing 4.

Listing 4. Creating and installing a public or private key
1 $ cd ~
2 $ mkdir .ssh
3 $ chmod 700 .ssh
4 $ cd .ssh
5 $ ssh-keygen -t dsa 
Generating public/private dsa key pair.
Enter file in which to save the key (/home/mstreicher/.ssh/id_dsa): ./id_dsa
Enter passphrase (empty for no passphrase): 
Enter same passphrase again: 
Your identification has been saved in ./id_dsa.
Your public key has been saved in ./id_dsa.pub.
The key fingerprint is:
40:6c:26:e7:53:df:d1:7b:c4:79:c5:a8:cd:6b:fe:8e mstreicher@db.linux-mag.com
6 $ ls
id_dsa  id_dsa.pub
7 $ chmod 600 *

8 $ scp id_dsa.pub www.example.com:
Login: marvin
Password: ******
id_dsa  100%  668     0.7KB/s   00:00  

9 $ ssh www.example.com
Login: marvin
Password: ******
A $ mkdir .ssh
B $ chmod 700 .ssh
C $ cd .ssh
D $ cat ../id_dsa.pub >> authorized_keys
E $ rm ../id_dsa.pub
F $ chmod 600 *
G $ logout

10 $ ssh www.example.com

a $ hostname
www.example.com
b $ logout

Commands 1 through 3 create a private local directory named .ssh in your home directory. This directory must be mode 700, or ssh won't use public or private key authentication. (You can see the same sequence of commands run on the remote computer in steps A through C.) Command 5 creates the key pair using DSA. For now, leave the two passphrases blank. (They provide an extra level of security but add another authentication step.) ssh-keygen generates two files: id_dsa (the private key) and id_dsa.pub (the public key). Step 6 shows the files, while Step 7 protects both keys. Your keys must be mode 0600 or mode 0400.

Passing wildcards to remote shells

Let's say you want to list all the C source files in your remote home directory. Locally, you'd type something like ls -l *.c, so you try that through ssh:

$ ssh www.example.com ls -l *.c

Two things might happen: If you don't have any C files in the current working directory on your local computer, the shell complains, zsh: no matches found: *.c, or, if you have C files in the current working directory that aren't in your home directory on the remote computer, the shell on the remote computer might complain, ls: whosit.c: No such file or directory. Scratching your head?

The problem is that the wildcard * is being expanded by the local shell first, before the ssh command is sent. What you intended was for * to be expanded on the remote system.

To do that, you must prevent the local shell from interpreting the wildcard (again). You can place the * in single quotation marks or use a backslash (\) to escape the asterisk. Then, the asterisk is passed as a regular character to the remote shell, where it is interpreted in the context of the remote computer.

Here are those two approaches -- use the one that suits each situation:

$ ssh www.example.com ls -l \*.c
$ ssh www.example.com ls -l '*'.c

Step 8 copies the public key to the remote computer. For now, you must type your password, but this is the last time. Commands A through C create the private .ssh directory, and Step D adds the public key to the list of authorized keys. The name of the file -- authorized_keys -- is intentional. Do not name the file differently. Step E removes the copy of the key; Step F protects the files, as in Step 7.

When you log out and log back in, a password is no longer required. ssh (and scp and sftp) tests your private key against the remote public key. If a match is found, your credentials are sound and you can log in without further identification.

Some systems will always require a password; other systems might prefer RSA over DSA. Check with your systems administrator to find out how to log in to a specific computer. Systems administrators can set some global settings, too, to make a system more accessible.

Online, everywhere, all the time

Nowadays, the Internet connects people far and wide in ways not experienced before in human history. Whether sharing the details of your day in your blog or downloading source code for your next project, wires have replaced tires as the way to get around.

Web surfing is still a popular sport, but to make time for real surfing, developers have created ways to automate file transfers of all kinds. Using scripts and a few UNIX utilities, you can keep your external Web and download sites current. You can download and upload files with just a few keystrokes, making the process quick and easy. And if you create a .netrc file, you can hasten the effort even more. No more passwords.

Now that your mind is clear, put the top down and take a road trip on the information superhighway. See you at the Restaurant at the End of the Fiber. Last one there picks up the tab!

Resources

Learn

  • Speaking UNIX: Check out other parts in this series.
  • curl: Discover clever ways to use Curl to download a variety of resources, using many of the most common Internet protocols.
  • AIX and UNIX: Visit the developerWorks AIX and UNIX zone to expand your UNIX skills.
  • New to AIX and UNIX: Visit the New to AIX and UNIX page to learn more about AIX and UNIX.
  • developerWorks technical events and webcasts: Stay current with developerWorks technical events and webcasts.
  • AIX 5L Wiki: A collaborative environment for technical information related to AIX.
  • Podcasts: Tune in and catch up with IBM technical experts.

Get products and technologies

Discuss

Comments

developerWorks: Sign in

Required fields are indicated with an asterisk (*).


Need an IBM ID?
Forgot your IBM ID?


Forgot your password?
Change your password

By clicking Submit, you agree to the developerWorks terms of use.

 


The first time you sign into developerWorks, a profile is created for you. Information in your profile (your name, country/region, and company name) is displayed to the public and will accompany any content you post, unless you opt to hide your company name. You may update your IBM account at any time.

All information submitted is secure.

Choose your display name



The first time you sign in to developerWorks, a profile is created for you, so you need to choose a display name. Your display name accompanies the content you post on developerWorks.

Please choose a display name between 3-31 characters. Your display name must be unique in the developerWorks community and should not be your email address for privacy reasons.

Required fields are indicated with an asterisk (*).

(Must be between 3 – 31 characters.)

By clicking Submit, you agree to the developerWorks terms of use.

 


All information submitted is secure.

Dig deeper into AIX and Unix on developerWorks


static.content.url=http://www.ibm.com/developerworks/js/artrating/
SITE_ID=1
Zone=AIX and UNIX
ArticleID=157709
ArticleTitle=Speaking UNIX, Part 3: Do everything right from the command line
publish-date=09052006