Annelise Anderson

andrsn@andrsn.stanford.edu

August 15, 1997 Congratulations on installing FreeBSD! This introduction is for people new to both FreeBSD and Un*x—so it starts with basics. It assumes you're using version 2.0.5 or later of FreeBSD as distributed by Walnut Creek or FreeBSD.ORG, your system (for now) has a single user (you)—and you're probably pretty good with DOS/Windows or OS/2. Log in (when you see login:) as a user you created during installation or as root. (Your FreeBSD installation will already have an account for root; root can go anywhere and do anything, including deleting essential files, so be careful!) The symbols % and # in the following stand for the prompt (yours may be different), with % indicating an ordinary user and # indicating root. To log out (and get a new login: prompt) type # exit as often as necessary. Yes, press enter after commands, and remember that Unix is case-sensitive—exit, not EXIT. To shut down the machine type: # /sbin/shutdown -h now Or to reboot type # /sbin/shutdown -r now or # /sbin/reboot You can also reboot with CtrlAltDelete. Give it a little time to do its work. This is equivalent to /sbin/reboot in recent releases of FreeBSD, and is much, much better than hitting the reset button. You don't want to have to reinstall this thing, do you? If you didn't create any users when you installed the system and are thus logged in as root, you should probably create a user now with # adduser The first time you use adduser, it might ask for some defaults to save. You might want to make the default shell csh instead of sh, if it suggests sh as the default. Otherwise just press enter to accept each default. These defaults are saved in /etc/adduser.conf, an editable file. Suppose you create a user jack with full name Jack Benimble. Give jack a password if security (even kids around who might pound on the keyboard) is an issue. When it asks you if you want to invite jack into other groups, type wheel Login group is ``jack''. Invite jack into other groups: wheel This will make it possible to log in as jack and use the su command to become root. Then you won't get scolded any more for logging in as root. You can quit adduser any time by typing CtrlC, and at the end you'll have a chance to approve your new user or simply type n for no. You might want to create a second new user (jill?) so that when you edit jack's login files, you'll have a hot spare in case something goes wrong. Once you've done this, use exit to get back to a login prompt and log in as jack. In general, it's a good idea to do as much work as possible as an ordinary user who doesn't have the power—and risk—of root. If you already created a user and you want the user to be able to su to root, you can log in as root and edit the file /etc/group, adding jack to the first line (the group wheel). But first you need to practice vi, the text editor--or use the simpler text editor, ee, installed on recent version of FreeBSD. To delete a user, use the rmuser command. Logged in as an ordinary user, look around and try out some commands that will access the sources of help and information within FreeBSD. Here are some commands and what they do: id Tells you who you are! pwd Shows you where you are—the current working directory. ls Lists the files in the current directory. ls -F Lists the files in the current directory with a * after executables, a / after directories, and an @ after symbolic links. ls -l Lists the files in long format—size, date, permissions. ls -a Lists hidden dot files with the others. If you're root, thedot files show up without the -a switch. cd Changes directories. cd .. backs up one level; note the space after cd. cd /usr/local goes there. cd ~ goes to the home directory of the person logged in—e.g., /usr/home/jack. Try cd /cdrom, and then ls, to find out if your CDROM is mounted and working. view filename Lets you look at a file (named filename without changing it. Try view /etc/fstab. :q to quit. cat filename Displays filename on screen. If it's too long and you can see only the end of it, press ScrollLock and use the up-arrow to move backward; you can use ScrollLock with man pages too. Press ScrollLock again to quit scrolling. You might want to try cat on some of the dot files in your home directory—cat .cshrc, cat .login, cat .profile. You'll notice aliases in .cshrc for some of the ls commands (they're very convenient). You can create other aliases by editing .cshrc. You can make these aliases available to all users on the system by putting them in the system-wide csh configuration file, /etc/csh.cshrc. Here are some useful sources of help. Text stands for something of your choice that you type in—usually a command or filename. apropos text Everything containing string text in the whatis database. man text The man page for text. The major source of documentation for Un*x systems. man ls will tell you all the ways to use the ls command. Press Enter to move through text, Ctrlb to go back a page, Ctrlf to go forward, q or Ctrlc to quit. which text Tells you where in the user's path the command text is found. locate text All the paths where the string text is found. whatis text Tells you what the command text does and its man page. Typing whatis * will tell you about all the binaries in the current directory. whereis text Finds the file text, giving its full path. You might want to try using whatis on some common useful commands like cat, more, grep, mv, find, tar, chmod, chown, date, and script. more lets you read a page at a time as it does in DOS, e.g., ls -l | more or more filename. The * works as a wildcard—e.g., ls w* will show you files beginning with w. Are some of these not working very well? Both locate and whatis depend on a database that's rebuilt weekly. If your machine isn't going to be left on over the weekend (and running FreeBSD), you might want to run the commands for daily, weekly, and monthly maintenance now and then. Run them as root and give each one time to finish before you start the next one, for now. # /etc/daily output omitted # /etc/weekly output omitted # /etc/monthly output omitted If you get tired waiting, press AltF2 to get another virtual console, and log in again. After all, it's a multi-user, multi-tasking system. Nevertheless these commands will probably flash messages on your screen while they're running; you can type clear at the prompt to clear the screen. Once they've run, you might want to look at /var/mail/root and /var/log/messages. Basically running such commands is part of system administration—and as a single user of a Unix system, you're your own system administrator. Virtually everything you need to be root to do is system administration. Such responsibilities aren't covered very well even in those big fat books on Unix, which seem to devote a lot of space to pulling down menus in windows managers. You might want to get one of the two leading books on systems administration, either Evi Nemeth et.al.'s UNIX System Administration Handbook (Prentice-Hall, 1995, ISBN 0-13-15051-7)—the second edition with the red cover; or Æleen Frisch's Essential System Administration (O'Reilly & Associates, 1993, ISBN 0-937175-80-3). I used Nemeth. To configure your system, you need to edit text files. Most of them will be in the /etc directory; and you'll need to su to root to be able to change them. You can use the easy ee, but in the long run the text editor vi is worth learning. There's an excellent tutorial on vi in /usr/src/contrib/nvi/docs/tutorial if you have that installed; otherwise you can get it by ftp to ftp.cdrom.com in the directory FreeBSD/FreeBSD-current/src/contrib/nvi/docs/tutorial. Before you edit a file, you should probably back it up. Suppose you want to edit /etc/rc.conf. You could just use cd /etc to get to the /etc directory and do: # cp rc.conf rc.conf.orig This would copy rc.conf to rc.conf.orig, and you could later copy rc.conf.orig to rc.conf to recover the original. But even better would be moving (renaming) and then copying back: # mv rc.conf rc.conf.orig # cp rc.conf.orig rc.conf because the mv command preserves the original date and owner of the file. You can now edit rc.conf. If you want the original back, you'd then mv rc.conf rc.conf.myedit (assuming you want to preserve your edited version) and then # mv rc.conf.orig rc.conf to put things back the way they were. To edit a file, type # vi filename Move through the text with the arrow keys. Esc (the escape key) puts vi in command mode. Here are some commands: x delete letter the cursor is on dd delete the entire line (even if it wraps on the screen) i insert text at the cursor a insert text after the cursor Once you type i or a, you can enter text. Esc puts you back in command mode where you can type :w to write your changes to disk and continue editing :wq to write and quit :q! to quit without saving changes /text to move the cursor to text; /Enter (the enter key) to find the next instance of text. G to go to the end of the file nG to go to line n in the file, where n is a number CtrlL to redraw the screen Ctrlb and Ctrlf go back and forward a screen, as they do with more and view. Practice with vi in your home directory by creating a new file with vi filename and adding and deleting text, saving the file, and calling it up again. vi delivers some surprises because it's really quite complex, and sometimes you'll inadvertently issue a command that will do something you don't expect. (Some people actually like vi—it's more powerful than DOS EDIT—find out about the :r command.) Use Esc one or more times to be sure you're in command mode and proceed from there when it gives you trouble, save often with :w, and use :q! to get out and start over (from your last :w) when you need to. Now you can cd to /etc, su to root, use vi to edit the file /etc/group, and add a user to wheel so the user has root privileges. Just add a comma and the user's login name to the end of the first line in the file, press Esc, and use :wq to write the file to disk and quit. Instantly effective. (You didn't put a space after the comma, did you?) At this point you probably don't have the printer working, so here's a way to create a file from a man page, move it to a floppy, and then print it from DOS. Suppose you want to read carefully about changing permissions on files (pretty important). You can use the command man chmod to read about it. The command # man chmod | col -b > chmod.txt will remove formatting codes and send the man page to the chmod.txt file instead of showing it on your screen. Now put a dos-formatted diskette in your floppy drive a, su to root, and type # /sbin/mount -t msdos /dev/fd0 /mnt to mount the floppy drive on /mnt. Now (you no longer need to be root, and you can type exit to get back to being user jack) you can go to the directory where you created chmod.txt and copy the file to the floppy with: % cp chmod.txt /mnt and use ls /mnt to get a directory listing of /mnt, which should show the file chmod.txt. You might especially want to make a file from /sbin/dmesg by typing % /sbin/dmesg > dmesg.txt and copying dmesg.txt to the floppy. /sbin/dmesg is the boot log record, and it's useful to understand it because it shows what FreeBSD found when it booted up. If you ask questions on freebsd-questions@FreeBSD.ORG or on a USENET group—like FreeBSD isn't finding my tape drive, what do I do?—people will want to know what dmesg has to say. You can now dismount the floppy drive (as root) to get the disk out with # /sbin/umount /mnt and reboot to go to DOS. Copy these files to a DOS directory, call them up with DOS EDIT, Windows Notepad or Wordpad, or a word processor, make a minor change so the file has to be saved, and print as you normally would from DOS or Windows. Hope it works! man pages come out best if printed with the dos print command. (Copying files from FreeBSD to a mounted dos partition is in some cases still a little risky.) Getting the printer printing from FreeBSD involves creating an appropriate entry in /etc/printcap and creating a matching spool directory in /var/spool/output. If your printer is on lpt0 (what dos calls LPT1), you may only need to go to /var/spool/output and (as root) create the directory lpd by typing: mkdir lpd, if it doesn't already exist. Then the printer should respond if it's turned on when the system is booted, and lp or lpr should send a file to the printer. Whether or not the file actually prints depends on configuring it, which is covered in the FreeBSD handbook. df shows file space and mounted systems. ps aux shows processes running. ps ax is a narrower form. rm filename remove filename. rm -R dir removes a directory dir and all subdirectories—careful! ls -R lists files in the current directory and all subdirectories; I used a variant, ls -AFR > where.txt, to get a list of all the files in / and (separately) /usr before I found better ways to find files. passwd to change user's password (or root's password) man hier man page on the Unix file system Use find to locate filename in /usr or any of its subdirectories with % find /usr -name "filename" You can use * as a wildcard in "filename" (which should be in quotes). If you tell find to search in / instead of /usr it will look for the file(s) on all mounted file systems, including the CDROM and the dos partition. An excellent book that explains Unix commands and utilities is Abrahams & Larson, Unix for the Impatient (2nd ed., Addison-Wesley, 1996). There's also a lot of Unix information on the Internet. Try the Unix Reference Desk. You should now have the tools you need to get around and edit files, so you can get everything up and running. There is a great deal of information in the FreeBSD handbook (which is probably on your hard drive) and FreeBSD's web site. A wide variety of packages and ports are on the Walnut Creek CDROM as well as the web site. The handbook tells you more about how to use them (get the package if it exists, with pkg_add /cdrom/packages/All/packagename, where packagename is the filename of the package). The cdrom has lists of the packages and ports with brief descriptions in cdrom/packages/index, cdrom/packages/index.txt, and cdrom/ports/index, with fuller descriptions in /cdrom/ports/*/*/pkg/DESCR, where the *s represent subdirectories of kinds of programs and program names respectively. If you find the handbook too sophisticated (what with lndir and all) on installing ports from the cdrom, here's what usually works: Find the port you want, say kermit. There will be a directory for it on the cdrom. Copy the subdirectory to /usr/local (a good place for software you add that should be available to all users) with: # cp -R /cdrom/ports/comm/kermit /usr/local This should result in a /usr/local/kermit subdirectory that has all the files that the kermit subdirectory on the CDROM has. Next, create the directory /usr/ports/distfiles if it doesn't already exist using mkdir. Now check check /cdrom/ports/distfiles for a file with a name that indicates it's the port you want. Copy that file to /usr/ports/distfiles; in recent versions you can skip this step, as FreeBSD will do it for you. In the case of kermit, there is no distfile. Then cd to the subdirectory of /usr/local/kermit that has the file Makefile. Type # make all install During this process the port will ftp to get any compressed files it needs that it didn't find on the cdrom or in /usr/ports/distfiles. If you don't have your network running yet and there was no file for the port in /cdrom/ports/distfiles, you will have to get the distfile using another machine and copy it to /usr/ports/distfiles from a floppy or your dos partition. Read Makefile (with cat or more or view) to find out where to go (the master distribution site) to get the file and what its name is. Its name will be truncated when downloaded to DOS, and after you get it into /usr/ports/distfiles you'll have to rename it (with the mv command) to its original name so it can be found. (Use binary file transfers!) Then go back to /usr/local/kermit, find the directory with Makefile, and type make all install. The other thing that happens when installing ports or packages is that some other program is needed. If the installation stops with a message can't find unzip or whatever, you might need to install the package or port for unzip before you continue. Once it's installed type rehash to make FreeBSD reread the files in the path so it knows what's there. (If you get a lot of path not found messages when you use whereis or which, you might want to make additions to the list of directories in the path statement in .cshrc in your home directory. The path statement in Unix does the same kind of work it does in DOS, except the current directory is not (by default) in the path for security reasons; if the command you want is in the directory you're in, you need to type ./ before the command to make it work; no space after the slash.) You might want to get the most recent version of Netscape from their ftp site. (Netscape requires the X Window System.) There's now a FreeBSD version, so look around carefully. Just use gunzip filename and tar xvf filename on it, move the binary to /usr/local/bin or some other place binaries are kept, rehash, and then put the following lines in .cshrc in each user's home directory or (easier) in /etc/csh.cshrc, the system-wide csh start-up file: setenv XKEYSYMDB /usr/X11R6/lib/X11/XKeysymDB setenv XNLSPATH /usr/X11R6/lib/X11/nls This assumes that the file XKeysymDB and the directory nls are in /usr/X11R6/lib/X11; if they're not, find them and put them there. If you originally got Netscape as a port using the CDROM (or ftp), don't replace /usr/local/bin/netscape with the new netscape binary; this is just a shell script that sets up the environmental variables for you. Instead rename the new binary to netscape.bin and replace the old binary, which is /usr/local/lib/netscape/netscape.bin. Your shell is the most important part of your working environment. In DOS, the usual shell is command.com. The shell is what interprets the commands you type on the command line, and thus communicates with the rest of the operating system. You can also write shell scripts, which are like DOS batch files: a series of commands to be run without your intervention. Two shells come installed with FreeBSD: csh and sh. csh is good for command-line work, but scripts should be written with sh (or bash). You can find out what shell you have by typing echo $SHELL. The csh shell is okay, but tcsh does everything csh does and more. It It allows you to recall commands with the arrow keys and edit them. It has tab-key completion of filenames (csh uses the escape key), and it lets you switch to the directory you were last in with cd -. It's also much easier to alter your prompt with tcsh. It makes life a lot easier. Here are the three steps for installing a new shell: 1. Install the shell as a port or a package, just as you would any other port or package. Use rehash and which tcsh (assuming you're installing tcsh) to make sure it got installed. 2. As root, edit /etc/shells, adding a line in the file for the new shell, in this case /usr/local/bin/tcsh, and save the file. (Some ports may do this for you.) 3. Use the chsh command to change your shell to tcsh permanently, or type tcsh at the prompt to change your shell without logging in again. Note: It can be dangerous to change root's shell to something other than sh or csh on early versions of FreeBSD and many other versions of Unix; you may not have a working shell when the system puts you into single user mode. The solution is to use su -m to become root, which will give you the tcsh as root, because the shell is part of the environment. You can make this permanent by adding it to your .tcshrc file as an alias with alias su su -m. When tcsh starts up, it will read the /etc/csh.cshrc and /etc/csh.login files, as does csh. It will also read the .login file in your home directory and the .cshrc file as well, unless you provide a .tcshrc file. This you can do by simply copying .cshrc to .tcshrc. Now that you've installed tcsh, you can adjust your prompt. You can find the details in the manual page for tcsh, but here is a line to put in your .tcshrc that will tell you how many commands you have typed, what time it is, and what directory you are in. It also produces a > if you're an ordinary user and a # if you're root, but tsch will do that in any case: set prompt = "%h %t %~ %# " This should go in the same place as the existing set prompt line if there is one, or under "if($?prompt) then" if not. Comment out the old line; you can always switch back to it if you prefer it. Don't forget the spaces and quotes. You can get the .tcshrc reread by typing source .tcshrc. You can get a listing of other environmental variables that have been set by typing env at the prompt. The result will show you your default editor, pager, and terminal type, among possibly many others. A useful command if you log in from a remote location and can't run a program because the terminal isn't capable is setenv TERM vt100. As root, you can dismount the CDROM with /sbin/umount /cdrom, take it out of the drive, insert another one, and mount it with /sbin/mount_cd9660 /dev/cd0a /cdrom assuming cd0a is the device name for your CDROM drive. The most recent versions of FreeBSD let you mount the cdrom with just /sbin/mount /cdrom. Using the live file system—the second of FreeBSD's CDROM disks—is useful if you've got limited space. What is on the live file system varies from release to release. You might try playing games from the cdrom. This involves using lndir, which gets installed with the X Window System, to tell the program(s) where to find the necessary files, because they're in the /cdrom file system instead of in /usr and its subdirectories, which is where they're expected to be. Read man lndir. If you use this guide I'd be interested in knowing where it was unclear and what was left out that you think should be included, and if it was helpful. My thanks to Eugene W. Stark, professor of computer science at SUNY-Stony Brook, and John Fieber for helpful comments. Annelise Anderson, andrsn@andrsn.stanford.edu