Printing Clients and Servers

by Michael W. Lucas
02/05/2004

Last time, we configured two commodity home printers to work with a FreeBSD system. Now we'll learn how to use FreeBSD's built-in tools to configure this system as a print server for other UNIX boxes and how to configure client machines to use this server. While there are many add-on programs that claim to make client printing easier, I find lpd(8) and its friends very quick and easy.

The Server

Start by making the printer work on the system you want to be the server. While the lpd(8) daemon that gets print jobs to the printer listens to the network by default, it doesn't accept print requests from just anyone. You must list the hostnames or IP addresses of permitted clients in /etc/hosts.lpd, each on a single line. I recommend using IP addresses so that a DNS failure won't take printing with it.

The Client

Printer client configuration is identical for any server that runs lpd(8). This means that your FreeBSD client can use any print server from the integrated lpd(8), add-on programs such as LPRng, or even Microsoft's Print Services for UNIX, without changing your client configuration. Here's a typical /etc/printcap configuration for a printing client.

lp|networkdept:\
        :lp=:\
        :rp=networkdept:\
        :rm=print:\
        :sd=/var/spool/output/lpd/networkdept:\
        :lf=/var/log/lpd-errs:

Just like termcap(5), each printer entry is technically a single line. The backslash character indicates that an entry continues on the next line. Also, each configuration option begins and ends with colons.

The first line is the printer name. This client has two names for this printer, lp and networkdept. lp is the standard UNIX name for the default printer, but I added networkdept as a friendly reminder. A printer can have any number of names — separate them with the pipe character, as shown above.

The second line, lp, lists the device name for this printer. As we're configuring a client for a network printer, leave the device name empty.

The third line, rp, gives the name that the print server uses for the printer in question. The print server's /etc/printcap lists the printer outside my office as networkdept, so that's the printer name I use.

The rm entry gives the hostname of the print server, in this case print. If your print server is in a different domain, use a fully qualified domain name here.

The fourth line, sd, lists the spool directory used for this printer. Every printer in your /etc/printcap needs a unique spool directory, which you must create before trying to use the printer.

Finally, the lf entry gives the full path to the log file. Unlike the spool directory, printers can share a log file.

Once you have configured a printer, fire up the printer daemon on your client. It doesn't matter if your printer is attached to the local machine or elsewhere on the network, the system still needs a daemon to send print jobs to the printer. You can now hit Print in your application and it should "just work."

Configuring Multiple Print Servers

So far, so easy? Now let's get slightly more complicated. This particular client machine is my laptop. I have one print server at work and another at home. The configuration above is for the work printer, but I want to come home and have everything magically work. I don't mind choosing a different printer in the OpenOffice Print dialog, but I don't want to change anything in /etc. Here's my laptop's configuration for my home laser printer. (We configured this printer in my last article.)

lp2|homelaser:\
        :lp=:\
        :rp=lp:\
        :rm=sardines.blackhelicopters.org:\
        :sd=/var/spool/output/lpd/hlp:\
        :lf=/var/log/lpd-errs:

The configuration looks quite similar, but with a few vital changes. The printer is named lp2 or homelaser. lp2 is the old-fashioned UNIX name for the second laser printer on the system. The print server knows this printer as lp, so that's the remote printer name. My home print server is called sardines.blackhelicopters.org. (I use the full domain name because the laptop has a hostname under the office's corporate domain.) By making this entry in /etc/printcap, I can access either printer. Of course, I must restart lpd(8) after making the edit, which will interrupt existing printing jobs. If this is a problem, as it might be on a high-volume print server, check out lpc(8) to make lpd reread its configuration file, restart individual printers, and perform other maintenance tasks. For a print client, lpc(8) is usually overkill.

Print Job Management

One common question printer clients have is, "Where is my print job?" The easiest way to tell is to check the print queue with lpq(1). Use the -P flag to identify a printer. For example, if I want to know if my last print job went to my home printer, I can run:

# lpq -P homelaser
no entries
#

The print server has no entries in its queue for this printer. My print job has probably gone on to the printer. Of course, the printer could have be chewing up the paper or spitting out reams of garbage, but that's a different problem. If the print server were still chewing on my print job, the entry would look a little more like this.

# lpq
Rank   Owner      Job  Files                                 Total Size
1st    mwlucas    38   (standard input)                      21640 bytes
#

If the print job were still sitting there after a while, I'd have to go log into the print server and look at the error log to see what was happening.

Many graphical programs, such as Mozilla, have a field where you can set a flag to choose a printer. Desktop environments such as KDE and Gnome have places where you can set printers and flags in a nice menu-driven format. Applications such as OpenOffice have a special tool used to configure application printers.

My only real problem with a multi-environment printing setup is that I continually send print jobs to the wrong printer. This isn't just a matter of "Oh, I have to walk across the office to the other printer" -- the printer I sent the job to doesn't even exist in the environment I'm in! If you're wondering where your print job went, check the queue for the other environment. For example, if I'm working at home on the weekend and just hit the Print button, the print job will go to the laptop's default (i.e., workplace) queue and hang there.

# lpq
mwlaptop.company.com: Warning: no daemon present
Rank   Owner      Job  Files                                 Total Size
1st    mwlucas    38   (standard input)                      21640 bytes

#

The second line, with the "no daemon present" warning, indicates that something's wrong with the printer. The easy guess is that I can't talk to that printer from the wrong side of the corporate firewall. If I leave those print jobs in the queue, when I go into work and plug in, the printer outside of my office will start to spew everything I wanted to print over the weekend. This may include things that I don't want to leave sitting on the work printer, so it's a good idea to clean out that queue.

lprm(1) removes jobs from the printer queue. lprm takes a whole variety of arguments, but the most useful for this purpose is a simple dash (-). This tells lpd(8) to remove all print jobs that belong to the user who is running the program. You can choose a printer with the -P flag, exactly as you can when viewing the print queue.

# lprm -P networkdept -
dfA039mwlaptop.company.com dequeued
cfA039mwlaptop.company.com dequeued
#

You can also use this to cancel print jobs if you hit print by mistake, but modern print servers work very quickly. By the time you type lprm -, the print server will have almost certainly already sent the job to the printer, and it will be far too late to cancel the job.