Table of Contents
Standalone servers are independent of domain controllers on the network. They are not domain members and function more like workgroup servers. In many cases a standalone server is configured with a minimum of security control with the intent that all data served will be readily accessible to all users.
Standalone servers can be as secure or as insecure as needs dictate. They can have simple or complex configurations. Above all, despite the hoopla about domain security, they remain a common installation.
If all that is needed is a server for read-only files, or for printers alone, it may not make sense to effect a complex installation. For example, a drafting office needs to store old drawings and reference standards. Noone can write files to the server because it is legislatively important that all documents remain unaltered. A share-mode read-only standalone server is an ideal solution.
Another situation that warrants simplicity is an office that has many printers that are queued off a single central server. Everyone needs to be able to print to the printers, there is no need to effect any access controls, and no files will be served from the print server. Again, a share-mode standalone server makes a great solution.
The term standalone server means that it will provide local authentication and access control for all resources that are available from it. In general this means that there will be a local user database. In more technical terms, it means resources on the machine will be made available in either share mode or in user mode.
No special action is needed other than to create user accounts. Standalone servers do not provide network logon services. This means that machines that use this server do not perform a domain logon to it. Whatever logon facility the workstations are subject to is independent of this machine. It is, however, necessary to accommodate any network user so the logon name he or she uses will be translated (mapped) locally on the standalone server to a locally known user name. There are several ways this can be done.
Samba tends to blur the distinction a little in defining a standalone server. This is because the authentication database may be local or on a remote server, even if from the SMB protocol perspective the Samba server is not a member of a domain security context.
Through the use of Pluggable Authentication Modules (PAM) (see the chapter on PAM)
and the name service switcher (NSS), which maintains the UNIX-user database, the source of authentication may
reside on another server. We would be inclined to call this the authentication server. This means that the
Samba server may use the local UNIX/Linux system password database (
/etc/shadow), may use a local smbpasswd file, or may use an LDAP backend, or even via PAM
and Winbind another CIFS/SMB server for authentication.
The example Reference Documentation Server and Central Print Serving are designed to inspire simplicity. It is too easy to attempt a high level of creativity and to introduce too much complexity in server and network design.
Configuration of a read-only data server that everyone can access is very simple. By default, all shares are
read-only, unless set otherwise in the
smb.conf file. The example - Reference
Documentation Server is the
smb.conf file that will do this. Assume that all the reference documents
are stored in the directory
/export, and the documents are owned by a user other than
nobody. No home directories are shared, and there are no users in the
system database. This is a simple system to administer.
Example 7.1. smb.conf for Reference Documentation Server
I would have spoken more briefly, if I'd had more time to prepare.
In this example, the machine name is set to GANDALF, and the workgroup is set to the name of the local workgroup (MIDEARTH) so the machine will appear together with systems with which users are familiar. The only password backend required is the “guest” backend to allow default unprivileged account names to be used. As there is a WINS server on this network, we of course make use of it.
A US Air Force Colonel was renowned for saying: “Better is the enemy of good enough!” There are often sound reasons for avoiding complexity as well as for avoiding a technically perfect solution. Unfortunately, many network administrators still need to learn the art of doing just enough to keep out of trouble.
The print server must require no administration.
The print spooling and processing system on our print server will be CUPS. (Please refer to CUPS Printing Support, for more information).
The print server will service only network printers. The network administrator will correctly configure the CUPS environment to support the printers.
All workstations will use only PostScript drivers. The printer driver of choice is the one shipped with the Windows OS for the Apple Color LaserWriter.
In this example our print server will spool all incoming print jobs to
/var/spool/samba until the job is ready to be submitted by
Samba to the CUPS print processor. Since all incoming connections will be as
the anonymous (guest) user, two things will be required to enable anonymous printing.
Enabling Anonymous Printing
testparm -s -v | grep "guest account"
root#passwd -l pcguest
The exact command may vary depending on your UNIX/Linux distribution.
chown nobody.nobody /var/spool/samba
chmod a+rwt /var/spool/samba
The contents of the
smb.conf file is shown in the Anonymous Printing example.
smb.conf for Anonymous Printing
On CUPS-enabled systems there is a facility to pass raw data directly to the printer without intermediate
processing via CUPS print filters. Where use of this mode of operation is desired, it is necessary to
configure a raw printing device. It is also necessary to enable the raw mime handler in the
/etc/mime.types files. Refer to CUPS Printing Support, Explicitly Enable raw Printing
The example in the Anonymous Printing example uses CUPS for direct printing
via the CUPS libarary API. This means that all printers will be exposed to Windows users without need to
configure a printcap file. If there is necessity to expose only a sub-set of printers, or to define a special
type of printer (for example, a PDF filter) the
printcap name = cups can be replaced
with the entry
printcap name = /etc/samba/myprintcap. In this case the file specified
should contain a list of the printer names that should be exposed to Windows network users.