NAME
yp
—
description of the YP/NIS
system
SYNOPSIS
yp |
DESCRIPTION
TheYP
subsystem allows network management of passwd,
group, netgroup, hosts, services, rpc, bootparams and ethers file entries
through the functions
getpwent(3),
getgrent(3),
getnetgrent(3),
gethostent(3),
getnetent(3),
getrpcent(3), and
ethers(3). The
bootparamd(8) daemon makes direct NIS library calls since there
are no functions in the standard C library for reading bootparams. NIS support
is enabled in
nsswitch.conf(5).
The YP
subsystem is started automatically
in /etc/rc if it has been initialized in
/etc/rc.conf and if the directory
/var/yp exists (which it does in the default
distribution). The default NIS domain must also be set with the
domainname(1) command, which will happen automatically at
system startup if it is specified in
/etc/rc.conf.
NIS is an RPC-based client/server system that allows a group of machines within an NIS domain to share a common set of configuration files. This permits a system administrator to set up NIS client systems with only minimal configuration data and add, remove or modify configuration data from a single location.
The canonical copies of all NIS information are stored on a single machine called the NIS master server. The databases used to store the information are called NIS maps. In DragonFly, these maps are stored in /var/yp/⟨domainname⟩ where ⟨domainname⟩ is the name of the NIS domain being served. A single NIS server can support several domains at once, therefore it is possible to have several such directories, one for each supported domain. Each domain will have its own independent set of maps.
In DragonFly, the NIS maps are Berkeley DB
hashed database files (the same format used for the
passwd(5) database files). Other operating systems that support NIS
use old-style ndbm
databases instead (largely
because Sun Microsystems originally based their NIS implementation on
ndbm
, and other vendors have simply licensed Sun's
code rather than design their own implementation with a different database
format). On these systems, the databases are generally split into
.dir and .pag files which
the ndbm
code uses to hold separate parts of the
hash database. The Berkeley DB hash method instead uses a single file for
both pieces of information. This means that while you may have
passwd.byname.dir and
passwd.byname.pag files on other operating systems
(both of which are really parts of the same map),
DragonFly will have only one file called
passwd.byname. The difference in format is not
significant: only the NIS server,
ypserv(8), and related tools need to know the database format of the
NIS maps. Client NIS systems receive all NIS data in ASCII form.
There are three main types of NIS systems:
- NIS clients, which query NIS servers for information.
- NIS master servers, which maintain the canonical copies of all NIS maps.
- NIS slave servers, which maintain backup copies of NIS maps that are periodically updated by the master.
A NIS client establishes what is called a binding to a particular NIS server using the ypbind(8) daemon. The ypbind(8) utility checks the system's default domain (as set by the domainname(1) command) and begins broadcasting RPC requests on the local network. These requests specify the name of the domain for which ypbind(8) is attempting to establish a binding. If a server that has been configured to serve the requested domain receives one of the broadcasts, it will respond to ypbind(8), which will record the server's address. If there are several servers available (a master and several slaves, for example), ypbind(8) will use the address of the first one to respond. From that point on, the client system will direct all of its NIS requests to that server. The ypbind(8) utility will occasionally “ping” the server to make sure it is still up and running. If it fails to receive a reply to one of its pings within a reasonable amount of time, ypbind(8) will mark the domain as unbound and begin broadcasting again in the hopes of locating another server.
NIS master and slave servers handle all NIS requests with the ypserv(8) daemon. The ypserv(8) utility is responsible for receiving incoming requests from NIS clients, translating the requested domain and map name to a path to the corresponding database file and transmitting data from the database back to the client. There is a specific set of requests that ypserv(8) is designed to handle, most of which are implemented as functions within the standard C library:
yp_order
()- check the creation date of a particular map
yp_master
()- obtain the name of the NIS master server for a given map/domain
yp_match
()- lookup the data corresponding to a given in key in a particular map/domain
yp_first
()- obtain the first key/data pair in a particular map/domain
yp_next
()- pass ypserv(8) a key in a particular map/domain and have it
return the key/data pair immediately following it (the functions
yp_first
() andyp_next
() can be used to do a sequential search of an NIS map) yp_all
()- retrieve the entire contents of a map
There are a few other requests which
ypserv(8) is capable of handling (i.e., acknowledge whether or not
you can handle a particular domain (YPPROC_DOMAIN
),
or acknowledge only if you can handle the domain and be silent otherwise
(YPPROC_DOMAIN_NONACK
)) but these requests are
usually generated only by
ypbind(8) and are not meant to be used by standard utilities.
On networks with a large number of hosts, it is often a good idea
to use a master server and several slaves rather than just a single master
server. A slave server provides the exact same information as a master
server: whenever the maps on the master server are updated, the new data
should be propagated to the slave systems using the
yppush(8) command. The NIS Makefile
(/var/yp/Makefile) will do this automatically if the
administrator comments out the line which says
“NOPUSH=true
”
(NOPUSH is set to true by default because the default
configuration is for a small network with only one NIS server). The
yppush(8) command will initiate a transaction between the master and
slave during which the slave will transfer the specified maps from the
master server using
ypxfr(8). (The slave server calls
ypxfr(8) automatically from within
ypserv(8); therefore it is not usually necessary for the
administrator to use it directly. It can be run manually if desired,
however.) Maintaining slave servers helps improve NIS performance on large
networks by:
- Providing backup services in the event that the NIS master crashes or becomes unreachable
- Spreading the client load out over several machines instead of causing the master to become overloaded
- Allowing a single NIS domain to extend beyond a local network (the ypbind(8) daemon might not be able to locate a server automatically if it resides on a network outside the reach of its broadcasts. It is possible to force ypbind(8) to bind to a particular server with ypset(8) but this is sometimes inconvenient. This problem can be avoided simply by placing a slave server on the local network.)
The DragonFly ypserv(8) is specially designed to provide enhanced security (compared to other NIS implementations) when used exclusively with DragonFly and FreeBSD client systems. The DragonFly password database system (which is derived directly from 4.4BSD) includes support for shadow passwords. The standard password database does not contain users' encrypted passwords: these are instead stored (along with other information) in a separate database which is accessible only by the super-user. If the encrypted password database were made available as an NIS map, this security feature would be totally disabled, since any user is allowed to retrieve NIS data.
To help prevent this, DragonFly's NIS server handles the shadow password maps (master.passwd.byname and master.passwd.byuid) in a special way: the server will only provide access to these maps in response to requests that originate on privileged ports. Since only the super-user is allowed to bind to a privileged port, the server assumes that all such requests come from privileged users. All other requests are denied: requests from non-privileged ports will receive only an error code from the server. Additionally, DragonFly's ypserv(8) includes support for Wietse Venema's tcp wrapper package; with tcp wrapper support enabled, the administrator can configure ypserv(8) to respond only to selected client machines.
While these enhancements provide better security than stock NIS, they are by no means 100% effective. It is still possible for someone with access to your network to spoof the server into disclosing the shadow password maps.
On the client side, DragonFly's getpwent(3) functions will automatically search for the master.passwd maps and use them if they exist. If they do, they will be used, and all fields in these special maps (class, password age and account expiration) will be decoded. If they are not found, the standard passwd maps will be used instead.
COMPATIBILITY
When using a
non-DragonFly/FreeBSD
NIS server for
passwd(5) files, it is unlikely that the default MD5-based format
that DragonFly uses for passwords will be accepted
by it. If this is the case, the value of the
passwd_format setting in
login.conf(5) should be changed to
"des
" for compatibility.
Some systems, such as SunOS 4.x, need NIS to be running in order
for their hostname resolution functions
(gethostbyname
(),
gethostbyaddr
(), etc.) to work properly. On these
systems, ypserv(8) performs DNS lookups when asked to return
information about a host that does not exist in its
hosts.byname or hosts.byaddr
maps. DragonFly's resolver uses DNS by default (it
can be made to use NIS, if desired), therefore its NIS server does not do
DNS lookups by default. However,
ypserv(8) can be made to perform DNS lookups if it is started with a
special flag. It can also be made to register itself as an NIS v1 server in
order to placate certain systems that insist on the presence of a v1 server
(DragonFly uses only NIS v2, but many other systems,
including SunOS 4.x, search for both a v1 and v2 server when binding).
DragonFly's
ypserv(8) does not actually handle NIS v1 requests, but this
“kludge mode” is useful for silencing stubborn systems that
search for both a v1 and v2 server.
(Please see the ypserv(8) manual page for a detailed description of these special features and flags.)
HISTORY
The YP
subsystem was written from the
ground up by Theo de Raadt to be compatible to Sun's
implementation. Bug fixes, improvements and NIS server support were later
added by Bill Paul. The server-side code was
originally written by Peter Eriksson and
Tobias Reber and is subject to the GNU Public
License. No Sun code was referenced.
BUGS
While DragonFly now has both NIS client
and server capabilities, it does not yet have support for
ypupdated(8) or the yp_update
()
function. Both of these require secure RPC, which
DragonFly does not support yet either.
The getservent(3) and getprotoent(3) functions do not yet have NIS support. Fortunately, these files do not need to be updated that often.
Many more manual pages should be written, especially ypclnt(3). For the time being, seek out a local Sun machine and read the manuals for there.
Neither Sun nor this author have found a clean way to handle the problems that occur when ypbind cannot find its server upon bootup.