| RPC.YPPASSWDD(8) | MidnightBSD System Manager's Manual | RPC.YPPASSWDD(8) |
rpc.yppasswdd —
server for updating NIS passwords
rpc.yppasswdd |
[-t master.passwd template
file] [-d default
domain] [-p path]
[-s] [-f]
[-a] [-m]
[-i] [-v]
[-u] [-h] |
The rpc.yppasswdd utility allows users to
change their NIS passwords and certain other information using the
yppasswd(1) and
ypchpass(1) commands.
The rpc.yppasswdd utility is an RPC-based server
that accepts incoming password change requests, authenticates them, places
the updated information in the /var/yp/master.passwd
template file and then updates the NIS master.passwd
and passwd maps.
The rpc.yppasswdd utility allows a normal
NIS user to change his or her NIS password, full name (also known as 'GECOS'
field) or shell. These updates are typically done using the
yppasswd(1),
ypchfn(1),
ypchsh(1), or
ypchpass(1) commands.
(Some administrators do not want users to be able to change their full name
information or shells; the server can be invoked with option flags that
disallow such changes.) When the server receives an update request, it
compares the address of the client making the request against the
securenets rules outlined in
/var/yp/securenets. (See the
ypserv(8) manual page for
more information on securenets; the rpc.yppasswdd
utility uses the same access control mechanism as
ypserv(8).)
The server then checks the 'old' password supplied by the user to make sure it is valid, then performs some sanity checks on the updated information (these include checking for embedded control characters, colons or invalid shells). Once it is satisfied that the update request is valid, the server modifies the template password file (the default is /var/yp/master.passwd) and then runs the /usr/libexec/yppwupdate script to rebuild the NIS maps. (This script has two arguments passed to it: the absolute pathname of the password template that was modified and the name of the domain that is to be updated. These in turn are passed to /var/yp/Makefile).
The FreeBSD version of
rpc.yppasswdd also allows the super-user on the NIS
master server to perform more sophisticated updates on the NIS passwd maps.
The super-user can modify any field in any user's master.passwd entry in any
domain, and can do so without knowing the user's existing NIS password (when
the server receives a request from the super-user, the password
authentication check is bypassed). Furthermore, if the server is invoked
with the -a flag, the super-user can even add new
entries to the maps using
ypchpass(1). Again,
this only applies to the super-user on the NIS master server: none of these
special functions can be performed over the network.
The rpc.yppasswdd utility can only be run
on a machine that is an NIS master server.
The following options are available:
-t
master.passwd template filerpc.yppasswdd assumes that the
template file used to generates the master.passwd
and passwd maps for the default domain is called
/var/yp/master.passwd. This default can be
overridden by specifying an alternate file name with the
-t flag.
Note: if the template file specified with this flag is
/etc/master.passwd,
rpc.yppasswdd will also automatically invoke
pwd_mkdb(8) to
rebuild the local password databases in addition to the NIS maps.
-d
domainrpc.yppasswdd utility can support multiple
domains, however it must choose one domain as a default. It will try to
use the system default domain name as set by the
domainname(1)
command for this default. However, if the system domain name is not set, a
default domain must be specified on the command line. If the system
default domain is set, then this option can be used to override it.-p
path-s-f-arpc.yppasswdd is started with this flag, it will
also allow the super-user to add new records to the NIS passwd maps, just
as is possible when using
chpass(1) to modify the
local password database.-mrpc.yppasswdd can only operate on a single NIS
domain, which is generally the same as the system default domain of the
NIS master server. The FreeBSD
rpc.yppasswdd attempts to overcome this problem in
spite of the inherent limitations of the yppasswd
protocol, which does not allow for a domain
argument in client requests. In multi-domain mode,
rpc.yppasswdd will search through all the passwd
maps of all the domains it can find under /var/yp
until it finds an entry that matches the user information specified in a
given update request. (Matches are determined by checking the username,
UID and GID fields.) The matched entry and corresponding domain are then
used for the update.
Note that in order for multi-domain mode to work, there have to be separate template files for each domain. For example, if a server supports three domains, foo, bar, and baz, there should be three separate master.passwd template files called /var/yp/foo/master.passwd, /var/yp/bar/master.passwd, and /var/yp/baz/master.passwd. If foo happens to be the system default domain, then its template file can be either /var/yp/foo/master.passwd or /var/yp/master.passwd. The server will check for the latter file first and then use the former if it cannot find it.
Multi-domain mode is off by default since it can fail if there are duplicate or near-duplicate user entries in different domains. The server will abort an update request if it finds more than one user entry that matches its search criteria. Even so, paranoid administrators may wish to leave multi-domain mode disabled.
-irpc.yppasswdd is invoked with this flag, it
will perform map updates in place. This means that instead of just
modifying the password template file and starting a map update, the server
will modify the map databases directly. This is useful when the password
maps are large: if, for example, the password database has tens of
thousands of entries, it can take several minutes for a map update to
complete. Updating the maps in place reduces this time to a few
seconds.-v-v flag will cause it to log informational
messages for all updates.-urpc.yppasswdd. This is either because the
yppasswd(1) program
is not installed set-uid root, or because the RPC implementation does not
place any emphasis on binding to reserved ports when establishing client
connections for the super-user. By default,
rpc.yppasswdd expects to receive requests from
clients using reserved ports; requests received from non-privileged ports
are rejected. Unfortunately, this behavior prevents any client systems
that to not use privileged ports from successfully submitting password
updates. Specifying the -u flag to
rpc.yppasswdd disables the privileged port check
so that it will work with
yppasswd(1) clients
that do not use privileged ports. This reduces security to a certain small
degree, but it might be necessary in cases where it is not possible to
change the client behavior.-hrpc.yppasswdd.rpc.yppasswdd to update and
push the NIS maps after an update.Bill Paul <wpaul@ctr.columbia.edu>
As listed in the yppasswd.x protocol definition, the
YPPASSWDPROC_UPDATE procedure takes two arguments: a V7-style passwd
structure containing updated user information and the user's existing
unencrypted (cleartext) password. Since
rpc.yppasswdd is supposed to handle update requests
from remote NIS client machines, this means that
yppasswd(1) and similar
client programs will in fact be transmitting users' cleartext passwords over
the network.
This is not a problem for password updates since the plaintext password sent with the update will no longer be valid once the new encrypted password is put into place, but if the user is only updating his or her 'GECOS' information or shell, then the cleartext password sent with the update will still be valid once the update is completed. If the network is insecure, this cleartext password could be intercepted and used to gain unauthorized access to the user's account.
| February 8, 1996 | midnightbsd-3.1 |