netgroup

- list of network groups

Synopsis

/etc/netgroup

Description

A netgroup defines a network-wide group of hosts and users. Use a netgroup to restrict access to shared NFS filesystems and to restrict remote login and shell access.

Network groups are stored in a network information services, such as LDAP or NIS, not in a local file.

This manual page describes the format for a file that is used to supply input to a program such as ldapaddent(1M) for LDAP or makedbm(1M) for NIS. These programs build maps used by their corresponding network information services.

Each line of the file defines the name and membership of a network group. The line should have the format:

groupname member... 

The items on a line can be separated by a combination of one or more spaces or tabs.

The groupname is the name of the group being defined. This is followed by a list of members of the group. Each member is either another group name, all of whose members are to be included in the group being defined, or a triple of the form:

(hostname,username,domainname)

In each triple, any of the three fields hostname, username, and domainname, can be empty. An empty field signifies a wildcard that matches any value in that field. Thus:

everything ( , ,this.domain)

defines a group named "everything" for the domain "this.domain" to which every host and user belongs.

The domainname field refers to the domain in which the triple is valid, not the domain containing the host or user. In fact, applications using netgroup generally do not check the domainname. Therefore, using

(,,domain)

is equivalent to

(,,)

You can also use netgroups to control NFS mount access (see share_nfs(1M)) and to control remote login and shell access (see hosts.equiv(4)). You can also use them to control local login access (see passwd(4), shadow(4), and compat in nsswitch.conf(4)).

When used for these purposes, a host is considered a member of a netgroup if the netgroup contains any triple in which the hostname field matches the name of the host requesting access and the domainname field matches the domain of the host controlling access.

Similarly, a user is considered a member of a netgroup if the netgroup contains any triple in which the username field matches the name of the user requesting access and the domainname field matches the domain of the host controlling access.

Note that when netgroups are used to control NFS mount access, access is granted depending only on whether the requesting host is a member of the netgroup. Remote login and shell access can be controlled both on the basis of host and user membership in separate netgroups.

Files

/etc/netgroup

Used by a network information service's utility to construct a map or table that contains netgroup information. For example, ldapaddent(1M) uses /etc/netgroup to construct an LDAP container.

Note that the netgroup information must always be stored in a network information service, such as LDAP or NIS. The local file is only used to construct a map or table for the network information service. It is never consulted directly.

See Also

ldapaddent(1M), makedbm(1M), share_nfs(1M), innetgr(3C), hosts(4), hosts.equiv(4), nsswitch.conf(4), passwd(4), shadow(4)

Notes

netgroup requires a network information service such as LDAP or NIS.

Applications may make general membership tests using the innetgr() function. See innetgr(3C).

Because the "-" character will not match any specific username or hostname, it is commonly used as a placeholder that will match only wildcarded membership queries. So, for example:

onlyhosts    (host1,-,our.domain) (host2,-,our.domain)
onlyusers    (-,john,our.domain) (-,linda,our.domain)

effectively define netgroups containing only hosts and only users, respectively. Any other string that is guaranteed not to be a legal username or hostname will also suffice for this purpose.

Use of placeholders will improve search performance.

When a machine with multiple interfaces and multiple names is defined as a member of a netgroup, one must list all of the names. See hosts(4). A manageable way to do this is to define a netgroup containing all of the machine names. For example, for a host "gateway" that has names "gateway-subnet1" and "gateway-subnet2" one may define the netgroup:

gateway (gateway-subnet1, ,our.domain) (gateway-subnet2, ,our.domain)

and use this netgroup “gateway” whenever the host is to be included in another netgroup.