/usr/sbin/ipaddrsel -f file
Use the ipaddrsel utility to configure the IPv6 default address selection policy table. The policy table is a longest-matching-prefix lookup table that is used for IPv6 source address selection and for destination address ordering when resolving names to AF_INET6 addresses. For a description of how the policy table is used for source address selection, see inet6(7P). For a description of how the policy table is used for destination address ordering, see getaddrinfo(3SOCKET).
The unmodified policy table is valid for all typical IPv6 deployments. Modify the table only if a circumstance exists for which the default behavior of the IPv6 source address selection or destination address ordering mechanism is unsatisfactory. See the section for examples of such circumstances. You should carefully consider your addressing strategy before you change the table from the provided default.
When the ipaddrsel command is issued without any arguments, the address selection policy currently in use is printed. The format of the output is compatible with the format of the configuration file that the -f option accepts.
The configuration file that the -f option accepts can contain either comment lines or policy entries. Comment lines have a '#' character as the first non-blank character. and they are ignored by the ipaddrsel utility. Policy entry lines have the following format:
prefix/prefix_length precedence label [# comment ]
The prefix must be an IPv6 prefix in a format consistent with inet(3SOCKET). The prefix_length is an integer ranging from 0 to 128. The IPv6 source address selection and destination address ordering algorithms determine the precedence or label of an address by doing a longest-prefix-match lookup using the prefixes in this table, much like next-hop determination for a destination is done by doing a longest-prefix-match lookup using an IP routing table.
The precedence is a non-negative integer that represents how the destination address ordering mechanism will sort addresses returned from name lookups. In general, addresses with a higher precedence will be in front of addresses with a lower precedence. Other factors, such as destinations with undesirable source addresses can, however, override these precedence values.
The label is a string of at most fifteen characters, not including the NULL terminator. The label allows particular source address prefixes to be used with destination prefixes of the same label. Specifically, for a particular destination address, the IPv6 source address selection algorithm prefers source addresses whose label is equal that of the destination.
The label may be followed by an optional comment.
The file must contain a default policy entry, which is an entry with "::0/0" as its prefix and prefix_length. This is to ensure that all possible addresses match a policy.
The ippadrsel utility supports the following options:
example# ipaddrsel -d example# ipaddrsel > /etc/inet/ipaddrsel.conf
Example 1 The Default Policy in /etc/inet/ipaddrsel.conf
The following example is the default policy that is located in /etc/inet/ipaddrsel.conf:
# Prefix Precedence Label ::1/128 50 Loopback ::/0 40 Default 2002::/16 30 6to4 ::/96 20 IPv4_Compatible ::ffff:0.0.0.0/96 10 IPv4
Example 2 Assigning a Lower Precedence to Link-local and Site-local Addresses
By default, the destination address ordering rules sort addresses of smaller scope before those of larger scope. For example, if a name resolves to a global and a site-local address, the site local address would be ordered before the global address. An administrator can override this ordering rule by assigning a lower precedence to addresses of smaller scope, as the following table demonstrates.
# Prefix Precedence Label ::1/128 50 Loopback ::/0 40 Default 2002::/16 30 6to4 fec0::/10 27 Site-Local fe80::/10 23 Link-Local ::/96 20 IPv4_Compatible ::ffff:0.0.0.0/96 10 IPv4
Example 3 Assigning Higher Precedence to IPv4 Destinations
By default, IPv6 addresses are ordered in front of IPv4 addresses in name lookups. ::ffff:0.0.0.0/96 has the lowest precedence in the default table. In the following example, IPv4 addresses are assigned higher precedence and are ordered in front of IPv6 destinations:
# Prefix Precedence Label ::1/128 50 Loopback ::/0 40 Default 2002::/16 30 6to4 ::/96 20 IPv4_Compatible ::ffff:0.0.0.0/96 60 IPv4
Example 4 Ensuring that a particular source address is only used when communicating with destinations in a particular network.
The following policy table assigns a label of 5 to a particular source address on the local system, 2001:1111:1111::1. The table assigns the same label to a network, 2001:2222:2222::/48. The result of this policy is that the 2001:1111:1111::1 source address will only be used when communicating with destinations contained in the 2001:2222:2222::/48 network. For this example, this network is the "ClientNet", which could represent a particular client's network.
# Prefix Precedence Label ::1/128 50 Loopback 2001:1111:1111::1/128 40 ClientNet 2001:2222:2222::/48 40 ClientNet ::/0 40 Default 2002::/16 30 6to4 ::/96 20 IPv4_Compatible ::ffff:0.0.0.0/96 10 IPv4
This example assumes that the local system has one physical interface, and that all global prefixes are assigned to that physical interface.
ipaddrsel returns the following exit values:
See attributes(5) for descriptions of the following attributes:
nscd(1M), inet(3SOCKET), getaddrinfo(3SOCKET), ipaddrsel.conf(4), attributes(5), inet6(7P)
The ipnodes cache kept by nscd(1M) contains addresses that are ordered using the destination address ordering algorithm, which is one of the reasons why ipaddrsel is called before nscd in the boot sequence. If ipaddrsel is used to change the address selection policy after nscd has started, you should invalidate the nscd ipnodes cache invalidated by invoking the following command:
example# /usr/sbin/nscd -i ipnodes