This section describes rationales for some of the design issues that went into
ipmasq
.
The rules files ipmasq
utilizes are named according to a common
scheme consisting of a capital letter, a two digit number, and a short
description of what the rule is doing. As the names of the rules files are run
through sort
, the letter and two digit number determine the order
in which rules will be run.
The capital letter denotes what kind of action the rule takes, according to the following list:
ipmasq
host fall into this category.
ipmasq
host.
More information about the rules files can be found in the comments of the rules files themselves.
The rules files supplied with ipmasq
have been broken down to a
fine granularity. This granularity is designed to allow individual portions to
be overridden easily. For instance, if all input rules were shipped in one
rules file, in order to override the input rule for external interfaces, the
remaining rules in the file would have to be duplicated.
New rules files should contain commands suitable for setting up the operation of one service. This is to both keep a reasonable level of granularity, and to allow those rules files to be self-contained entities.
ipmasq
makes a few assumptions about the networks it is
masquerading between (these hold for the majority of cases):
default-if
program by selecting the interface that contains
the system's default route.
ipmasq
has run. Should it change, ipmasq
must be
re-run. Please see Integrating with Other
Systems, Chapter 4 for more information on the specifics of how to do this
for your particular connection method.
Should any of these assumptions prove false for your specific network, please see Dealing with Oddball Networks, Chapter 3.
Ipmasq User's Manual
Brian Bassettbrianb@debian.org
osamu@debian.org