SMAILRTRS(5) Local SMAILRTRS(5)NAME
/etc/smail/routers - smail router configuration
THE ROUTERS FILE
The routers file describes the operations, and their order, for han‐
dling remote addresses. For example, some possibilities are to look up
domain names in a routing database, to match a special domain-literal
form, or call a program to return a list of neighbouring hosts.
Generic Routers Attributes
The following list describes the possible generic attributes for router
file entries:
always
type: boolean
Routers will not always find a complete match for a particular
domain name. For example, if a routing database has a route to the
domain ``amdahl.com'' but not to the domain
``futatsu.uts.amdahl.com'' then the routing driver might return the
route to just ``amdahl.com''. Generally, the route for the longest
match of the target domain is used. However, if the always
attribute is set, then any match found by the current router will
be used in preference to routes that might have been returned by
routers later in the router list. This is useful for hardwiring a
certain number of routes within a small database. For example,
this is useful for Internet sites that gateway for a small number
of UUCP sites in the UUCP zone, or for hardwiring routes to inter‐
nal SMTP hosts that may also be reachable over the public Internet
via other gateways.
driver
type: string
The driver attribute names a specific set of low-level functions
which will do the work of routing remote mail. This attribute is
required.
method
type: string
transport
type: string
The router entry has the option of specifically setting a transport
to use for remote delivery. If the driver does not do so, then
either a method or a transport attribute must exist to specify how
the mail is to be delivered. A method attribute specifies a file
which contains a table relating domain names to transports. A
transport attribute specifies a specific transport. If the method
file does not contain a match for all domains, then the transport
named with the transport attribute is used. The format of a method
file is given in the smailmeth(5) manual page.
THE ROUTER DRIVERS
This section details the usage and driver-specific attributes for all
of the router drivers distributed with smail.
The Pathalias Driver
WARNING: The pathalias router driver is in dire need of a full inter‐
nals rewrite. It should not be used unless necessary. Use the rewrite
router driver instead.
The pathalias router driver computes routes from a database in the
style produced by the pathalias program. It has the following driver-
specific attributes:
domain
type: string
A colon-separated list of default domains for hosts in the data‐
base. Targets ending in any of these domains will have the domain
stripped (including a preceding dot) before the database is
searched. A target containing only the domain (e.g., just
``.uucp'') will remain unchanged.
file
type: string
This defines the name of the file containing the database. Except
when it does not make sense as a result of the proto used, if this
does not begin with a slash (`/'), it will be referenced relative
to the smail_lib_dir directory.
interval
type: number
The number of seconds to wait between open retries.
optional
type: boolean
If set, then if the open fails, assume an empty alias file. This
is useful for optional databases. If not set, and if tryagain is
also not set, then the message will be frozen in the error queue
and tagged with a configuration error.
This option might be useful in a networking environment where work‐
stations may be all configured with the same routers file and thus
have the option of using a private alias file, without the neces‐
sity of creating such a file on each host.
proto
type: string
Names the protocol used in opening and searching the database.
Possibilities are discussed below.
reopen
type: boolean
If set, the pathalias will be closed and reopened after each call
to the pathalias driver. This is useful in environments where file
descriptors are scarce but many databases are desired.
required
type: string
A colon-separated list of domains which the target name is required
to be within for this router entry to match the target and act to
route it. The names are not stripped before the database is
searched, unless they are also specified by the domain attribute.
retries
type: number
The maximum number of retries for opening a file. This is useful
on systems without an atomic rename system call. On such systems
there will be a window of vulnerability when new databases are
moved into place where no file will exist.
try type: string
A list of domains to be stripped only if the target was not found
in the database without the domain stripped. (This is currently
not supported.)
tryagain
type: boolean
If set, then if the open fails, the resolution of local addresses
will be attempted at a later time. If not set, and if optional is
also not set, then the message will be frozen in the error queue
and tagged with a configuration error.
This is useful in a networking environment where failure to open a
database (such as a remote YP/NIS database) may be a result of the
database server machine being down or temporarily inaccessible.
The current list of possible values for the proto attribute is:
bsearch Use a binary search to look through a sorted file arranged as
lines which begin with a key and are followed by the value
associated with the key, separated by a colon or whitespace.
dbm Use the BSD dbm(3x) or ndbm(3x) [or db(3x) ndbm emulation] rou‐
tines to search for the key. The keys and data in the dbm
database must end in a NUL byte. If only the dbm library is
available then only one dbm database can be used by smail,
while the ndbm routines will allow any number of databases.
However, it is always okay for multiple routers and directors
to use the same dbm database, if this is useful.
lsearch Use a linear search using the same read routine used to read
config files. `#'-style comments are allowed and the beginning
of each file entry should be the key, followed by whitespace or
a colon character. The rest of the entry should be the value
associated with the key.
yp Use the Sun YP service to access a paths database stored on a
remote machine. In this case the value for the file attribute
is of the form:
domain_name:database_name
where the ``domain_name:'' portion is optional and defaults to
the default YP domain for the local host.
aliasyp This is a variant of the YP protocol that is compatible with
the standard Sun mail.aliases YP service. This database has a
different format from other databases which must be taken into
account when sending requests. Typically this is not useful
for a path database.
nisplus Look up the key in a remote Sun NIS+ (Sun NIS version 3) data‐
base. NIS+ is not compatible with NIS (otherwise known as YP).
The file parameter must be a NIS+ indexed name, which is
described in the nis(1) manual page. The search string is
replaced with ``%s'' where it can be filled in by smail's NIS+
lookup routine. See the nisplus entry in the smail manual page
and the more detailed example in the nisplus entry in the
smaildrct(X_MAN_5_EXT) manual page.
All database lookups are either independent of case or, when case inde‐
pendent lookups are impossible, case-folded. Thus, keys in DBM or YP
databases must be in lower case.
As an example of the use of the domain, try, and required driver
attributes, if the domain value is ``uucp'' then any domain name that
ends in ``.uucp'' will have the ``.uucp'' part removed before being
looked up. Alternately, if the the try value is ``uucp'' then
``.uucp'' is stripped only if the original was not found. If the
required value is ``uucp'' then a domain name is not a candidate for a
match unless it ends in ``.uucp''. The effects of domain and try are
not cumulative and a domain name is applied to the required attribute
value before being applied to domain and try in that order.
Note that the length of any stripped string is added to the count of
characters matched for purposes of deciding which router had the most
complete match.
A sample pathalias router entry is:
pathalias:
transport=uux,
driver=pathalias;
file=paths,
proto=bsearch,
domain=uucp
An example of a pathalias file for the site ``walldrug'' is:
.amdahl.com amdahl!%s
amdahl amdahl!%s
dgcad namei!glotz!nsavax!dgcad!%s
glotz namei!glotz!%s
hoptoad hoptoad!%s
kgbvax kgbvax!%s
kremvax kremvax!%s
namei namei!%s
nsavax namei!glotz!nsavax!%s
This data base associates a host or domain name, on the left hand side,
with a path, on the right hand side. The right hand side should be a
UUCP-style `!'-path ending in a ``%s''. The format is intended to be
more general, with the ``%s'' showing where to put a user name. Indeed
pathalias(8) can produce paths involving both right and left operators
such as:
decwrl decwrl!%s@ucbvax
This form is not recommended in general because of a historical dis‐
agreement over the precedence of ``!'' and ``@'' which is not always
resolved correctly by the pathalias program. By standardising on
UUCP-style `!'-paths, as will be produced from the USENET maps, we
avoid many routing ambiguities. If the path is just ``%s'', it denotes
a name for the local host.
The Uuname Driver
Some sites in the UUCP zone may wish to use a direct path to their
neighbours, without relying on a pathalias database. Alternately, a
site may wish to use the pathalias routes to their neighbours only in
the case that a domain address is given (i.e., ``neighbour.uucp'').
A simple driver that meets these requirements is the uuname driver,
which reads a list of hostnames from the output of a shell command
(normally /usr/bin/uuname) and uses this list to match neighbouring
hostnames.
An example entry for a site that wishes to route to their neighbours
for domain names that may end in ``.uucp'' is:
uuname:
transport=uux,
driver=uuname;
cmd=/usr/bin/uuname,
domain=uucp
An alternative is a site that wishes to bypass the pathalias router
only for explicit target matches, so that the form ``neighbour.UUCP
will be routed through, as an example, pathalias. This can be done by
not specifying a domain, as in:
uuname:
transport=uux,
driver=uuname;
cmd=/usr/bin/uuname
The uuname driver also supports the domain, required, and try
attributes, and all three domain-style attributes can be colon sepa‐
rated lists. See the pathalias driver for more details on these
attributes.
In addition, there is a string attribute, statfile which names a file
which can be used to determine when the output of the shell command
will change. For example, when used with the uuname command this would
normally be one of /usr/lib/uucp/L.sys, /usr/lib/uucp/Systems, or
/etc/uucp/Systems depending on whether a site is using HoneyDanBer or
stock V7-style UUCP. If a statfile attribute is defined, then smail
daemons that expect to deliver multiple messages will cache the output
of the shell command. The specified pathname will then be checked at
regular intervals to determine if the command should be reissued to
refresh the cached information.
The Queryprogram Driver
If no other router driver meets a particular routing requirement, but a
shell script or C program can be written which does, the queryprogram
driver can be used to access this script or program. This driver runs
a program, with arguments, and examines its exit status and standard
output for information. The program should return an exit status of
zero to indicate a match for the target host, non-zero otherwise.
Optionally, this program can write a path and a transport on its stan‐
dard output to provide additional information. If a path is not
returned then a one-hop path to the target host is assumed. The pro‐
gram is always executed as an unprivileged user.
The queryprogram driver supports the domain, required, and try
attributes, as defined in the description of the pathalias driver. The
program and its arguments are described in terms of a cmd attribute
which is a string consisting of white-space-separated tokens which are
expanded to form a program name and the arguments to pass to that pro‐
gram. The variable $host can be used to obtain the target host string.
White space or punctuation from an expansion variable do not delimit
arguments (i.e. variables expand into one argument). Two additional
boolean attributes affect the driver's behaviour:
read_path
type: boolean
If this attribute is enabled, then a !-path is read as the first
white-space delimited field in the first line of output from the
program. If no output is produced by the program, or this
attribute is not enabled, then a one-hop path to the target host is
assumed. A path should consist of a list of hostnames separated by
exclamation point (`!') characters.
read_transport
type: boolean
If this attribute is enabled, then a transport is read as a white-
space delimited field from the first line of output from the pro‐
gram. If the read_path attribute is also enabled, then the trans‐
port is taken from the second field, otherwise it is taken from the
first. If no output was generated by the program, or the field did
not exist in the output, then the normal methods for determining
the transport are used.
A trivial example of the use of this driver is the router file entry:
match_bar:
driver=queryprogram,
transport=bar;
cmd="/bin/test X${lc:host} = Xfoo"
This router will use the program test(1) to determine if the target
host is ``foo''. If this is the target host, then it is assumed to be
a neighbouring host which can be reached over the transport ``bar''.
A more complex example is to create a shell script such as:
# match_fubar.sh - match hosts on the fubar networks
case "$1" in
foo) echo "foo foo_transport"; exit 0;;
bar) echo "foo!bar foo_transport"; exit 0;;
curds) echo "curds curds_transport"; exit 0;;
whey) echo "curds!whey curds_transport"; exit 0;;
esac
exit 1
By setting up a router file entry such as:
match_fubar:
driver=queryprogram;
cmd="/bin/sh $lib_dir/match_fubar.sh ${lc:host}",
read_path, read_transport
a complete, though not particularly optimal, router is created which
can match the hosts ``foo'', ``bar'', ``curds'', and ``whey'' and cause
messages to these hosts to be delivered using one of the transports
``foo_transport'' or ``curds_transport''.
The queryprogram driver is not very efficient, as it requires a fork()
& exec() for each host. However, it is a very simple to construct a
new router using this driver, so it is useful for prototyping. In gen‐
eral, a new form of routing requirement is most efficiently handled by
writing a new router driver.
The Gethostbyaddr Driver
Also in a network environment, it is useful to be able to specify
explicit Internet addresses using a target such as: ``[192.168.2.1]''
which might be the internet address for a private internal host to
which e-mail might be routed to.
The gethostbyaddr driver matches targets of this form, which is defined
by square brackets surrounding only digits and dot characters.
The gethostbyaddr driver has the following private attributes:
check_for_local
type: boolean
If set, see if the hostname returned by gethostbyaddr(3n) matches
one of the known names for the local host. The address
INADDR_LOOPBACK, which normally represents 127.0.0.1 (the address
most commonly used for host loopback interfaces), also counts as a
local host. If a local host is detected then the remainder address
will be re-parsed and re-routed thus short-cutting one hop in the
delivery of the message.
NOTE: This attribute is set by default.
fail_if_error
type: boolean
If set, then any literal IP address target which does not fit the
form of an internet address is considered an error. If not set,
then such addresses are merely not matched by the driver.
NOTE: This attribute is set by default.
The Bind Driver
The bind driver uses the Berkeley Internet Name Domain server, to
resolve target domain names. It is fully compliant with the RFC 974
and RFC 1134 standards, and uses MX records. This router driver is
available only on systems that have a 4.3BSD compatible or newer
resolver library. It is highly recommended that this router driver be
used for machines that are on the Internet, rather than using the geth‐
ostbyname router.
The following private attributes vary the behaviour of the bind driver:
defer_no_connect
type: boolean
If this boolean attribute is set, then we must be able to connect
to the nameserver. If attempts to connect to the nameserver fail,
then routing is retried at a later time, on the assumption that the
nameserver is only down temporarily. If this attribute is not set,
then a failure to connect to the nameserver will cause the router
to be ignored. This allows a bind router to be optional based on
whether or not the nameserver exists.
NOTE: This attribute is set by default.
local_mx_okay
type: boolean
If this boolean attribute is not set, then it is considered an
error for an unfiltered (i.e. lowest priority) MX resource record
to be encountered which points to the local host. If this
attribute is set, then such an MX resource record will be permitted
but will be ignored -- i.e. it will cause the address not to be
matched.
NOTE: This attribute is NOT set by default. If you think you need
to turn this flag on to make your configuration work the way you
want then you probably have your router instances in the wrong
order, so move your UUCP router(s) above the BIND router(s).
WARNING: This attribute is deprecated and will disappear in future
versions of Smail.
defnames
type: boolean
This boolean attribute is passed to the BIND resolver routines as
the ``RES_DEFNAMES'' flag. If set, then hostnames which do not
contain dots will have a default domain appended to them before
lookups. For example, this allows for hosts on a local area net‐
work to be referenced using the basename of the host rather than
requiring the full domain name.
ignore_domains
type: string
If this string is set, then it is used as a colon-separated list of
domain names. If a hostname ends in any of these domains, then the
router will not match the host, and will not result in any queries.
This can be set to a list of domains that are known not to exist,
such as ``uucp'' and ``bitnet'', to prevent expensive lookups which
will never be successful.
domain_required
type: boolean
If this boolean attribute is set, then a hostname with only one
domain component (i.e., no dots) will not be matched. This should
probably be set if defnames is not set, to prevent lookups for UUCP
hostnames that will not be found in the DNS.
NOTE: This attribute is set by default.
mx_only
type: boolean
If this boolean attribute is set, which it should normally be by
default, then match a host only if it has a valid MX record.
Note that an MX record is only valid if it points to a canonical
hostname. See however the descriptions of smtp_bad_mx_targets and
allow_one_mx_target_cname_hack in smailconf(5).
If all domain names that can accept mail have an MX record pointing
to a valid mail exchanger host, then no attempts will made to send
mail to other ``hosts'' such as printers or terminal concentrators
which have only A records. However RFC 974 still requires that
SMTP mailers route to hosts without MX records (and indeed far too
much of the Public Internet still relies on this several decade old
silliness), so if a router instance using the gethostbyname driver
follows a router using this driver then Smail can still meet the
ancient RFC 974 requirements.
Note that RFC 2821 does not allow an Internet mailer to route to a
host with an A record if there were also any MX records for that
same host, even if all those MX records were unusable for any rea‐
son. This means that a router using this bind driver MUST occur
before any instance of any router using the gethostbyname driver if
Smail is to conform with this requirement.
NOTE: This attribute is set by default.
WARNING: This attribute is deprecated and will disappear in future
versions of Smail.
mx_domains
type: string
This is a list of colon separated domains which, if they match the
tail of a fully qualified domain name, will force mx_only mode for
that domain. Negative matches can be forced by preceding the
domain with an exclamation mark (!).
For example this allows you to send to any host in the york.ac.uk
domain but only to MX registered hosts in the ac.uk domain:
mx_domains = ``!york.ac.uk:ac.uk''
gateways
type: string
A list of gateways and corresponding domains for explicit routing
using the bind processing logic for the gateway address. The string
consists of a sequence of lists. Each list consists of strings
separated by colons, and the lists are separated by a string con‐
sisting of a single `+' character. Each list consists of the name
of a gateway, followed by a list of domains that should be routed
to that gateway. The gatewayed domains will match either against a
complete address, or against the terminating domain of an address.
For example:
gateways = "uknet.ac.uk:uucp:+:\
earn-relay.ac.uk:bitnet:ie:earn"
In this example, if an address of the form ``<anything>.uucp'' is
being processed, it will be converted to ``uknet.ac.uk'' before any
other processing is done. The new name will then be looked up in
the DNS and routed accordingly. Thus mail for ``<anything>.uucp''
will always be sent to ``uknet.ac.uk'', by whatever route is regis‐
tered in the DNS for that address. Similarly, mail for the domains
``bitnet'', ``ie'', and ``earn'' will be routed to ``earn-
relay.ac.uk''.
WARNING: This attribute is deprecated and will disappear in future
versions of Smail.
dns_search
type: boolean
If set allow the resolver to search its domain list for matches.
This experimental and might not have the effect you expect depend‐
ing on your resolver search capabilities.
widen_domains
type: string
A list of domains, separated by colons, to be tacked onto the end
of the address if it isn't found directly in the DNS. For example:
widen_domains = "cam.ac.uk:ac.uk"
With this setting, an address such as ``psy.ox'' will first be
tried as it stands, then ``psy.ox.cam.ac.uk'' will be tried, and if
that fails, ``psy.ox.ac.uk'' will be tried.
All matches found by the bind driver are considered full matches, for
the purposes of prioritising results from different routers. Matches
are considered full even in the presence of wildcards in MX records.
The Gethostbyname Driver
In a networking environment, hostnames on a network can be matched
using this driver, which calls the gethostbyname(3N) library routine.
This routine is only available on systems that supply a BSD compatible
networking library. Driver attributes for this driver include domain,
required, and try attributes, which are compatible with the pathalias
and uuname drivers.
Additional private attributes are:
only_local_domain
type: boolean
Do not match hosts that have a domain component (i.e., that contain
a dot) - i.e. domain names. The domain attribute is applied before
the only_local_domain attribute, so hosts that have a domain compo‐
nent consisting only of one of the strings listed with the domain
attribute can be matched.
the gethostbyname driver will only match a target host completely.
The hostname given to the transport will be the host as given by
the h_name field of the hostent structure.
Note that RFC 2821 does not allow an Internet mailer to route to a
host with an A record if there were also any MX records for that
same host, even if all those MX records were unusable for any rea‐
son. This means that a router using the bind driver MUST occur
before any instance of any router using this gethostbyname driver
if Smail is to conform with this requirement.
NOTE: The hostname given to gethostbyname() will be down-cased, so
that upper-case or mixed-case names can be matched correctly.
The Smarthost Driver
Sometimes it is handy to be able to redirect mail to a host that you
don't know about to some other host that has more routing information.
This other host is called a ``smart'' host, and can be named by using
the smarthost driver.
The smarthost driver has the following private driver attribute:
path
type: string
This define a host or UUCP-style `!'-path path that defines the
smart host.
If no transport or method attribute is given for this router file
entry, this address will be resent through all the routing drivers. An
exception is that an address cannot be sent to the smarthost driver
twice, as this would mean that the path to the smarthost was not known.
The global configuration variable auth_domains affects the smarthost
router driver by restricting the set of domain names that can be
matched by the smarthost router to domain names that are outside of the
domains listed in auth_domains.
For example, if the site ``namei'' wanted to use ``amdahl''s routing
database for mail delivery to non-neighbouring sites they could use a
router file entry of:
smart_host: driver=smarthost; path=amdahl
or
smart_host: transport=uusmtp,
driver=smarthost; path=amdahl
NOTE: At the present time, a transport or method attribute is
required, as the software is not yet in place that can rescan the
routers in the required ways.
Then, a recipient address of ``Ted.Husted.Jr@walldrug.uucp'' will be
rewritten so that the hostname is ``amdahl'' and the remaining part of
the address is ``Ted.Hustead.Jr@walldrug.uucp.''
Alternately, in the second form for the entry, the transport, uux,
would be known immediately and be passed the hostname ``amdahl'' imme‐
diately, with a $user value of ``Ted.Hustead.Jr@walldrug.uucp.''
By specifying a UUCP-style `!'-path it is possible to route mail to a
smart host that you would not otherwise be able to route to. For exam‐
ple, if the machine ``glotz'' wished to use ``amdahl'' as its smart
host, through its neighbouring site ``namei'', it could use:
smart_host: driver=smarthost; path=namei!amdahl
or
smart_host: transport=demand, driver=smarthost; path=namei!amdahl
It is possible to set the path and/or the transport in the config file.
This is done by setting the config file attributes smart_path and
smart_transport. For example, the config file could contain the fol‐
lowing:
smart_path = amdahl
smart_transport = uusmtp
Then, if the entry in the routers file contained:
smart_host: driver=smarthost
the configuration would be essentially the same as in the second
smart_host example above. This makes it possible to share copies of
the router file among several machines, with the smart host configura‐
tion specific to one machine being set in its private configuration
file.
These config file attributes are used only if the path driver attribute
is not given in the smart_host entry. If the smart_path attribute from
the config file is used, the smart_transport will override a transport
or method attribute.
The Rewrite Driver
The rewrite driver is a pathalias-like driver that matches addresses
against a database and gives back a new address that is subsequently
re-parsed.
The rewrite driver supports the following private attributes in exactly
the same way as in the pathalias router: domain, file, interval,
optional, proto, required, retries, reopen, tryagain.
The algorithm of the rewrite router driver is given a target domain
name and looks it up in the database using the same search strategy as
the pathalias driver. The best match wins, as with pathalias. If a
match is found, the information on the matching line specifies how to
rewrite the address, or whether not to rewrite it at all. If the
address is not rewritten, the driver behaves as if it had not been
matched at all. If the address is rewritten, the driver sets the input
address up as the parent (just as with an alias) of the rewritten
address, and returns the rewritten address to be re-parsed. The
match_always attribute has no effect on this driver - partial matches
are treated like full matches.
Each line in the rewrite database has the following format:
domain flag format
The ``domain'' portion is matches exactly as with the pathalias driver.
It can be a simple hostname, fully qualified domain name, or a partial
domain name beginning with a dot (`.').
The flag is one of `-', meaning rewriting should not be done (the
router then behaves as if the target had not been matched), or `+',
meaning rewriting should be done, using the specified format if the
remainder was a simple mailbox (user name), and leaving the remainder
unchanged if it was more complex.
The format is a string to be variable expanded to produce the rewritten
address. The expansion is performed in a context in which $user refers
to the remainder and $host refers to the target. See the smail(5) man‐
ual page for more information about the syntax of string expansions.
For example:
.special.com + special-$user
special.com -
.foo.org + $user-$host
foo.org + ${lookup:user:lsearch{foo-aliases}{$value}{postmaster}}
foo.com + ${lookup:user:lsearch{foo-aliases}{$value}{"$user@$host"}}
Note the last entry, with the double quotes, will produce a nice look‐
ing bounce making this driver a superb way of implementing virtual
domains without requiring extra directors or other such integral con‐
figuration changes.
The Rerouter Driver
The rerouter driver is an experimental router driver that can do opti‐
misation of UUCP-zone !-paths addresses. For example, given an address
of:
host1!host2!host3!host4!host!user
The rerouter driver could realize that a more efficient path exists to
``host'' and rewrite this as:
host6!host!user
Before you say to yourself how neat this would be and how much of a
savings this would be for the world, you must realize that such re-
routing is dangerous and should be configured-in only with a great deal
of consideration and attention to detail in your re-routing database.
The rerouter driver is probably only of use in a privately managed net‐
work.
The rerouter driver supports the following private attributes in
exactly the same way as in the pathalias router: domain, file, inter‐
val, optional, proto, required, retries, reopen, tryagain.
The rerouter driver also supports the following unique private
attributes:
matchall
type: boolean
If set, reroute all bang path addresses. The re-routing database
isn't used at all in this mode, so even hosts that have an exclu‐
sion flag (`-') in the database will be rerouted. WARNING: Use
this only for debugging purposes. Unconditional re-routing is an
offence against the community!
matchlocal
type: boolean
If set, then address containing forms for the local host (within
the bounds of the required attributes) are considered candidates
for re-routing. The path starting with the component directly fol‐
lowing the last occurrence of the local hostname is used re-rout‐
ing. For example, if the local hostname is ``geminix'', then the
input path:
host1!host2!geminix!host!user
might be rewritten as:
host4!host!user
matchdb
type: boolean
If set, a database lookup is done for the target. Otherwise, the
reroute database isn't used. This must be set of you intend to set
the file attribute to point to a database.
bounceonly
type: boolean
This attribute combines with the matchlocal attribute to do limited
re-routing that bypasses multiple occurrences of the local host,
but only for bounce messages (messages generated for returning
errors to the originator of a message). This attribute affects
only the use of the matchlocal attribute, and has no other affect
on the rerouter driver.
The rerouter driver operates only on pure !-path addresses. For such
addresses, the !-path is scanned from right to left for hostnames that
are considered candidates for re-routing. If such a host is found, the
path is truncated to contain everything from that component to the end
of the path. In the case of a match resulting from the matchlocal
attribute, the matching component is also removed from the path.
Given a truncated path generated by the rerouter driver, the address is
passed back into the smail address resolver engine to be routed based
on the other routers. This goes all the way back to the beginning of
the address resolution process, rather than continuing on with the list
of routers, as would happen for a non-matching target hostname.
The Rerouter Database Format
If a database file is used to define candidate hosts for re-routing,
the file will be presumed to contain entries consisting of a hostname
or a domain name beginning with a period, followed by `+' or `-'. For
example, the database file might contain:
.do.main +
.bar +
.foo.bar -
host1.foo.bar +
host2 +
host2.abc +
host3 +
host4 +
If a match is found on the given hostname, then the `+' or `-' is used
to indicate whether the hostname is a candidate for re-routing. A `+'
indicates that the host is a candidate and a `-' indicates that the
host is not a candidate.
The rules for finding an entry in the database are:
1. Look for an exact match of the target hostname. If not found, and
the domain begins with a dot, look for a match without the dot.
2. If that fails, strip the first component of the domain, leaving the
initial dot on the second component. Look for a match on the
resulting name. Repeat this step until a match is found or no com‐
ponents remain.
This lookup algorithm, together with the `+' or `-' field can be used
to enable re-routing for all but a certain subset of hosts within a
domain.
General Recommendations for Rerouter Driver Use
While re-routing can be useful, and can improve the flow of mail
through the UUCP zone, it can be disastrous if done wrong.
Here are two situations where it can be useful to reroute mail:
1. If your site is on the Internet, and it is a backbone for a domain
of UUCP sites that are connected to your site, then your site might
have access to routing informations that isn't available to the
UUCP sites (MX records etc.). Using shortcuts over the Internet
instead of sending the mail along the original slow and maybe
expensive (telephone charges) path can be a major advantage.
2. If one or more links that in your direct neighbourhood (one or two
hops away) have changed and the new UUCP maps haven't yet propa‐
gated widely enough to make other sites aware of these changes, you
could use re-routing to temporarily correct the incoming bang paths
that were generated from the old maps.
However, before you enable re-routing you should make sure that your
routing databases are really truly accurate. It's not enough that you
install the posted USENET maps automatically. These maps are notori‐
ously inaccurate, and often contain very unrealistic link costs,
resulting in non-optimal routes. Therefore, it is necessary to check
all routes that might be used for re-routing line by line whenever it
is rebuilt from the maps.
If you can't or don't want to do this, don't use re-routing! Or at
least only use it for sites in your direct neighbourhood. Re-routing
with a poorly maintained routing database is a disservice to the net!
The simplest way to setup re-routing is to limit the set of hosts that
you use for re-routing and ensure that the routes to all of those hosts
are stable and accurate. Making a separate paths file-based router is
the simplest way of ensuring that stable paths are used.
Also, if you think that you have the administrative resources for re-
routing, do so only when you are willing to keep in touch with the
maintainers of other re-routing sites in your area. Mail loops between
two re-routing sites are a serious problem that can be reduced only by
discussing changes you want to make to your routing in advance of actu‐
ally making them.
So in short, folks, don't use re-routing if it isn't absolutely neces‐
sary. Please don't use it as a means to inflate your ego. Instead,
maintaining a re-routing host is a responsible task that needs lots of
time and determination. Please restrain yourself, for the sake of fast
and flawless mail delivery.
Rerouter Hostname Non-Uniqueness
One more point, which is actually an important one, is that hostnames
within UUCP-zone are not always unique. Just because you recognizee a
hostname on the end of path does not mean that it is the host that you
think it is. To avoid conflicts arising from non-unique names it is
best to avoid re-routing for hostnames that are not qualified within a
domain.
The only safe way to reroute an address is to reroute from left to
right rather than right to left, and to stop as soon as a host is found
which is not a re-routing candidate. However, such a policy greatly
limits the usefulness of re-routing or at least it greatly increases
the number of entries in your re-routing database that are necessary
for re-routing to be useful.
DEFAULT CONFIGURATION
The default internal routers configuration can be viewed by typing
smail -oR /no-such-file -v -bP ROUTERS
The default router configuration is normally something like this:
rewrite
a router using the rewrite driver and the always attribute to
allow forced rewriting of addresses using the /etc/smail/rewrite
database.
inet_addrs
match literals forms of IP internet addresses, deliver with the
smtp transport.
bind_hosts
a router using the bind driver to route using normal Internet
DNS MX records and deliver via the inet_zone_bind_smtp trans‐
port.
inet_hosts
a router using the gethostbyname driver to match hosts on an IP
network, delivering with the smtp transport.
uucp_neighbours
a router using the uuname driver to route to direct UUCP neigh‐
bours via the uux transport.
FILES
/etc/smail/routers
Optional configuration for smail routers, i.e., configured methods
for resolving or routing to remote hosts. This file replaces the
compiled-in router configuration.
/etc/smail/aliases
A file of aliases for local addresses.
/etc/smail/paths
A file of paths to remote hosts.
/etc/smail/lists/
A directory of mailing list files.
~/.forward
Lists of forwarding addresses for local users.
/var/log/smail/logfile
A log of smail transactions.
/var/log/smail/paniclog
A log of configuration or system errors encountered by smail.
SEE ALSOsmail(5), smailconf(5), smaildrct(5), smailmeth(5), smailqual(5),
smailrtry(5), smailtrns(5), smail(8). Smail Administration and Instal‐
lation Guide. DARPA Internet Requests for Comments RFC 821, RFC 822,
RFC 974, RFC 976, and RFC 1123.
BUGS
Colons cannot be included in the value of a list element.
Database files cannot contain ``#'' in the left-hand field.
COPYRIGHT
Copyright (C) 1987, 1988 Ronald S. Karr and Landon Curt Noll Copyright
(C) 1992 Ronald S. Karr
See a file COPYING, distributed with the source code, or type smail -bc
to view distribution rights and restrictions associated with this soft‐
ware.
Smail-3 RELEASE-3_2_0_115 SMAILRTRS(5)