freebsd-dev/usr.sbin/mrouted/mrouted.8

.\"COPYRIGHT 1989 by The Board of Trustees of Leland Stanford Junior University.
.\"
.\" $FreeBSD$
.\"
.Dd May 8, 1995
.Dt MROUTED 8
.UC 5
.Sh NAME
.Nm mrouted
.Nd IP multicast routing daemon
.Sh SYNOPSIS
.Nm mrouted 
.Op Fl c Ar config_file
.Op Fl d Op Ar debug_level
.Op Fl p
.Sh DESCRIPTION
.Nm Mrouted 
is an implementation of the Distance-Vector Multicast Routing
Protocol (DVMRP), an earlier version of which is specified in RFC-1075.
It maintains topological knowledge via a distance-vector routing protocol
(like RIP, described in RFC-1058), upon which it implements a multicast
datagram forwarding algorithm called Reverse Path Multicasting.
.Pp
.Nm Mrouted 
forwards a multicast datagram along a shortest (reverse) path tree
rooted at the subnet on which the datagram originates. The multicast
delivery tree may be thought of as a broadcast delivery tree that has
been pruned back so that it does not extend beyond those subnetworks
that have members of the destination group. Hence, datagrams
are not forwarded along those branches which have no listeners of the
multicast group. The IP time-to-live of a multicast datagram can be
used to limit the range of multicast datagrams.
.Pp
In order to support multicasting among subnets that are separated by (unicast)
routers that do not support IP multicasting, 
.Nm 
includes support for
"tunnels", which are virtual point-to-point links between pairs of 
multicast routers
located anywhere in an internet.  IP multicast packets are encapsulated for
transmission through tunnels, so that they look like normal unicast datagrams
to intervening routers and subnets.  The encapsulation 
is added on entry to a tunnel, and stripped off
on exit from a tunnel.
The packets are encapsulated using the IP-in-IP protocol
(IP protocol number 4).
Older versions of 
.Nm
tunneled using IP source routing, which puts a heavy load on some
types of routers.
This version does not support IP source route tunnelling.
.Pp
The tunnelling mechanism allows 
.Nm 
to establish a virtual internet, for
the purpose of multicasting only, which is independent of the physical
internet, and which may span multiple Autonomous Systems.  This capability
is intended for experimental support of internet multicasting only, pending
widespread support for multicast routing by the regular (unicast) routers.
.Nm Mrouted 
suffers from the well-known scaling problems of any distance-vector
routing protocol, and does not (yet) support hierarchical multicast routing.
.Pp
.Nm Mrouted 
handles multicast routing only; there may or may not be unicast routing
software running on the same machine as 
.Nm mrouted .
With the use of tunnels, it
is not necessary for 
.Nm 
to have access to more than one physical subnet
in order to perform multicast forwarding.
.Pp
The following options are available:
.Bl -tag -width indent
.It Fl c Ar config_file
Specify an alternative file for configuration commands. Default is
.Pa /etc/mrouted.conf .
.It Fl d Op Ar debug_level
If no
.Fl d
option is given, or if the debug level is specified as 0, 
.Nm
detaches from the invoking terminal.  Otherwise, it remains attached to the
invoking terminal and responsive to signals from that terminal.
Regardless of the debug level, 
.Nm 
always writes warning and error messages to the system
log demon.  The
.Fl debug-level
argument is a comma-seperated list of any of the following:
.Bl -tag -width indent
.It "packet"
Display the type, source and destination of all packets sent or received.
.It "pruning"
Display more information about prunes sent or received.
.It "routing"
Display more information about routing update packets sent or received.
.It "route_detail"
Display routing updates in excruciating detail.  This is generally way too
much information.
.It "neighbors"
Display information about neighbor discovery.
.It "cache"
Display insertions, deletions and refreshes of entries in
the kernel forwarding cache.
.It "timeout"
Debug timeouts and periodic processes.
.It "interface"
Display information about interfaces and their configuration.
.It "membership"
Display information about group memberships on physical interfaces.
.It "traceroute"
Display information about multicast traceroute requests
passing through this router.
.It "igmp"
Display IGMP operation including group membership and querier election.
.It "icmp"
Monitor ICMP handling.
.It "rsrr"
Monitor RSRR operation.
.El
.Pp
Upon startup,
.Nm
writes its pid to the file
.Pa /var/run/mrouted.pid .
.Sh CONFIGURATION
.Nm Mrouted 
automatically configures itself to forward on all multicast-capable
interfaces, i.e., interfaces that have the IFF_MULTICAST flag set (excluding
the loopback "interface"), and it finds other DVMRP routers directly reachable
via those interfaces.  To override the default configuration, or to add
tunnel links to other multicast routers,
configuration commands may be placed in
.Pa /etc/mrouted.conf
(or an alternative file, specified by the
.Fl c
option).
.Pp
The file format is free-form; whitespace (including newlines) is not
significant.
The file begins with commands that apply to
.Nm mrouted Ns 's
overall operation or set defaults.
.Bl -tag -width indent
.It cache_lifetime Ar secs
Specifies, in seconds, the lifetime of a multicast forwarding cache
entry in the kernel.  Multicast forwarding cache entries in the kernel
are checked every
.Ar secs
seconds, and are refreshed if the source is still
active or deleted if not.  Care should be taken when setting this value,
as a low value can keep the kernel cache small at the cost of "thrashing"
the cache for periodic senders, but high values can cause the kernel
cache to grow unacceptably large.  The default is 300 seconds (5 minutes).
.It prune_lifetime Ar secs
Sepcifies, in seconds, the average lifetime of prunes that are sent towards
parents.  The actual lifetimes will be randomized in the range
[.5\fIsecs\fP,1.5\fIsecs\fP].  The default is 7200 (2 hours).  Smaller values
cause less state to be kept both at this router and the parent, at the
cost of more frequent broadcasts.  However, some routers (e.g. mrouted <3.3
and all currently known versions of cisco's IOS) do not use the
DVMRP generation ID to determine that a neighbor has rebooted.  Prunes
sent towards these neighbors should be kept short, in order to shorten
the time to recover from a reboot.  For use in this situation, the
prune_lifetime keyword may be specified on an interface as described
below.
.It noflood
.Nm Mrouted
uses a DVMRP optimization to prevent having to keep individual routing tables
for each neighbor; part of this optimization is that
.Nm
assumes that it is the forwarder for each of its attached subnets on
startup.  This can cause duplicates for a short period (approximately
one full route report interval), since both the router that just
started up and the proper forwarder will be forwarding traffic.  This
behavior can be turned off with the noflood keyword;
.Nm
will not assume that it is the forwarder on startup.
Turning on noflood can cause black holes on restart, which will generally
last approximately one full route report interval.
The noflood keyword can also be specified on individual interfaces.
.It rexmit_prunes Ar [on|off]
.Nm Mrouted 's
default is to retransmit prunes on all point-to-point interfaces
(including tunnels) but no multi-access interfaces.  This option
may be used to make the default on (or off) for all interfaces.
The rexmit_prunes keyword can also be specified on individual interfaces.
.It name Ar "boundary-name scoped-addr/mask-len"
Associates
.Ar boundary-name
with the boundary described by
.Ar scoped-addr/mask-len ,
to help make interface configurations
more readable and reduce repetition in the configuration file.
.El
.Pp
The second section of the configuration file, which may optionally
be empty, describes options that apply to physical interfaces.
.Bl -tag -width indent
.It phyint Ar "local-addr|ifname"
The phyint command does nothing by itself; it is simply a place holder
which interface-specific commands may follow.  An interface address or
name may be specified.
.It disable
Disables multicast forwarding on this interface.  By default,
.Nm
discovers all locally attached multicast capable interfaces and forwards
on all of them.
.It netmask Ar netmask
If the kernel's netmask does not accurately reflect
the subnet (e.g. you're using proxy-ARP in lieu of IP subnetting), use the
netmask command to describe the real netmask.
.It altnet Ar network/mask-len
If a phyint is attached to multiple IP subnets, describe each additional subnet
with the altnet keyword.  This command may be specified multiple times
to describe multiple subnets.
.It igmpv1
If there are any IGMPv1 routers on the phyint, use the \fBigmpv1\fP
keyword to force
.Nm
into IGMPv1 mode.  All routers on the phyint
must use the same version of IGMP.
.It force_leaf
Force
.Nm
to ignore other routers on this interface.
mrouted will never send or accept neighbor probes or
route reports on this interface.
.El
.Pp
In addition, the common vif commands described later may all be used on
a phyint.
.Pp
The third section of the configuration file, also optional, describes
the configuration of any DVMRP tunnels this router might have.
.Bl -tag -width indent
.It tunnel Ar "local-addr|ifname" Ar "remote-addr|remote-hostname"
This command establishes a DVMRP tunnel between this host (on the interface
described by
.Ar local-addr
or
.Ar ifname )
and a remote host (identified by
.Ar remote-addr
or
.Ar remote-hostname ).
A remote hostname may only be used if
it maps to a single IP address.
A tunnel must be configured on both routers before it can be used.

Be careful that the unicast route to the remote address goes out the
interface specified by the
.Ar "local-addr|ifname"
argument.  Some UNIX
kernels rewrite the source address of
.Nm mrouted 's
packets on their way out to contain the address of the transmission
interface.  This is best assured via a static host route.
.El
.Pp
The common vif commands described below
may all be used on tunnels or phyints.
.Bl -tag -width indent
.It metric Ar m
The metric is the "cost" associated with receiving a datagram on the given
interface or tunnel; it may be used to influence the choice of routes.
The metric defaults to 1.  Metrics should be kept as small as possible,
because DVMRP cannot route along paths with a sum of metrics greater
than 31.
.It advert_metric Ar m
The advert_metric is the "cost" associated with sending a datagram
on the given interface or tunnel; it may be used to influence the choice
of routes.  The advert_metric defaults to 0.  Note that the effective
metric of a link is one end's metric plus the other end's advert_metric.
.It threshold Ar t
The threshold is the minimum IP time-to-live required for a multicast datagram
to be forwarded to the given interface or tunnel.  It is used to control the
scope of multicast datagrams.  (The TTL of forwarded packets is only compared
to the threshold, it is not decremented by the threshold.  Every multicast
router decrements the TTL by exactly 1.)  The default threshold is 1.

In general, all multicast routers
connected to a particular subnet or tunnel should
use the same metric and threshold for that subnet or tunnel.
.It rate_limit Ar r
The rate_limit option allows the network administrator to specify a 
certain bandwidth in Kbits/second which would be allocated to multicast
traffic.  It defaults 0 (unlimited).
.It boundary Ar "boundary-name|scoped-addr/mask-len"
The boundary option allows an interface
to be configured as an administrative boundary for the specified
scoped address. Packets belonging to this address will not
be forwarded on a scoped interface.  The boundary option accepts either
a name or a boundary spec.  This command may be specified several times
on an interface in order to describe multiple boundaries.
.It passive
No packets will be sent on this link or tunnel until we hear from the other
end.  This is useful for the "server" end of a tunnel that goes over
a dial-on-demand link; configure the "server" end as passive and
it will not send its periodic probes until it hears one from the other
side, so will not keep the link up.  If this option is specified on both
ends of a tunnel, the tunnel will never come up.
.It noflood
As described above, but only applicable to this interface/tunnel.
.It prune_lifetime Ar secs
As described above, but only applicable to this interface/tunnel.
.It rexmit_prunes Ar "[on|off]"
As described above, but only applicable to this interface/tunnel.
Recall that prune retransmission
defaults to on on point-to-point links and tunnels, and off on
multi-access links.
.It allow_nonpruners
By default,
.Nm
refuses to peer with DVMRP neighbors that
do not claim to support pruning.  This option allows such peerings
on this interface.
.It notransit
A specialized case of route filtering; no route learned from an interface
marked "notransit" will be advertised on another interface marked
"notransit".  Marking only a single interface "notransit" has no meaning.
.It accept|deny Ar "(route/mask-len [exact])+" Op bidir
The
.Li accept
and
.Li deny
commands allow rudimentary route filtering.  The
.Li accept
command causes
.Nm
to accept only the listed routes on the configured interface; the
.Li deny
command causes
.Nm
to accept all but the listed routes.
Only one of
.Li accept
or
.Li deny
commands may be used on a given interface.

The list of routes follows the
.Li accept
or
.Li deny
keyword.  If the keyword
.Ar exact
follows a route, then only that route is matched; otherwise, that route
and any more specific route is matched.  For example,
.Li deny 0/0
denys all routes, while
.Li deny 0/0 exact
denys only the default route.  The default route may also be specified
with the
.Li default
keyword.

The
.Ar bidir
keyword enables bidirectional route filtering; the filter will be applied
to routes on both output and input.  Without the
.Ar bidir
keyword,
.Li accept
and 
.Li deny
filters are only applied on input.  Poison reverse routes are never
filtered out.
.El
.Pp
.Nm Mrouted
will not initiate execution if it has fewer than two enabled vifs,
where a vif (virtual interface) is either a physical multicast-capable
interface or a tunnel.  It will log a warning if all of its vifs are
tunnels; such an 
.Nm
configuration would be better replaced by more
direct tunnels (i.e. eliminate the middle man).
.Sh "EXAMPLE CONFIGURATION"
This is an example configuration for a mythical multicast router at a big
school.
.sp
.nf
#
# mrouted.conf example
#
# Name our boundaries to make it easier
name LOCAL 239.255.0.0/16
name EE 239.254.0.0/16
#
# le1 is our gateway to compsci, don't forward our
#     local groups to them
phyint le1 boundary EE
#
# le2 is our interface on the classroom net, it has four
#     different length subnets on it.
# note that you can use either an ip address or an
# interface name
phyint 172.16.12.38 boundary EE altnet 172.16.15.0/26
	altnet 172.16.15.128/26 altnet 172.16.48.0/24
#
# atm0 is our ATM interface, which doesn't properly
#      support multicasting.
phyint atm0 disable
#
# This is an internal tunnel to another EE subnet
# Remove the default tunnel rate limit, since this
#   tunnel is over ethernets
tunnel 192.168.5.4 192.168.55.101 metric 1 threshold 1
	rate_limit 0
#
# This is our tunnel to the outside world.
# Careful with those boundaries, Eugene.
tunnel 192.168.5.4 10.11.12.13 metric 1 threshold 32
	boundary LOCAL boundary EE
.fi
.Sh SIGNALS
.Nm Mrouted 
responds to the following signals:
.Bl -tag -width indent
.It HUP
Restarts
.Nm mrouted .
The configuration file is reread every time this signal is evoked.
.It INT
Terminate execution gracefully (i.e., by sending
good-bye messages to all neighboring routers).
.It TERM
Same as INT.
.It USR1
Dump the internal routing tables to
.Pa /var/tmp/mrouted.dump .
.It USR2
Dump the internal cache tables to
.Pa /var/tmp/mrouted.cache .
.It QUIT
Dump the internal routing tables to stderr (only if
.Nm 
was invoked with a non-zero debug level). 
.El
.Pp
For convenience in sending signals,
.Nm
writes its pid to
.Pa /var/run/mrouted.pid
upon startup.
.Sh EXAMPLE
The routing tables look like this:
.nf
.ft C

Virtual Interface Table
 Vif  Local-Address                    Metric  Thresh  Flags
  0   36.2.0.8      subnet: 36.2/16       1       1    querier
                    groups: 224.0.2.1
                            224.0.0.4
                   pkts in: 3456
                  pkts out: 2322323

  1   36.11.0.1     subnet: 36.11/16      1       1    querier
                    groups: 224.0.2.1      
                            224.0.1.0      
                            224.0.0.4      
                   pkts in: 345
                  pkts out: 3456

  2   36.2.0.8      tunnel: 36.8.0.77     3       1   
                     peers: 36.8.0.77 (3.255)
                boundaries: 239.0.1/24
                          : 239.1.2/24
                   pkts in: 34545433
                  pkts out: 234342

  3   36.2.0.8	    tunnel: 36.6.8.23	  3       16

Multicast Routing Table (1136 entries)
 Origin-Subnet   From-Gateway    Metric Tmr In-Vif  Out-Vifs
 36.2                               1    45    0    1* 2  3*
 36.8            36.8.0.77          4    15    2    0* 1* 3*
 36.11                              1    20    1    0* 2  3*
 .
 .
 .

.fi
.Pp
In this example, there are four vifs connecting to two subnets and two
tunnels.  The vif 3 tunnel is not in use (no peer address). The vif 0 and
vif 1 subnets have some groups present; tunnels never have any groups.  This
instance of 
.Nm
is the one responsible for sending periodic group
membership queries on the vif 0 and vif 1 subnets, as indicated by the
"querier" flags. The list of boundaries indicate the scoped addresses on that
interface. A count of the no. of incoming and outgoing packets is also
shown at each interface.
.Pp
Associated with each subnet from which a multicast datagram can originate
is the address of the previous hop router (unless the subnet is directly-
connected), the metric of the path back to the origin, the amount of time
since we last received an update for this subnet, the incoming vif for
multicasts from that origin, and a list of outgoing vifs.  "*" means that
the outgoing vif is connected to a leaf of the broadcast tree rooted at the
origin, and a multicast datagram from that origin will be forwarded on that
outgoing vif only if there are members of the destination group on that leaf.
.Pp
.Nm Mrouted 
also maintains a copy of the kernel forwarding cache table. Entries
are created and deleted by 
.Nm mrouted .
.Pp
The cache tables look like this:
.nf
.ft C

Multicast Routing Cache Table (147 entries)
 Origin             Mcast-group     CTmr  Age Ptmr IVif Forwvifs
 13.2.116/22        224.2.127.255     3m   2m    -  0    1 
>13.2.116.19
>13.2.116.196
 138.96.48/21       224.2.127.255     5m   2m    -  0    1 
>138.96.48.108
 128.9.160/20       224.2.127.255     3m   2m    -  0    1 
>128.9.160.45
 198.106.194/24     224.2.135.190     9m  28s   9m  0P  
>198.106.194.22

.fi 
.Pp
Each entry is characterized by the origin subnet number and mask and the
destination multicast group.
.Pp
The 'CTmr' field indicates the lifetime
of the entry.  The entry is deleted from the cache table
(or refreshed, if traffic is flowing)
when the timer decrements to zero.  The 'Age' field is the time since
this cache entry was originally created.  Since cache entries get refreshed
if traffic is flowing, routing entries can grow very old.
.Pp
The 'Ptmr' field is simply a dash if no prune was sent upstream, or the
amount of time until the upstream prune will time out.
.Pp
The 'Ivif' field indicates the
incoming vif for multicast packets from that origin.  Each router also
maintains a record of the number of prunes received from neighboring
routers for a particular source and group. If there are no members of
a multicast group on any downward link of the multicast tree for a
subnet, a prune message is sent to the upstream router. They are
indicated by a "P" after the vif number.
.Pp
The Forwvifs field shows the
interfaces along which datagrams belonging to the source-group are
forwarded. A "p" indicates that no datagrams are being forwarded along
that interface. An unlisted interface is a leaf subnet with no
members of the particular group on that subnet. A "b" on an interface
indicates that it is a boundary interface, i.e. traffic will not be
forwarded on the scoped address on that interface.
.Pp
An additional line with a ">" as the first character is printed for
each source on the subnet.  Note that there can be many sources in
one subnet.
An additional line with a "<" as the first character is printed
describing any prunes received from downstream dependent neighbors
for this subnet and group.
.Sh FILES
.Bl -tag -width /var/tmp/mrouted.cache -compact
.It Pa /etc/mrouted.conf
.It Pa /var/run/mrouted.pid
.It Pa /var/tmp/mrouted.dump
.It Pa /var/tmp/mrouted.cache
.El
.Sh SEE ALSO
.Xr map-mbone 8 ,
.Xr mrinfo 8 ,
.Xr mtrace 8
.Pp
DVMRP is described, along with other multicast routing algorithms, in the
paper "Multicast Routing in Internetworks and Extended LANs" by S. Deering,
in the Proceedings of the ACM SIGCOMM '88 Conference.
.Sh AUTHORS
.An Steve Deering ,
.An Ajit Thyagarajan ,
.An Bill Fenner .