Traffic Shaping/Control
Tom
Eastep
2005-02-19
2001-2005
Thomas M. Eastep
Permission is granted to copy, distribute and/or modify this
document under the terms of the GNU Free Documentation License, Version
1.2 or any later version published by the Free Software Foundation; with
no Invariant Sections, with no Front-Cover, and with no Back-Cover
Texts. A copy of the license is included in the section entitled
GNU Free Documentation
License
.
Introduction
Shorewall does not do any type of Traffic Shaping/Bandwidth
management itself but it does contain some facilities to intergrate with
traffic shaping/control solutions. In order to use traffic shaping with
Shorewall, it is essential that you get a copy of the Linux Advanced Routing and Shaping
HOWTO, version 0.3.0 or later or The Traffic Control
HOWTO. It is also necessary to be running Linux Kernel 2.4.18 or
later. Shorewall traffic shaping support consists of the following:
A new TC_ENABLED parameter in
/etc/shorewall.conf.
A new CLEAR_TC parameter in
/etc/shorewall.conf (Added in Shorewall 1.3.13). When Traffic Shaping
is enabled (TC_ENABLED=Yes), the setting of this variable determines
whether Shorewall clears the traffic shaping configuration during
Shorewall [re]start and Shorewall stop.
/etc/shorewall/tcrules - A file
where you can specify firewall marking of packets. The firewall mark
value may be used to classify packets for traffic
shaping/control.
/etc/shorewall/tcstart - A
user-supplied file that is sourced by Shorewall during
shorewall start
and which you can use to define your
traffic shaping disciplines and classes. I have provided a sample that
does table-driven CBQ shaping but if you read the traffic shaping
sections of the HOWTO mentioned above, you can probably code your own
faster than you can learn how to use my sample. I personally use
HTB (see
below). HTB support may eventually become an integral part of
Shorewall since HTB is a lot simpler and better-documented than CBQ.
As of 2.4.20, HTB is a standard part of the kernel but iproute2 must
be patched in order to use it.
In tcstart, when you want to run the tc
utility,
use the run_tc function supplied by shorewall if you want tc errors to
stop the firewall.
You can generally use off-the-shelf traffic shaping scripts by
simply copying them to /etc/shorewall/tcstart. I use The Wonder Shaper (HTB
version) that way (i.e., I just copied wshaper.htb to
/etc/shorewall/tcstart and modified it according to the Wonder Shaper
README). WARNING: If you use use
Masquerading or SNAT (i.e., you only have one external IP address)
then listing internal hosts in the NOPRIOHOSTSRC variable in the
wshaper[.htb] script won't work. Traffic shaping occurs after SNAT has
already been applied so when traffic shaping happens, all outbound
traffic will have as a source address the IP addresss of your
firewall's external interface.
/etc/shorewall/tcclear - A
user-supplied file that is sourced by Shorewall when it is clearing
traffic shaping. This file is normally not required as Shorewall's
method of clearing qdisc and filter definitions is pretty
general.
Shorewall allows you to start traffic shaping when Shorewall itself
starts or it allows you to bring up traffic shaping when you bring up your
interfaces.
To start traffic shaping when Shorewall starts:
Set TC_ENABLED=Yes and CLEAR_TC=Yes
Supply an /etc/shorewall/tcstart script to configure your
traffic shaping rules.
Optionally supply an /etc/shorewall/tcclear script to stop
traffic shaping. That is usually unnecessary.
If your tcstart script uses the fwmark
classifier, you can mark packets using entries in
/etc/shorewall/tcrules.
To start traffic shaping when you bring up your network interfaces,
you will have to arrange for your traffic shaping configuration script to
be run at that time. How you do that is distribution dependent and will
not be covered here. You then should:
Set TC_ENABLED=Yes and CLEAR_TC=No
Do not supply /etc/shorewall/tcstart or /etc/shorewall/tcclear
scripts.
If your tcstart script uses the fwmark
classifier, you can mark packets using entries in
/etc/shorewall/tcrules.
Kernel Configuration
This screen shot show how I've configured QoS in my Kernel:
/etc/shorewall/tcrules
The fwmark classifier provides a convenient way to classify packets
for traffic shaping. The /etc/shorewall/tcrules file provides a means for
specifying these marks in a tabular fashion.
Normally, packet marking occurs in the PREROUTING chain before any
address rewriting takes place. This makes it impossible to mark inbound
packets based on their destination address when SNAT or Masquerading are
being used. Beginning with Shorewall 1.3.12, you can cause packet marking
to occur in the FORWARD chain by using the MARK_IN_FORWARD_CHAIN option in
shorewall.conf.
Unlike entries in /etc/shorewall/rules,
evaluation of entries in /etc/shorewall/tcrules
continues after a match. So the final mark assigned to each packet is
determined by the last matching entry
in the /etc/shorewall/tcrules file.
Columns in the file are as follows:
MARK - Specifies the mark value is to be assigned in case of a
match. This is an integer in the range 1-255. Beginning with Shorewall
version 1.3.14, this value may be optionally followed by
:
and either F
or P
to
designate that the marking will occur in the FORWARD or PREROUTING
chains respectively. If this additional specification is omitted, the
chain used to mark packets will be determined by the setting of the
MARK_IN_FORWARD_CHAIN option in shorewall.conf.
This possible values in this field were expanded in Shorewall
version 2.2.0:
If your kernel and iptables include CONNMARK support then
you can also mark the connection rather than the packet
The mark value may be optionally followed by "/" and a
mask value (used to determine those bits of the connection
mark to actually be set).
The mark and optional mask are then followed by one
of:
C: Mark the connection in the chain determined by
the setting of MARK_IN_FORWARD_CHAIN
CF: Mark the conneciton in the FORWARD
chain
CP: Mark the connection in the PREROUTING
chain.
A classification of the form <major>:<minor>
where <major> and <minor> are integers. Corresponds to
the 'class' specification in these traffic shaping
modules:
- atm
- cbq
- dsmark
- pfifo_fast
- htb
- prio
Classification always occurs in the POSTROUTING
chain. This feature requires Shorewall 2.2.0 and requires that
your kernel and iptables include CLASSIFY target support.
RESTORE[/mask] -- restore the packet's mark from the
connection's mark using the supplied mask if any. Your kernel and
iptables must include CONNMARK support. As iabove, may be followed
by ":P" or ":F
SAVE[/mask] -- save the packet's mark to the connection's
mark using the supplied mask if any. Your kernel and iptables must
include CONNMARK support. As above, may be followed by ":P" or
":F
CONTINUE -- don't process any more marking rules in the
table. As above, may be followed by ":P" or ":F".
SOURCE - The source of the packet. If the packet originates on
the firewall, place $FW
in this column. Otherwise, this
is a comma-separated list of interface names, IP addresses, MAC
addresses in Shorewall Format and/or Subnets.
Examples eth0
192.168.2.4,192.168.1.0/24
Beginning with Shorewall version 2.2.2, "$fw" may be optionally
followed by a colon (":") and a host/net address or an address
range.
DEST -- Destination of the packet. Comma-separated list of IP
addresses and/or subnets.
PROTO - Protocol - Must be the name of a protocol from
/etc/protocol, "ipp2p", a number or "all". For "ipp2p", your kernel
and iptables must have ipp2p match support from Netfilter
Patch_o_matic_ng.
PORT(S) - Destination Ports. A comma-separated list of Port
names (from /etc/services), port numbers or port ranges (e.g., 21:22);
if the protocol is icmp
, this column is interpreted as
the destination icmp type(s). If the protocol is "ipp2p", then this
column is interpreted as an ipp2p option without the leading "--"
(default "ipp2p"). For a list of value ipp2p options, as root type
iptables -m ipp2p --help.
CLIENT PORT(S) (Renamed SOURCE PORT(S) in Shorewall 2.2.0) -
(Optional) Source port(s). If omitted, any source port is acceptable.
Specified as a comma-separate list of port names, port numbers or port
ranges.
USER (Added in Shorewall version 1.4.10) - (Optional) This
column may only be non-empty if the SOURCE is the firewall itself.
When this column is non-empty, the rule applies only if the program
generating the output is running under the effective user and/or
group. It may contain :
[<user name or number>]:[<group name or
number>]
The colon is optionnal when specifying only a user.
Examples : john: / john / :users / john:users
TEST (added in Shorewall version 2.2.0). Defines a test on the
existing packet or connection mark. The rule will match only if the
test returns true. Tests have the format
[!]<value>[/<mask>][:C]
where
! Inverts the test (not equal)
<value> Value of the packet or
connection mark.
<mask> A mask to be applied to
the mark before testing
:C Designates a connection mark. If omitted, the packet
mark's value is tested.
All packets arriving on eth1 should be marked with 1. All packets
arriving on eth2 and eth3 should be marked with 2. All packets
originating on the firewall itself should be marked with 3.
#MARK SOURCE DESTINATION PROTO PORT(S) SOURCE PORT(S) USER/GROUP TEST
1 eth1 0.0.0.0/0 all
2 eth2 0.0.0.0/0 all
2 eth3 0.0.0.0/0 all
3 fw 0.0.0.0/0 all
All GRE (protocol 47) packets not originating on the firewall and
destined for 155.186.235.151 should be marked with 12.
#MARK SOURCE DESTINATION PROTO PORT(S) SOURCE PORT(S) USER/GROUP TEST
12 0.0.0.0/0 155.182.235.151 47
All SSH packets originating in 192.168.1.0/24 and destined for
155.186.235.151 should be marked with 22.
#MARK SOURCE DESTINATION PROTOCOL PORT(S) SOURCE PORT(S) USER/GROUP TEST
22 192.168.1.0/24 155.182.235.151 tcp 22
My Current Setup
I am currently using the HTB version of The Wonder Shaper (I just
copied wshaper.htb to /etc/shorewall/tcstart and modified it as shown in
the Wondershaper README). WonderShaper DOES NOT USE THE
/etc/shorewall/tcrules file.
My Old Setup
I have also run with the following set of hand-crafted rules in my
/etc/shorewall/tcstart file.
run_tc qdisc add dev eth0 root handle 1: htb default 30
run_tc class add dev eth0 parent 1: classid 1:1 htb rate 384kbit burst 15k
echo Added Top Level Class -- rate 384kbit
run_tc class add dev eth0 parent 1:1 classid 1:10 htb rate 140kbit ceil 384kbit burst 15k prio 1
run_tc class add dev eth0 parent 1:1 classid 1:20 htb rate 224kbit ceil 384kbit burst 15k prio 0
run_tc class add dev eth0 parent 1:1 classid 1:30 htb rate 20kbit ceil 384kbit burst 15k quantum 1500 prio 1
echo Added Second Level Classes -- rates 140kbit, 224kbit, 20kbit
run_tc qdisc add dev eth0 parent 1:10 pfifo limit 5
run_tc qdisc add dev eth0 parent 1:20 pfifo limit 10
run_tc qdisc add dev eth0 parent 1:30 pfifo limit 5
echo Enabled PFIFO on Second Level Classes
run_tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 1 fw classid 1:10
run_tc filter add dev eth0 protocol ip parent 1:0 prio 0 handle 2 fw classid 1:20
run_tc filter add dev eth0 protocol ip parent 1:0 prio 1 handle 3 fw classid 1:30
echo Defined fwmark filters
My tcrules file that went with this tcstart file is shown in Example
1 above. When I was using these rules:
I wanted to allow up to 140kbits/second for traffic outbound
from my DMZ (eth1 -- note that the ceiling is set to 384kbit so
outbound DMZ traffic can use all available bandwidth if there is no
traffic from the local systems or from my laptop or firewall).
My laptop (which at that time connected via eth3) and local
systems (eth2) could use up to 224kbits/second.
My firewall could use up to 20kbits/second.
Once www.shorewall.net was moved off-site, I no longer needed these
shaping rules and The Wonder Shaper does all that I now require.