updated readme and added script

README

@@ -0,0 +1,5 @@
+Source code and script that go with 
+
+Journey to the Center of the Linux Kernel: Traffic Control, the QoS
+
+http://wiki.linuxwall.info/doku.php/en:ressources:dossiers:networking:traffic_control

lnw_gateway_tc.sh

@@ -0,0 +1,174 @@
+#! /usr/bin/env bash
+#
+### BEGIN INIT INFO
+# Provides:          lnw_gateway_qos.sh
+# Required-Start:    $syslog
+# Required-Stop:     $syslog
+# Should-Start:      $local_fs
+# Should-Stop:       $local_fs
+# Default-Start:     2
+# Default-Stop:      0 1 6
+# Short-Description: starts/stops lnw_gateway_qos.sh
+# Description:       load or stop the Traffic Control rules on the network interface
+### END INIT INFO
+
+NETCARD=eth0    # network card to apply the QoS to
+UPLINK=1600     # upload in kbits/s
+DOWNLINK=16000  # download in kbits/s
+
+TC=/sbin/tc
+IPT=/sbin/iptables
+
+
+case "$1" in
+	start)
+		echo "~~~~ LOADING $NETCARD TRAFFIC CONTROL RULES FOR `uname -n` ~~~~"
+		echo
+		echo "#-cleanup"
+		$TC qdisc del dev $NETCARD root
+		$IPT -t mangle -F
+
+        # the r2q value is calculated to let ~3 full packets pass before
+        # rescheduling, so r2q = (UPLINK / (3 * MTU * 8))
+		echo "#-define a HTB root qdisc"
+		$TC qdisc add dev $NETCARD root handle 1: htb default 999 r2q 40
+
+        # this is the main branch that sets the total bandwidth available to the leaves
+		echo "#--uplink - rate $UPLINK kbit ceil $UPLINK kbit"
+		$TC class add dev $NETCARD parent 1:0 classid 1:1 htb \
+            rate $(($UPLINK))kbit ceil $(($UPLINK))kbit
+
+
+        # INTERACTIVE CLASS: for low latency, high priority packets such as VoIP and DNS
+        # it has little bandwidth assigned to it, but pass before everything else
+		echo "#---interactive - id 100 - rate $(( $UPLINK / 10 )) kbit ceil $(( $UPLINK )) kbit"
+		$TC class add dev $NETCARD parent 1:1 classid 1:100 htb \
+            rate $(( $UPLINK / 10 ))kbit ceil $(( $UPLINK / 3 ))kbit \
+            burst 5k prio 1
+
+        # this class transmits uses a pfifo before exiting
+		echo "#--- ~ sub interactive: pfifo"
+		$TC qdisc add dev $NETCARD parent 1:100 handle 1100: pfifo limit 10
+
+        # filter definition, that will catch all the packets carrying the mark 100
+		echo "#--- ~ interactive filter"
+		$TC filter add dev $NETCARD parent 1:0 protocol ip prio 1 \
+            handle 100 fw flowid 1:100
+
+        # netfilter/iptables rules, will mark all UDP packets
+		echo "#--- ~ netfilter rule - all UDP traffic at 100"
+		$IPT -t mangle -A POSTROUTING -o $NETCARD -p udp -j CONNMARK --set-mark 100
+
+
+        # TCP ACKs class: it's important to let the ACKs leave the network as fast as possible
+        # when a host of the network is downloading.
+		echo "#---tcp acks - id 200 - rate $(( $UPLINK / 5 )) kbit ceil $(( $UPLINK )) kbit"
+		$TC class add dev $NETCARD parent 1:1 classid 1:200 htb \
+            rate $(( $UPLINK / 5 ))kbit ceil $(( $UPLINK / 10 ))kbit burst 15k prio 2
+
+		echo "#--- ~ sub tcp acks: pfifo"
+		$TC qdisc add dev $NETCARD parent 1:200 handle 1200: pfifo limit 10
+
+		echo "#--- ~ filtre tcp acks"
+		$TC filter add dev $NETCARD parent 1:0 protocol ip prio 2 handle 200 fw flowid 1:200
+
+		echo "#--- ~ netfilter rule for TCP ACKs will be loaded at the end"
+
+
+        # SSH class: for outgoing connections to avoid lag when somebody else is downloading
+        # however, an SSH connection cannot fill up the connection to more than 70%
+		echo "#---ssh - id 300 - rate $(( $UPLINK / 10 )) kbit ceil $(( $UPLINK / 10 * 7 )) kbit"
+		$TC class add dev $NETCARD parent 1:1 classid 1:300 htb \
+            rate $(( $UPLINK / 10 ))kbit ceil $(( $UPLINK / 8 ))kbit burst 15k prio 3
+
+        # SFQ will mix the packets if there are several SSH connections in parallel
+        # and ensure that none has the priority
+		echo "#--- ~ sub ssh: sfq"
+		$TC qdisc add dev $NETCARD parent 1:300 handle 1300: sfq perturb 10 limit 32
+
+		echo "#--- ~ ssh filter"
+		$TC filter add dev $NETCARD parent 1:0 protocol ip prio 3 handle 300 fw flowid 1:300
+
+		echo "#--- ~ netfilter rule - SSH at 300"
+		$IPT -t mangle -A POSTROUTING -o $NETCARD -p tcp --tcp-flags SYN,ACK SYN,ACK \
+            --dport 22 -j CONNMARK --set-mark 300
+
+
+        # HTTP class: All web browsing (80/443) gets half the bandwidth and can borrow full
+		echo "#---http branch - id 400 - rate $(( $UPLINK / 2 )) kbit ceil $UPLINK kbit"
+		$TC class add dev $NETCARD parent 1:1 classid 1:400 htb \
+            rate $(( $UPLINK / 2 ))kbit ceil $(( $UPLINK ))kbit burst 30k prio 4
+
+		echo "#--- ~ sub http branch: sfq"
+		$TC qdisc add dev $NETCARD parent 1:400 handle 1400: sfq perturb 10 limit 32
+
+		echo "#--- ~ http branch filter"
+		$TC filter add dev $NETCARD parent 1:0 protocol ip prio 4 handle 400 fw flowid 1:400
+
+		echo "#--- ~ netfilter rule - http/s"
+		$IPT -t mangle -A POSTROUTING -o $NETCARD -p tcp --tcp-flags SYN SYN \
+            --dport 80 -j CONNMARK --set-mark 400
+		$IPT -t mangle -A POSTROUTING -o $NETCARD -p tcp --tcp-flags SYN SYN \
+            --dport 443 -j CONNMARK --set-mark 400
+
+
+        # DEFAULT Class: the root qdisc will send all non filtered traffic to this one
+        # we also set a netfilter rule to send large file transfert to this queue
+        # and avoid impacting the HTTP queue
+		echo "#---default - id 999 - rate $(( $UPLINK / 10 ))kbit ceil $(( $UPLINK ))kbit"
+		$TC class add dev $NETCARD parent 1:1 classid 1:999 htb \
+            rate $(( $UPLINK / 10 ))kbit ceil $(( $UPLINK ))kbit burst 2k prio 9
+
+		echo "#--- ~ sub default: sfq"
+		$TC qdisc add dev $NETCARD parent 1:999 handle 1999: sfq perturb 10 limit 32
+
+		echo "#--- ~ filtre default"
+		$TC filter add dev $NETCARD parent 1:0 protocol ip prio 99 handle 999 fw flowid 1:999
+		$IPT -t mangle -A POSTROUTING -o $NETCARD -p tcp -m connbytes --connbytes 10000000: \
+            --connbytes-mode bytes --connbytes-dir both -j CONNMARK --set-mark 999
+
+
+
+		echo "#--- ~ propagating marks on connections"
+		$IPT -t mangle -A POSTROUTING -j CONNMARK --restore-mark
+
+		echo "#--- ~ Mark TCP ACKs flags at 200"
+		$IPT -t mangle -A POSTROUTING -p tcp --tcp-flags URG,ACK,PSH,RST,SYN,FIN ACK \
+            -m length --length 40:64 -j MARK --set-mark 200
+
+
+		echo
+		echo "Traffic Control is up and running"
+		echo
+
+	;;
+	stop)
+		$IPT -t mangle -F
+		$IPT -t mangle -X
+		echo "iptables rules removed"
+
+		$TC qdisc del dev $NETCARD root handle 1
+        echo "traffic control rules removed"
+
+		exit 0
+        ;;
+	restart)
+		$0 stop
+		$0 start
+		exit 0
+        ;;
+	show)
+        echo;echo " ---- qdiscs details -----"
+		$TC -d qdisc show dev $NETCARD
+        echo;echo " ---- qdiscs statistics --"
+		$TC -s qdisc show dev $NETCARD
+		exit 0
+        ;;
+	*)
+		echo;echo "usage:$0 {start|stop|restart|show}"
+        echo "load or stop the Traffic Control rules on the network interface"
+		echo;echo "julien vehent - 2011 - julien@linuxwall.info"
+	;;
+esac
+
+