I'm troubleshooting a modified version of the wondershaper and am
encountering some trouble that I was hoping you could help me look into.
I have 4 queues that I add traffic to:
1. VoIP phone stuff
2. ssh traffic
3. HTTP
4. everything else (should be default).
Latency is definitely increased when I run it. I was trying the catch
all rule (#4) and removed the specific rule for HTTP. Presumably this
should mean that http traffic is caught by rule 4.
I'm not sure what happens to http traffic, but I do know that I have a
horrible latency after removing a specific rule to catch http.
Could you possibly look over my rules and give me some feedback.
I'm also confused by your Stochastic Fairness rules. I don't understand
why you need them if we are using HTB as our qdisc.
S.
--
_ pictures ________________________________ http://www.imaginator.com
_ contact _________________________ http://imaginator.com/contact.php
#!/bin/sh
# Based on the Wonder Shaper, with modifications by simon@imaginator.com
# this script sets up 4 HTB queues and assigns traffic of diffrent priority to
# each HTB. The script also shapes incoming traffic to stop buffering at our
# ISP which kills latency.
# I have a 1.5Mbps downlink and a 384Kbps uplink ADSL connection.
# The central office is close enought that I get the full speed.
# I set the link up for a slightly lower speed each way.
# This is the inboud and outbound speed in Kbps
DOWNLINK=1300
UPLINK=300
DEV=eth0
if [ "$1" = "start" ]
then
echo "installing rules";
# uplink first
# install root HTB, point default traffic to 1:30:
tc qdisc add dev $DEV root handle 1: htb default 40
# shape everything at $UPLINK speed - this prevents huge queues in your
# DSL modem which destroy latency:
tc class add dev $DEV parent 1: classid 1:1 htb rate ${UPLINK}kbit burst 6k
# setup limit queues
# I believe that this sets up 4 buckes with bucket one having priority
# over bucket 4. Bucket one is emptied before anything in bucket 2 is
# touched.
tc class add dev $DEV parent 1:1 classid 1:10 htb rate $[10*$UPLINK/10]kbit \
burst 6k prio 1
tc class add dev $DEV parent 1:1 classid 1:20 htb rate $[10*$UPLINK/10]kbit \
burst 6k prio 2
tc class add dev $DEV parent 1:1 classid 1:30 htb rate $[10*$UPLINK/10]kbit \
burst 6k prio 3
tc class add dev $DEV parent 1:1 classid 1:40 htb rate $[10*$UPLINK/10]kbit \
burst 6k prio 4
# the original wonder shaper had these lines in and they seemed superfluous
# because we are using HTB:
# all get Stochastic Fairness:
#tc qdisc add dev $DEV parent 1:10 handle 10: sfq perturb 10
#tc qdisc add dev $DEV parent 1:20 handle 20: sfq perturb 10
#tc qdisc add dev $DEV parent 1:30 handle 30: sfq perturb 10
# We now have 4 queues that we can start assigning traffic to
# I think that this works like iptables rules ie scanning through
# the rules until it finds a match and then dumping the packet into
# the matched queue
# This is the ping latency test to see how things are working...
# Smokeping pings hosts every 5 minutes and records the round trip time.
# Here packets going to 1 network get top priority and the other network,
# a low priority. The results are at:
# http://imaginator.com/cgi-bin/smokeping.cgi?target=Latecy
tc filter add dev $DEV parent 1:0 protocol ip prio 10 u32 \
match ip dst 64.3.149.1 flowid 1:10
tc filter add dev $DEV parent 1:0 protocol ip prio 10 u32 \
match ip dst 64.3.151.1 flowid 1:40
# VOIP stuff gets top priority
tc filter add dev $DEV parent 1:0 protocol ip prio 10 u32 \
match ip dport 10000 0xffff flowid 1:10
# ssh connections are second priority for outgoing
tc filter add dev $DEV parent 1:0 protocol ip prio 10 u32 \
match ip dport 22 0xffff flowid 1:20
# outbound webpages and my streaming mp3s
tc filter add dev $DEV parent 1:0 protocol ip prio 10 u32 \
match ip sport 80 0xffff flowid 1:30
tc filter add dev $DEV parent 1:0 protocol ip prio 10 u32 \
match ip sport 7000 0xffff flowid 1:30
# rest is 'non-interactive' ie 'bulk' and ends up in 1:40
tc filter add dev $DEV parent 1:0 protocol ip prio 10 u32 \
match ip dst 0.0.0.0/0 flowid 1:40
# returning traffic is slowed slightly so return packets don't get
# buffered at the ISP and kill our latency.
# attach ingress filter:
tc qdisc add dev $DEV handle ffff: ingress
# filter *everything* to it (0.0.0.0/0), drop everything that's
# coming in too fast:
tc filter add dev $DEV parent ffff: protocol ip prio 50 u32 match ip src \
0.0.0.0/0 police rate ${DOWNLINK}kbit burst 10k drop flowid :1
# that's it, we're shaping.
fi
if [ "$1" = "status" ]
then
echo "showing status";
tc -s qdisc ls dev $DEV
tc -s class ls dev $DEV
exit
fi
if [ "$1" = "restart" ]
then
$0 stop
$0 start
fi
if [ "$1" = "stop" ]
then
echo "shutting down";
# clean existing down- and uplink qdiscs, hide errors
tc qdisc del dev $DEV root 2> /dev/null > /dev/null
tc qdisc del dev $DEV ingress 2> /dev/null > /dev/null
fi
