├── .gitignore
├── LICENSE
├── README.md
└── lossy


/.gitignore:
--------------------------------------------------------------------------------
1 | .idea
2 | *.iml
3 | 


--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
 1 | The MIT License (MIT)
 2 | 
 3 | Copyright (c) 2016 Dmitry Murashenkov
 4 | 
 5 | Permission is hereby granted, free of charge, to any person obtaining a copy
 6 | of this software and associated documentation files (the "Software"), to deal
 7 | in the Software without restriction, including without limitation the rights
 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 9 | copies of the Software, and to permit persons to whom the Software is
10 | furnished to do so, subject to the following conditions:
11 | 
12 | The above copyright notice and this permission notice shall be included in all
13 | copies or substantial portions of the Software.
14 | 
15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 | SOFTWARE.
22 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
  1 | # lossy
  2 | Wrapper for linux tc and netem tools. Simulate lossy/shaped networks the easy way.
  3 | 
  4 | Lossy uses tc (traffic control) to build a chain of qdiscs (queues for packets) and simulate packet loss, delay 
  5 | or limited network bandwidth. It uses standard netem (network emulator) and tbf (token bucket filter) qdiscs for 
  6 | that purpose.
  7 | 
  8 | However using tc directly can be tricky because of several known issues, so this tool is made to simplify usage and 
  9 | provide built-in detection/workarounds for some problems.
 10 | 
 11 | Lossy sets rules that match incoming/outgoing traffic and applies policies to these packets: 
 12 | * To enabled packet loss/delay specify netem settings (it is better to check 'man netem' for detailed description.
 13 | * To limit network bandwidth specify tbf settings (see 'man tbf')
 14 | 
 15 | #Usage 
 16 | 
 17 | ```
 18 | Usage: sudo lossy -n 'delay 100ms 10ms 25% loss 0.1%' \
 19 |                   -t 'rate 0.5mbit burst 10kb limit 10k' \
 20 |                   --from 127.0.0.1:* \
 21 |                   --to *:80
 22 | ```
 23 | 
 24 | Options:
 25 | 
 26 | ```
 27 | -s
 28 |     Print status.
 29 | 
 30 | -c
 31 |     Clear all rules.
 32 | 
 33 | -t|--tbf <tbf-settings>
 34 |     Set network shaping setting that are passed to token bucket filter.
 35 |     Example 'rate 0.5mbit burst 10kb limit 10k'.
 36 |     See 'man tbf' for details. Default ''.
 37 | 
 38 | -n|--netem <netem-settings>
 39 |     Set delay/loss settings that are passed to netem.
 40 |     Example 'delay 100ms 10ms 25% loss 0.1%'.
 41 |     See 'man netem' for details. Default ''.
 42 | 
 43 | --from <host[/mask]:port|none>
 44 |     Match only packets from this address.
 45 |     Example '127.0.0.1:80', '*:80', '127.0.0.1/24:*', '*:*'.
 46 |     Special value 'none' - match none.
 47 |     Default 'none'.
 48 |     Host with mask can be specified.
 49 | 
 50 | --to <host[/mask]:port|none>
 51 |     Match only packets from to this address.
 52 |     Example '127.0.0.1:80', '*:80', '127.0.0.1/24:*', '*:*'.
 53 |     Special value 'none' - match none.
 54 |     Default 'none'.
 55 |     Host with mask can be specified.
 56 | 
 57 | -i|--interface <interface>
 58 |     Matches only on this interface.
 59 |     Example 'eth0', 'lo'.
 60 |     Default 'eth0'.
 61 | 
 62 | -r|--reorder-if-jitter <true|false>
 63 |     If jitter is specified for netem then should it reorder packets or send them in order.
 64 |     Default 'false' - send in order.
 65 | 
 66 | -e|--exclude-ports <port[,port...]>
 67 |     Exclude traffic to/from this port from any host from matching.
 68 |     Used to prevent setting up rule that affects ssh session you run lossy in.
 69 |     Example '22,80'.
 70 |     Several port can be specified separated by comma.
 71 |     Default '22'.
 72 | ```
 73 | 
 74 | # Important notes
 75 | 
 76 | * Must be run as root.
 77 | 
 78 | * All changes are cleared after reboot
 79 | 
 80 | * Clears existing rules on specified interface before adding new rules, so if you launched it with '-i eth0' and then '-i lo'
 81 |   then rules on eth0 will remain. When invoking lossy with -c you can also add -i to specify interface to clear.
 82 | 
 83 | * Options --from and --to each is applied both to incoming and outgoing traffic. Example: -netem 'delay 100ms' --to '*:*',
 84 |   we ping this machine and get 200ms roundtrip time because incoming packets matches *:* and reply packet also matches rule.
 85 |   If you want to affect only upstream/downstream you will have to specify ip or mask so that packets to/from this machine
 86 |   are not matched. Example: this machine has ip 192.168.1.1 and we add rule --to 192.168.1.1 this will result in 100ms
 87 |   rountrip, incoming packets will have 100ms delay while outgoing will have 0ms delay.
 88 | 
 89 | * Lossy uses ethtool to verify environment correctness, it is not required to run, but may tell you about some problems
 90 |   that can cause incorrect behavior.
 91 | 
 92 | * All settings are applied both to outgoing traffic and incoming traffic, so if you specify delay 100ms then roundtrip
 93 |   time will be 200ms. Use --from and --to to match packets more granularly.
 94 | 
 95 | * Remember that your ssh session is affected by lossy also (if you alter -e option at least), so if you specify large 
 96 |   delay, jitter, packet loss you may not be able to disable it via ssh. However all settings are reset after system reboot.
 97 | 
 98 | * If both tbf and netem enabled then netem goes after tbf.
 99 | 
100 | * No need to specify 'limit' in tbf settings if netem is enabled, because netem also has 'limit' option and internal
101 |   pfifo qdisc, so traffic will mainly be affected by this setting which we set to 1000 (packets) by default.
102 | 
103 | * If 'large segment offload' option (also known as TSO/GSO) is enabled in kernel (check with 'ethtool -k eth0') then
104 |   inside kernel packets can be larger than MTU and be broken into MTU-size chunks only by network interface. This leads
105 |   to tbf limiting speed to very low value (10-20Kb/sec) despite of settings. If this is the case recommended approach
106 |   is to disable TSO/GSO, this doesn't affect functionality, but may slightly increase CPU usage in case of large amounts
107 |   of traffic.
108 | 
109 | * Netem has a reordering bug/feature - if jitter is specified like 'delay 100ms 50ms' then packets will be reordered. Docs
110 |   suggest adding child pfifo qdisc, but it doesn't work in 3.2-4.2 kernels, but there is undocumented behavior that if
111 |   'rate' netem option is specified then no reordering occurs. Option was added in attempt to bring tbf functionality into
112 |   netem but has this side effect. So we simply have -r flag to prevent reordering, it results in adding large 'rate' value.
113 |   Seems several rates can be specified and the last one takes effect so we prepend our value to allow overriding.
114 | 
115 | * When configuring tbf set burst and limit high enough. Min burst is rate/250 (HZ=250 in many linux kernels) so 40k is
116 |   required to reach 10Mb/sec speed. Sometimes packets can even fail to queue at all if limit is low (-s shows drops
117 |   increasing, but sent bytes stays the same) so always verify your configuration with iperf and ping tools.
118 | 


--------------------------------------------------------------------------------
/lossy:
--------------------------------------------------------------------------------
  1 | #!/bin/bash +x
  2 | 
  3 | #netem has inner queue if packets (implemented as a pfifo qdisc) so if both netem and tbf are enabled
  4 | #and netem goes after tbf then total traffic will be limited by netem's "limit" option and tbf's "limit"
  5 | #won't have any effect
  6 | #netem "limit" option has same meaning as tbf "limit" option so in typical case of mss=1.5Kb
  7 | #10 pkts will mean 15Kb
  8 | TBF_BUFFER_SIZE_PKTS=1000
  9 | 
 10 | INTERFACE="eth0"
 11 | FROM='none'
 12 | TO='none'
 13 | TBF_SETTINGS=''
 14 | NETEM_SETTINGS=''
 15 | NETEM_REORDER_IF_JITTER='false'
 16 | ACTION="enable"
 17 | EXCLUDE_PORTS="22"
 18 | 
 19 | usage()
 20 | {
 21 | echo "
 22 | Usage: sudo lossy -n 'delay 100ms 10ms 25% loss 0.1%' \
 23 |                   -t 'rate 0.5mbit burst 10kb limit 10k' \
 24 |                   --from 127.0.0.1:* \
 25 |                   --to *:80
 26 | 
 27 | Wrapper for linux tc and netem tools. Simulate lossy/shaped networks the easy way.
 28 | 
 29 | Lossy uses tc (traffic control) to build a chain of qdiscs (queues for packets) and simulate packet loss, delay
 30 | or limited network bandwidth. It uses standard netem (network emulator) and tbf (token bucket filter) qdiscs for
 31 | that purpose.
 32 | 
 33 | However using tc directly can be tricky because of several known issues, so this tool is made to simplify usage and
 34 | provide built-in detection/workarounds for some problems.
 35 | 
 36 | Lossy sets rules that match incoming/outgoing traffic and applies policies to these packets:
 37 | * To enabled packet loss/delay specify netem settings (it is better to check 'man netem' for detailed description.
 38 | * To limit network bandwidth specify tbf settings (see 'man tbf')
 39 | 
 40 | Options:
 41 | 
 42 | -s
 43 |     Print status.
 44 | 
 45 | -c
 46 |     Clear all rules.
 47 | 
 48 | -t|--tbf <tbf-settings>
 49 |     Set network shaping setting that are passed to token bucket filter.
 50 |     Example 'rate 0.5mbit burst 10kb limit 10k'.
 51 |     See 'man tbf' for details. Default ''.
 52 | 
 53 | -n|--netem <netem-settings>
 54 |     Set delay/loss settings that are passed to netem.
 55 |     Example 'delay 100ms 10ms 25% loss 0.1%'.
 56 |     See 'man netem' for details. Default ''.
 57 | 
 58 | --from <host[/mask]:port|none>
 59 |     Match only packets from this address.
 60 |     Example '127.0.0.1:80', '*:80', '127.0.0.1/24:*', '*:*'.
 61 |     Special value 'none' - match none.
 62 |     Default 'none'.
 63 |     Host with mask can be specified.
 64 | 
 65 | --to <host[/mask]:port|none>
 66 |     Match only packets from to this address.
 67 |     Example '127.0.0.1:80', '*:80', '127.0.0.1/24:*', '*:*'.
 68 |     Special value 'none' - match none.
 69 |     Default 'none'.
 70 |     Host with mask can be specified.
 71 | 
 72 | -i|--interface <interface>
 73 |     Matches only on this interface.
 74 |     Example 'eth0', 'lo'.
 75 |     Default 'eth0'.
 76 | 
 77 | -r|--reorder-if-jitter <true|false>
 78 |     If jitter is specified for netem then should it reorder packets or send them in order.
 79 |     Default 'false' - send in order.
 80 | 
 81 | -e|--exclude-ports <port[,port...]>
 82 |     Exclude traffic to/from this port from any host from matching.
 83 |     Used to prevent setting up rule that affects ssh session you run lossy in.
 84 |     Example '22,80'.
 85 |     Several port can be specified separated by comma.
 86 |     Default '22'.
 87 | 
 88 | Important notes:
 89 | 
 90 | * Must be run as root.
 91 | 
 92 | * All changes are cleared after reboot
 93 | 
 94 | * Clears existing rules on specified interface before adding new rules, so if you launched it with '-i eth0' and then '-i lo'
 95 |   then rules on eth0 will remain. When invoking lossy with -c you can also add -i to specify interface to clear.
 96 | 
 97 | * Options --from and --to each is applied both to incoming and outgoing traffic. Example: -netem 'delay 100ms' --to '*:*',
 98 |   we ping this machine and get 200ms roundtrip time because incoming packets matches *:* and reply packet also matches rule.
 99 |   If you want to affect only upstream/downstream you will have to specify ip or mask so that packets to/from this machine
100 |   are not matched. Example: this machine has ip 192.168.1.1 and we add rule --to 192.168.1.1 this will result in 100ms
101 |   rountrip, incoming packets will have 100ms delay while outgoing will have 0ms delay.
102 | 
103 | * Lossy uses ethtool to verify environment correctness, it is not required to run, but may tell you about some problems
104 |   that can cause incorrect behavior.
105 | 
106 | * All settings are applied both to outgoing traffic and incoming traffic, so if you specify delay 100ms then roundtrip
107 |   time will be 200ms. Use --from and --to to match packets more granularly.
108 | 
109 | * Remember that your ssh session is affected by lossy also (if you alter -e option at least), so if you specify large
110 |   delay, jitter, packet loss you may not be able to disable it via ssh. However all settings are reset after system reboot.
111 | 
112 | * If both tbf and netem enabled then netem goes after tbf.
113 | 
114 | * No need to specify 'limit' in tbf settings if netem is enabled, because netem also has 'limit' option and internal
115 |   pfifo qdisc, so traffic will mainly be affected by this setting which we set to 1000 (packets) by default.
116 | 
117 | * If 'large segment offload' option (also known as TSO/GSO) is enabled in kernel (check with 'ethtool -k eth0') then
118 |   inside kernel packets can be larger than MTU and be broken into MTU-size chunks only by network interface. This leads
119 |   to tbf limiting speed to very low value (10-20Kb/sec) despite of settings. If this is the case recommended approach
120 |   is to disable TSO/GSO, this doesn't affect functionality, but may slightly increase CPU usage in case of large amounts
121 |   of traffic.
122 | 
123 | * Netem has a reordering bug/feature - if jitter is specified like 'delay 100ms 50ms' then packets will be reordered. Docs
124 |   suggest adding child pfifo qdisc, but it doesn't work in 3.2-4.2 kernels, but there is undocumented behavior that if
125 |   'rate' netem option is specified then no reordering occurs. Option was added in attempt to bring tbf functionality into
126 |   netem but has this side effect. So we simply have -r flag to prevent reordering, it results in adding large 'rate' value.
127 |   Seems several rates can be specified and the last one takes effect so we prepend our value to allow overriding.
128 | 
129 | * When configuring tbf set burst and limit high enough. Min burst is rate/250 (HZ=250 in many linux kernels) so 40k is
130 |   required to reach 10Mb/sec speed. Sometimes packets can even fail to queue at all if limit is low (-s shows drops
131 |   increasing, but sent bytes stays the same) so always verify your configuration with iperf and ping tools.
132 | "
133 | }
134 | 
135 | status()
136 | {
137 |     echo "Outgoing traffic settings for interface $INTERFACE"
138 |     tc qdisc show dev $INTERFACE
139 |     tc -s filter show dev $INTERFACE
140 |     echo
141 |     echo "Incoming traffic settings for interface $INTERFACE"
142 |     tc qdisc show dev ifb0
143 |     tc -s filter show dev ifb0
144 |     echo
145 |     echo "Statistics for interface $INTERFACE"
146 |     tc -s qdisc ls
147 | }
148 | 
149 | clear()
150 | {
151 |     echo "Clearing rules for interface $INTERFACE"
152 |     tc qdisc del dev $INTERFACE root
153 |     tc qdisc del dev $INTERFACE ingress
154 |     tc qdisc del dev ifb0 root
155 | }
156 | 
157 | checkTSO()
158 | {
159 |     if ! hash ethtool 2>/dev/null; then
160 |         echo "Recommend installing ethtool so we can verify environment correctness and report possible problems (not required)"
161 |         return
162 |     fi
163 | 
164 |     if [ "x" != "x$TBF_SETTINGS" ]; then
165 |         #If "Large segment offload" (also called TSO/GSO) kernel option is enabled
166 |         #then kernel may operate large packets (like MTU is 16K for example) internally
167 |         #and only when packet reaches network interface it is broken into network-MTU-size
168 |         #chunks.
169 |         #In this options is on and TBF enabled then it will limit speed to very low value (10-20Kb/sec)
170 |         #despite of applied settings. Disabling TSO/GSO doesn't affect functionality but
171 |         #may cause some CPU overhead (not a problem unless you have really much traffic).
172 |         if ethtool -k $INTERFACE | egrep 'tcp-segmentation-offload: on|generic-segmentation-offload: on' > /dev/null ; then
173 |             echo "TSO/GSO must be disabled for network interface with following command (otherwise TBF works incorrectly): ethtool -K $INTERFACE tso off gso off"
174 |             exit 1
175 |         fi
176 |     fi
177 | }
178 | 
179 | checkRoot()
180 | {
181 |     if [[ $EUID -ne 0 ]]; then
182 |        echo "This script must be run as root"
183 |        exit 1
184 |     fi
185 | }
186 | 
187 | buildFromFilter()
188 | {
189 |     FILTER=""
190 |     if [ "none" == "$FROM" ]; then
191 |         return
192 |     elif [ "x" != "x$FROM" ]; then
193 |         IP=`getIp "$FROM"`
194 |         PORT=`getPort "$FROM"`
195 |         if [ "x" != "x$IP" ]; then
196 |             FILTER="$FILTER match ip src $IP"
197 |         else
198 |             #if ip not specified then we need to match all packets
199 |             FILTER="$FILTER match ip src 0/0"
200 |         fi
201 | 
202 |         if [ "x" != "x$PORT" ]; then
203 |             FILTER="$FILTER match ip sport $PORT 0xffff"
204 |         fi
205 |     fi
206 | 
207 |     echo $FILTER
208 | }
209 | 
210 | buildToFilter()
211 | {
212 |     FILTER=""
213 |     if [ "none" == "$TO" ]; then
214 |         return
215 |     elif [ "x" != "x$TO" ]; then
216 |         IP=`getIp "$TO"`
217 |         PORT=`getPort "$TO"`
218 |         if [ "x" != "x$IP" ]; then
219 |             FILTER="$FILTER match ip dst $IP"
220 |         else
221 |             #if ip not specified then we need to match all packets
222 |             FILTER="$FILTER match ip dst 0/0"
223 |         fi
224 | 
225 |         if [ "x" != "x$PORT" ]; then
226 |             FILTER="$FILTER match ip dport $PORT 0xffff"
227 |         fi
228 |     fi
229 | 
230 |     echo $FILTER
231 | }
232 | 
233 | 
234 | addExcludePortFilter()
235 | {
236 |     if [ "x" != "x$EXCLUDE_PORTS" ]; then
237 |         IFS=',' read -r -a PORTS <<< "$EXCLUDE_PORTS"
238 |         for PORT in "${PORTS[@]}"
239 |         do
240 |             eval "tc filter add dev $INTERFACE protocol ip parent 1:0 prio 1 u32 match ip src 0/0 match ip sport $PORT 0xffff flowid 1:1"
241 |             eval "tc filter add dev $INTERFACE protocol ip parent 1:0 prio 1 u32 match ip src 0/0 match ip dport $PORT 0xffff flowid 1:1"
242 |             eval "tc filter add dev ifb0 protocol ip parent 2:0 prio 1 u32 match ip src 0/0 match ip dport $PORT 0xffff flowid 2:1"
243 |             eval "tc filter add dev ifb0 protocol ip parent 2:0 prio 1 u32 match ip src 0/0 match ip dport $PORT 0xffff flowid 2:1"
244 |         done
245 |     fi
246 | }
247 | 
248 | enableEgress()
249 | {
250 |     #first add prio queue to split traffic - with it we can apply limits to single band
251 |     tc qdisc add dev $INTERFACE root handle 1: prio
252 |     PARENT_QDISC_HANDLE="1:3"
253 | 
254 |     if [ "x" != "x$TBF_SETTINGS" ]; then
255 |         eval "tc qdisc add dev $INTERFACE parent $PARENT_QDISC_HANDLE handle 10: tbf $TBF_SETTINGS"
256 |         PARENT_QDISC_HANDLE="10:0"
257 |     fi
258 | 
259 |     if [ "x" != "x$NETEM_SETTINGS" ]; then
260 |         eval "tc qdisc add dev $INTERFACE parent $PARENT_QDISC_HANDLE handle 11: netem $NETEM_SETTINGS limit $TBF_BUFFER_SIZE_PKTS"
261 |     fi;
262 | 
263 |     FROM_FILTER=`buildFromFilter`
264 |     TO_FILTER=`buildToFilter`
265 |     if [ "x" != "x$FROM_FILTER" ]; then
266 |         eval "tc filter add dev $INTERFACE protocol ip parent 1:0 prio 3 u32 $FROM_FILTER flowid 1:3"
267 |     fi;
268 |     if [ "x" != "x$TO_FILTER" ]; then
269 |         eval "tc filter add dev $INTERFACE protocol ip parent 1:0 prio 3 u32 $TO_FILTER flowid 1:3"
270 |     fi;
271 | }
272 | 
273 | enableIngress()
274 | {
275 |     #tc exists to control how we send (egress) traffic and it is expected that we want
276 |     #to process received (ingress) traffic without introducing any queues however for test
277 |     #purposes we may want to shape incoming traffic so we have to redirect it via virtual interface
278 |     #and handle it like it is egress traffic
279 |     modprobe ifb
280 |     ip link set dev ifb0 up
281 |     tc qdisc add dev $INTERFACE ingress
282 |     tc filter add dev $INTERFACE parent ffff: protocol ip u32 match u32 0 0 flowid 1:1 action mirred egress redirect dev ifb0
283 | 
284 |     tc qdisc add dev ifb0 root handle 2: prio
285 |     PARENT_QDISC_HANDLE="2:3"
286 | 
287 |      if [ "x" != "x$TBF_SETTINGS" ]; then
288 |         eval "tc qdisc add dev ifb0 parent $PARENT_QDISC_HANDLE handle 20: tbf $TBF_SETTINGS"
289 |         PARENT_QDISC_HANDLE="20:0"
290 |     fi
291 | 
292 |     if [ "x" != "x$NETEM_SETTINGS" ]; then
293 |         eval "tc qdisc add dev ifb0 parent $PARENT_QDISC_HANDLE handle 21: netem $NETEM_SETTINGS limit $TBF_BUFFER_SIZE_PKTS"
294 |     fi;
295 | 
296 |     FROM_FILTER=`buildFromFilter`
297 |     TO_FILTER=`buildToFilter`
298 |     if [ "x" != "x$FROM_FILTER" ]; then
299 |         eval "tc filter add dev ifb0 protocol ip parent 2:0 prio 3 u32 $FROM_FILTER flowid 1:3"
300 |     fi;
301 |     if [ "x" != "x$TO_FILTER" ]; then
302 |         eval "tc filter add dev ifb0 protocol ip parent 2:0 prio 3 u32 $TO_FILTER flowid 1:3"
303 |     fi;
304 | }
305 | 
306 | enable()
307 | {
308 |     checkTSO
309 |     enableEgress
310 |     enableIngress
311 |     addExcludePortFilter
312 |     echo 'New settings applied'
313 | }
314 | 
315 | getIp()
316 | {
317 |     OIFS="$IFS"
318 |     IFS=':'
319 |     read -a IP_PORT <<< "$1"
320 |     IFS="$OIFS"
321 |     #if wildcard is given output empty string to match everything
322 |     echo "${IP_PORT[0]}" | tr -d '*'
323 | }
324 | 
325 | getPort()
326 | {
327 |     OIFS="$IFS"
328 |     IFS=':'
329 |     read -a IP_PORT <<< "$1"
330 |     IFS="$OIFS"
331 |     shift
332 |     #if wildcard is given output empty string to match everything
333 |     echo "${IP_PORT[1]}" | tr -d '*'
334 | }
335 | 
336 | printArgs()
337 | {
338 |     echo "Interface: $INTERFACE"
339 |     echo "From: ${FROM:-*:*}"
340 |     echo "To: ${TO:-*:*}"
341 |     echo "Exclude ports: $EXCLUDE_PORTS"
342 |     echo "TBF settings: $TBF_SETTINGS"
343 |     echo "Netem settings: $NETEM_SETTINGS"
344 |     echo "Reorder packets during jitter: $NETEM_REORDER_IF_JITTER"
345 | }
346 | 
347 | if [ "$#" == 0 ]; then
348 |     usage
349 |     exit 0
350 | fi
351 | 
352 | while [[ $# -gt 0 ]]
353 | do
354 |     key="$1"
355 | 
356 |     case $key in
357 | 
358 |         -h|--help|'-?')
359 |             usage
360 |             exit 0
361 |         ;;
362 | 
363 |         -s|--status)
364 |             ACTION="status"
365 |         ;;
366 | 
367 |         -t|--tbf)
368 |             shift
369 |             TBF_SETTINGS=$1
370 |         ;;
371 | 
372 |         -n|--netem)
373 |             shift
374 |             NETEM_SETTINGS=$1
375 |         ;;
376 | 
377 |        --from)
378 |             shift
379 |             FROM=$1
380 |         ;;
381 | 
382 |        --to)
383 |             shift
384 |             TO=$1
385 |         ;;
386 | 
387 |         -i|--interface)
388 |             shift
389 |             INTERFACE=$1
390 |         ;;
391 | 
392 |         -r|--reorder-if-jitter)
393 |             shift
394 |             NETEM_REORDER_IF_JITTER=$1
395 |         ;;
396 | 
397 |         -c|--clear)
398 |             ACTION="clear"
399 |         ;;
400 | 
401 |         -e|--exclude-port)
402 |             shift
403 |             EXCLUDE_PORTS=$1
404 |         ;;
405 | 
406 |         *)
407 |             echo "Unknown option $key"
408 |             exit 1
409 |         ;;
410 | 
411 |     esac
412 |     shift
413 | done
414 | 
415 | if [ "clear" == "$ACTION" ]; then
416 |     checkRoot
417 |     clear
418 |     exit 0
419 | elif [ "status" == "$ACTION" ]; then
420 |     status
421 |     exit 0
422 | elif  [ "enable" == "$ACTION" ]; then
423 |     if [ "false" == "$NETEM_REORDER_IF_JITTER" ] && [ "x" != "x$NETEM_SETTINGS" ]; then
424 |         #latest version of netem only prevents reorder if rate is added
425 |         NETEM_SETTINGS="rate 10Gbit $NETEM_SETTINGS"
426 |     fi
427 | 
428 |     checkRoot
429 |     printArgs
430 |     clear
431 |     enable
432 | 
433 |     if [ "false" == "$NETEM_REORDER_IF_JITTER" ] && [ "x" != "x$NETEM_SETTINGS" ]; then
434 |         #check that 'rate' param is supported and was set
435 |         if ! tc -s qdisc ls | grep netem | grep rate &>/dev/null; then
436 |             echo "This netem version always reorders packets based on jitter, -r option omitted"
437 |         fi
438 |     fi
439 | fi
440 | 


--------------------------------------------------------------------------------