i anew to netlink. its been somtime since i have been trying to figure out what rtnetlink_dump_ifinfo does. could someone please help me on this. thanks. --------------------------------- Do you Yahoo!? Yahoo! Mail - Find whayou need with new enhanced search. Learn more. -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20050301/fd5ba25c/attachment.htm Fronistnet_user ayahoo.com Tue Mar 1 14:21:29 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: wherdoes tc requests rtnetlink_dump_ifinfo Message-ID: <20050301222129.39635.qmail@xxxxxxxxxxxxxxxxxxxxxxx> hi whei executtc qdisc add dev eth0 root netem rtnetlink_dump_ifinfo is called which traverses the lisof network devices and calls rtnetlink_fill_ifinfo i anoable to figure out what why is the function rtnetlink_dump_ifinfo called and wheriis requested. fromy understanding its is simply dumping thinformation for all thinterfaces. i would appreciatif someoncould could point me to the call. thanks. __________________________________ Do you Yahoo!? Read only thmail you wan- Yahoo! Mail SpamGuard. http://promotions.yahoo.com/new_mail Froerwann.thoraval airisa.fr Wed Mar 2 06:33:46 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 12:51:15 2007 Subject: Using th'loss' option Message-ID: <4225CECA.30802@xxxxxxxx> Hello, Iseems thathe 'loss' option is not included in the 'tc' command. Heris thexample provided by the Netem website: tc qdisc changdev eth0 roonetem loss .1% And heris thresult of this command: RTNETLINK answers: No such filor directory Indeed, running 'tc qdisc changdev eth0 roonetem help' gives: Usage: ... nete[ limiPACKETS ] [ delay TIME [ JITTER [CORRELATION]]] [ drop PERCENT [CORRELATION]] [ duplicatPERCENT [CORRELATION]] [ distributio{uniform|normal|pareto|paretonormal} ] [ gap PACKETS ] No 'loss' optioprovided. I arunning a 2.6.10-gentoo-r7 with iproute2_2.6.10.20050124. Any idea ? Regards, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 Frostephen.earl abt.com Wed Mar 2 06:37:08 2005 From: stephen.earl abt.co(stephen.earl@xxxxxx) Date: Wed Apr 18 12:51:15 2007 Subject: Using th'loss' option Message-ID: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> You need to typ'add' instead of'change' to add thrule. 'change' only works if you already have a rule running on that interface. -----Original Message----- From: netem-bounces@xxxxxxxxxxxxxx [mailto:netem-bounces@xxxxxxxxxxxxxx]OBehalf Of Erwann Thoraval Sent: 02 March 2005 14:34 To: netem@xxxxxxxxxxxxxx Subject: Using th'loss' option Hello, Iseems thathe 'loss' option is not included in the 'tc' command. Heris thexample provided by the Netem website: tc qdisc changdev eth0 roonetem loss .1% And heris thresult of this command: RTNETLINK answers: No such filor directory Indeed, running 'tc qdisc changdev eth0 roonetem help' gives: Usage: ... nete[ limiPACKETS ] [ delay TIME [ JITTER [CORRELATION]]] [ drop PERCENT [CORRELATION]] [ duplicatPERCENT [CORRELATION]] [ distributio{uniform|normal|pareto|paretonormal} ] [ gap PACKETS ] No 'loss' optioprovided. I arunning a 2.6.10-gentoo-r7 with iproute2_2.6.10.20050124. Any idea ? Regards, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 _______________________________________________ Netemailing list Netem@xxxxxxxxxxxxxx http://lists.osdl.org/mailman/listinfo/netem Froerwann.thoraval airisa.fr Wed Mar 2 06:54:02 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 12:51:15 2007 Subject: Using th'loss' option In-Reply-To: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> References: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <4225D38A.9070206@xxxxxxxx> stephen.earl@bt.cowrote: > You need to typ'add' instead of'change' to add thrule. 'change' only works if you already have a rule running on that interface. Indeed, iworks with 'add', thanks ! Nevertheless, th'loss' option is nogiven by the 'help': >Usage: ... nete[ limiPACKETS ] > [ delay TIME [ JITTER [CORRELATION]]] > [ drop PERCENT [CORRELATION]] > [ duplicatPERCENT [CORRELATION]] > [ distributiouniform|normal|pareto|paretonormal}] > [ gap PACKETS ] Maybthis will bfixed soon. Regards, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 Frojw afemmecomp.com Wed Mar 2 07:05:28 2005 From: jw afemmecomp.co(John Wiley) Date: Wed Apr 18 12:51:15 2007 Subject: Using th'loss' option In-Reply-To: <4225D38A.9070206@xxxxxxxx> References: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <4225D38A.9070206@xxxxxxxx> Message-ID: <60948.69.143.124.86.1109775928.squirrel@xxxxxxxxxxxxxxxxx> Erwann, My understanding is thath"loss" option and the "drop" option are synonymous (packeloss and packe"drop" are essentially the same phenomenofronetem's standpoint)... BesRegards! -jw > stephen.earl@bt.cowrote: >> You need to typ'add' instead of'change' to add thrule. 'change' >> only works if you already hava rulrunning on that interface. > > Indeed, iworks with 'add', thanks ! > > Nevertheless, th'loss' option is nogiven by the 'help': > > >Usage: ... nete[ limiPACKETS ] > > [ delay TIME [ JITTER [CORRELATION]]] > > [ drop PERCENT [CORRELATION]] > > [ duplicatPERCENT [CORRELATION]] > > [ distribution > uniform|normal|pareto|paretonormal}] > > [ gap PACKETS ] > > Maybthis will bfixed soon. > > Regards, > -- > ErwanThoraval > IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France > T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 > > Tel. direct: +33 (0) 2 99 84 71 39 > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > http://lists.osdl.org/mailman/listinfo/netem Frojw afemmecomp.com Wed Mar 2 07:07:43 2005 From: jw afemmecomp.co(John Wiley) Date: Wed Apr 18 12:51:15 2007 Subject: Patch Archive... In-Reply-To: <4225D38A.9070206@xxxxxxxx> References: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <4225D38A.9070206@xxxxxxxx> Message-ID: <60957.69.143.124.86.1109776063.squirrel@xxxxxxxxxxxxxxxxx> All, Ilooking for an archivof the patches that have been floating around herfor thlast weeks... (even a message archive would be helpful!) Is thersuch a thing available? Thmailing list link on the netem web pagaOSDL doesn't seem to work... BesRegards! -jw Fronistnet_user ayahoo.com Wed Mar 2 17:12:00 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: dynamically increasing bands iprio Message-ID: <20050303011201.80951.qmail@xxxxxxxxxxxxxxxxxxxxxxx> is ipossiblto increase bands in prio dynamically. by defaulihas 3 bands, after attaching prio to an interface, cai increasthe number of bands to lets say 10? or is ionly possiblwhen you attach prio to ainterface? thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Thu Mar 3 09:52:57 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: dynamically increasing bands iprio In-Reply-To: <20050303011201.80951.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050303011201.80951.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050303095257.72990772@xxxxxxxxxxxxxxxxx> OWed, 2 Mar 2005 17:12:00 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > is ipossiblto increase bands in prio dynamically. > by defaulihas 3 bands, after attaching prio to an > interface, cai increasthe number of bands to lets > say 10? or is ionly possiblwhen you attach prio to > ainterface? > Thbesway to answer questions like this is to read the source. Thnumber of bands in thprio scheduler is controlled by the constanTCQ_PRIO_BANDS. Iis defined in include/linux/pkt_sched.h to b16. Fronistnet_user ayahoo.com Thu Mar 3 14:35:13 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: htb and netem- unexpected results Message-ID: <20050303223514.94354.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i experimented attaching neteto htb by doing the following tc qdisc add dev eth0 roohandl1: htb default 11 tc class add dev eth0 paren1:0 handl1:1 htb rate 1000mbiceil 1000mbit tc class add dev eth0 paren1:1 handl1:10 htb rate 500mbiceil 500mbit tc class add dev eth0 paren1:1 handl1:11 htb rate 500mbiceil 500mbit tc filter add dev eth0 protocol ip paren1: prio 1 u32 match ip ds11.0.0.1 flowid 1:10 __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Thu Mar 3 14:38:23 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: htb and netem- unexpected results In-Reply-To: <20050303223514.94354.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050303223514.94354.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050303143823.7c942022@xxxxxxxxxxxxxxxxx> OThu, 3 Mar 2005 14:35:13 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > i experimented attaching neteto htb by doing the > following > > > > tc qdisc add dev eth0 roohandl1: htb default 11 > > tc class add dev eth0 paren1:0 handl1:1 htb rate > 1000mbiceil 1000mbit > > tc class add dev eth0 paren1:1 handl1:10 htb rate > 500mbiceil 500mbit > > tc class add dev eth0 paren1:1 handl1:11 htb rate > 500mbiceil 500mbit > > tc filter add dev eth0 protocol ip paren1: prio 1 > u32 match ip ds11.0.0.1 flowid 1:10 > whahappened? Fronistnet_user ayahoo.com Thu Mar 3 14:39:52 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: netewith htb - mordelay than expected Message-ID: <20050303223953.50646.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i experimented attaching neteto htb by doing the following eth0 has ip 11.0.0.2 tc qdisc add dev eth0 roohandl1: htb default 11 tc class add dev eth0 paren1:0 handl1:1 htb rate 1000mbiceil 1000mbit tc class add dev eth0 paren1:1 handl1:10 htb rate 500mbiceil 500mbit tc class add dev eth0 paren1:1 handl1:11 htb rate 500mbiceil 500mbit tc filter add dev eth0 protocol ip paren1: prio 1 u32 match ip ds11.0.0.1 flowid 1:10 uptill now if i ping 11.0.0.1 i gea responstime of 0.1 ms nexi attached neteby tc qdisc add dev eth0 paren1:10 handl20: netem latency 10ms now id i ping 11.0.0.1 i gea responstime of 1000ms i anosure what i did wrong. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Thu Mar 3 18:00:17 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: Re: netewith htb - mordelay than expected Message-ID: <20050304020017.54118.qmail@xxxxxxxxxxxxxxxxxxxxxxx> whai havnoticed is that the delay gets round off to thneares1000th of ms. for 1ms i get 1000ms, for 1001ms i ge2000ms... i anot sure what i am doing wrong --- js si <nistnet_user@xxxxxxxxx> wrote: > i experimented attaching neteto htb by doing the > following > > eth0 has ip 11.0.0.2 > > tc qdisc add dev eth0 roohandl1: htb default 11 > > tc class add dev eth0 paren1:0 handl1:1 htb rate > 1000mbiceil 1000mbit > > tc class add dev eth0 paren1:1 handl1:10 htb > rate > 500mbiceil 500mbit > > tc class add dev eth0 paren1:1 handl1:11 htb > rate > 500mbiceil 500mbit > > tc filter add dev eth0 protocol ip paren1: prio 1 > u32 match ip ds11.0.0.1 flowid 1:10 > > uptill now if i ping 11.0.0.1 i gea responstime > of > 0.1 ms > > nexi attached neteby > > tc qdisc add dev eth0 paren1:10 handl20: netem > latency 10ms > > now id i ping 11.0.0.1 i gea responstime of > 1000ms > > i anosure what i did wrong. > > > > > __________________________________ > CelebratYahoo!'s 10th Birthday! > Yahoo! Netrospective: 100 Moments of thWeb > http://birthday.yahoo.com/netrospective/ > __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Thu Mar 3 19:23:15 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: Re: netewith htb - mordelay than expected Message-ID: <20050304032315.48348.qmail@xxxxxxxxxxxxxxxxxxxxxxx> another thing thai tried was to pua prio in betweehtb and netem. then iworks fine. --- js si <nistnet_user@xxxxxxxxx> wrote: > whai havnoticed is that the delay gets round off > to thneares1000th of ms. for 1ms i get 1000ms, > for > 1001ms i ge2000ms... i anot sure what i am doing > wrong > > > --- js si <nistnet_user@xxxxxxxxx> wrote: > > > i experimented attaching neteto htb by doing the > > following > > > > eth0 has ip 11.0.0.2 > > > > tc qdisc add dev eth0 roohandl1: htb default > 11 > > > > tc class add dev eth0 paren1:0 handl1:1 htb > rate > > 1000mbiceil 1000mbit > > > > tc class add dev eth0 paren1:1 handl1:10 htb > > rate > > 500mbiceil 500mbit > > > > tc class add dev eth0 paren1:1 handl1:11 htb > > rate > > 500mbiceil 500mbit > > > > tc filter add dev eth0 protocol ip paren1: prio > 1 > > u32 match ip ds11.0.0.1 flowid 1:10 > > > > uptill now if i ping 11.0.0.1 i gea response > time > > of > > 0.1 ms > > > > nexi attached neteby > > > > tc qdisc add dev eth0 paren1:10 handl20: netem > > latency 10ms > > > > now id i ping 11.0.0.1 i gea responstime of > > 1000ms > > > > i anosure what i did wrong. > > > > > > > > > > __________________________________ > > CelebratYahoo!'s 10th Birthday! > > Yahoo! Netrospective: 100 Moments of thWeb > > http://birthday.yahoo.com/netrospective/ > > > > > > > > __________________________________ > CelebratYahoo!'s 10th Birthday! > Yahoo! Netrospective: 100 Moments of thWeb > http://birthday.yahoo.com/netrospective/ > __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Fri Mar 4 11:33:22 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: Re: Asymmetrical delay? In-Reply-To: <14106EBD68D5D944AD82BD1672022192013C755C@xxxxxxxxxxxxxxxxxxxxxxxxxxx> References: <14106EBD68D5D944AD82BD1672022192013C755C@xxxxxxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050304113322.33c61e1c@xxxxxxxxxxxxxxxxx> OFri, 4 Mar 2005 11:19:47 -0800 "Hanson, Lars A" <lars.a.hanson@xxxxxxxxxx> wrote: > I havjuscompleted my successful installation of iproute2. One of > thmorvaluable features of NistNET was the ability to do asymmetrical > delays of outbound and inbound traffic. To emulatsatellitlink > latency, thforward link and return link arslightly different > latencies. > > Oa linux router, would I apply onlatency to eth0 and the other to > eth1, and hava cliendevice behind the router generating traffic? I > believusing NAT I would then havtwo filters, one specific to > outbound traffic frothclient address, and inbound traffic from the > defaulgateway. Does thasound correct to you? > Surthawould work. You could even do it with a bridge, without routing. tc add dev eth0 roonetedelay 1000ms tc add dev eth1 roonetedelay 1ms Fronistnet_user ayahoo.com Fri Mar 4 16:05:33 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: htb and netem Message-ID: <20050305000533.19948.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i anosure what htb does when it enqueues a packet and tries to dequeuibut does not get a packet back as ihas been delayed. i think this is causing the problebecausif i ping from multiple shell commands thethresponse time decreases. i am not sure what is causing this. thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Mon Mar 7 09:40:13 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: netewatchdog timer question Message-ID: <20050307174013.32454.qmail@xxxxxxxxxxxxxxxxxxxxxxx> whethnetem watchdog timer expires it calls qdisc_restart. will this call thdequeufunction of throoqdisc or the corresponding netem qdisc whose timer expired? __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Mon Mar 7 11:05:26 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: netewatchdog timer question In-Reply-To: <20050307174013.32454.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050307174013.32454.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050307110526.1b67d3fc@xxxxxxxxxxxxxxxxx> OMon, 7 Mar 2005 09:40:13 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > whethnetem watchdog timer expires it calls > qdisc_restart. will this call thdequeufunction of > throoqdisc or the corresponding netem qdisc whose > timer expired? Iwill call throot qdisc. For thsimplcase of just netem, this will call: qdisc_restart q->dequeu(root) netem_dequeuq->qdisc->dequeue pfifo_dequeue Ipulls thnext available packet from the transmit queue and sends ito thdevice. Fronistnet_user ayahoo.com Mon Mar 7 15:55:03 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: filter ids Message-ID: <20050307235504.27556.qmail@xxxxxxxxxxxxxxxxxxxxxxx> is ipossiblto have filter ids so that i can delete a specific filter whenorequired? i tried giving a wholdescription of thfilter to the tc filter del command buideletes all the filters at the specified device. thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Mon Mar 7 18:17:44 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: getting-setting filter handlof basfilter Message-ID: <20050308021744.15792.qmail@xxxxxxxxxxxxxxxxxxxxxxx> is thera way to sethe filter handle of a u32 type filter? or is thera way to gethe handle value after a filter has beeadded? __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froalexander acheskis.com Mon Mar 7 23:33:23 2005 From: alexander acheskis.co(Alexander Cheskis) Date: Wed Apr 18 12:51:15 2007 Subject: sequential losses In-Reply-To: <20050308021744.15792.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <200503080733.j287XTqi006781@xxxxxxxxxxxxx> Hi, I need to emulatwireless fading effects, by sequential packelost (f.e. 20 consequenpackets losses every 1000 packets) CaI do iby using random loss, only changing random distribution tables, or I need to add additional optioin netem? 10x, Alexander Froerwann.thoraval airisa.fr Tue Mar 8 07:33:36 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 12:51:15 2007 Subject: Debug messages Message-ID: <422DC5D0.8090905@xxxxxxxx> Hi all, I would likto seevery debug message generated by the pr_debug() function. Frothkernel sources, in include/linux/kernel.h: #ifdef DEBUG #definpr_debug(fmt,arg...) \ printk(KERN_DEBUG fmt,##arg) #else #definpr_debug(fmt,arg...) \ do { } whil(0) #endif I havtwo questions: - Wheris DEBUG defined ? - Wherthmessages go ? (in /var/log/kern.log or /var/log/debug ?) Thanks, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 Froshemminger aosdl.org Tue Mar 8 08:34:24 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: sequential losses In-Reply-To: <200503080733.j287XTqi006781@xxxxxxxxxxxxx> References: <20050308021744.15792.qmail@xxxxxxxxxxxxxxxxxxxxxxx> <200503080733.j287XTqi006781@xxxxxxxxxxxxx> Message-ID: <20050308083424.662a4c4b@xxxxxxxxxxxxxxxxx> OTue, 8 Mar 2005 09:33:23 +0200 "Alexander Cheskis" <alexander@xxxxxxxxxxx> wrote: > Hi, > > I need to emulatwireless fading effects, by sequential packelost (f.e. > 20 consequenpackets losses every 1000 packets) > CaI do iby using random loss, only changing random distribution tables, > or I need to add additional optioin netem? > > 10x, > Alexander Probably having a high correlatiovaluwould cause the same effect. Froshemminger aosdl.org Tue Mar 8 08:40:57 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: Debug messages In-Reply-To: <422DC5D0.8090905@xxxxxxxx> References: <422DC5D0.8090905@xxxxxxxx> Message-ID: <20050308084057.08afbc15@xxxxxxxxxxxxxxxxx> OTue, 08 Mar 2005 16:33:36 +0100 ErwanThoraval <erwann.thoraval@xxxxxxxx> wrote: > Hi all, > > I would likto seevery debug message generated by the pr_debug() > function. Frothkernel sources, in include/linux/kernel.h: > > > #ifdef DEBUG > #definpr_debug(fmt,arg...) \ > printk(KERN_DEBUG fmt,##arg) > #else > #definpr_debug(fmt,arg...) \ > do { } whil(0) > #endif > > > I havtwo questions: > > - Wheris DEBUG defined ? Iisn'normally. When I need them I just add '#define DEBUG 1' to thstarof the file. > - Wherthmessages go ? (in /var/log/kern.log or /var/log/debug ?) They comouvia kernel logging, the files are controlled by syslog.conf and ardistribution dependant. I jususe dmesg to look at the last stuff. > Thanks, > > -- > ErwanThoraval > IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France > T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 > > Tel. direct: +33 (0) 2 99 84 71 39 > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > http://lists.osdl.org/mailman/listinfo/netem Frojw afemmecomp.com Tue Mar 8 08:58:27 2005 From: jw afemmecomp.co(John Wiley) Date: Wed Apr 18 12:51:15 2007 Subject: sequential losses In-Reply-To: <20050308083424.662a4c4b@xxxxxxxxxxxxxxxxx> Message-ID: <002801c52400$0ce5e4c0$6600a8c0@ra> Juscreatthe pdf tables based on your expected fading pdf... The correlatiobetween packeloss and "RF Fade" is often not well defined, but you could simply decidthaif your BER exceeds, say, 1e-4, you start taking packehits... We'vdone some work in defining basic relationships betweePER (PackeError Rate) and BER, but this is very often implementatiospecific based on things likinterleaver length, FEC implementation, whether or noyou'vimplemented adaptive coding/modulation or adaptivpower control... Hopthahelps... Best! -jw --- JohWiley Senior Engineer FCI 14170 Newbrook DrivSuit100 Chantilly, VA 20151 703.961.1818x123 (voice) 703.817.1313 (fax) http://www.femmecomp.com > -----Original Message----- > From: netem-bounces@xxxxxxxxxxxxxx > [mailto:netem-bounces@xxxxxxxxxxxxxx] OBehalf Of Stephen Hemminger > Sent: Tuesday, March 08, 2005 11:34 AM > To: Alexander Cheskis > Cc: netem@xxxxxxxxxxxxxx > Subject: Re: sequential losses > > > OTue, 8 Mar 2005 09:33:23 +0200 > "Alexander Cheskis" <alexander@xxxxxxxxxxx> wrote: > > > Hi, > > > > I need to emulatwireless fading effects, by sequential > packelos > > (f.e. 20 consequenpackets losses every 1000 packets) Can > I do iby > > using randoloss, only changing randodistribution > tables, or I need > > to add additional optioin netem? > > > > 10x, > > Alexander > > Probably having a high correlatiovaluwould cause the same effect. > Fronistnet_user ayahoo.com Tue Mar 8 09:54:08 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: u32 filter handles Message-ID: <20050308175408.99634.qmail@xxxxxxxxxxxxxxxxxxxxxxx> does each of thu32 filter has a handlstarting with 800:: i tried add filters with thcommands tc filter add dev eth0 paren1:0 handl700::201 prio 1 u32 which gives aerror. buif i do tc filter add dev eth0 paren1:0 handl800::201 prio 1 u32 thea filter is created. even if i do tc filter add dev eth0 paren1:0 handl::201 prio 1 u32 thea filter with handl800::201 is created. is this somsorof a rule or just a coincidence? thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Tue Mar 8 10:39:58 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: htb deactivataffects netem Message-ID: <20050308183958.98896.qmail@xxxxxxxxxxxxxxxxxxxxxxx> whais thpurpose of deactivating a class in htb wheon dequeuno packet is returned? is it optimizing thcodor some other technical issue? can i disablthis class deactivation as iaffects the working of netefor latency and duplication? thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Frostephen.earl abt.com Fri Mar 11 06:46:42 2005 From: stephen.earl abt.co(stephen.earl@xxxxxx) Date: Wed Apr 18 12:51:15 2007 Subject: PHPNetemGUI 0.7 is available. Message-ID: <B1A7C3B14EAAED48B27E84CBE6FFB37307AF94DF@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> Versio0.7 of my neteweb interface is available for download from: www.smyles.plus.com/phpnetemgui Below is a lisof changes: 2005-03-04 V0.7 -Added aIP address matching featurso you can limit netem to a specific source and destination ip address or address block -Added thability to view thcommands that will be run when you click te 'apply settings' button instead of just running them in the background. This will allow you to copy and modify the commands for possible use on other machines. -Added a buttoto stop and starall running netem rules. The commands get stored in a text file so that you can always stop and start the last set of rules that were used. I acurrently working on version 0.8, which will havallow you to have more than one set of netem rules per interface, each set matching different ip source and destination address values. Fronistnet_user ayahoo.com Mon Mar 14 12:16:03 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:15 2007 Subject: neteand latency variation Message-ID: <20050314201603.17994.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i havtwo interfaces ( A & B ) connected back to back. oon(interface A) i configure netem to add a 1ms delay. whei ping froA to B then i get a delay of 1ms (with a variatioof .03 ms) buwhen i ping froB to A i gea vairation of 0.25 ms . what happens is thathlatency starts from 1ms, goes down to 0.1 ms and thecycles back to 1ms and so on. i am noablto understand the reason for this behaivior. is ith1ms resolution of the clock? or is it just coincidencthathe variation in one direction is less thathother? thanks. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ Froerwann.thoraval airisa.fr Wed Mar 16 05:46:51 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 12:51:15 2007 Subject: Trying to modify thsourccode Message-ID: <423838CB.4010400@xxxxxxxx> Hi, I try to modify thloss process in sch_netem.c to approximatthe 2-states Gilbermodel. To do so, I need to ussrand() and drand() fro<stdlib.h>. This is the firstimfor me to modify a module and I fail. I add #includ<stdlib.h> athe beginning of sch_netem.c. I try to compilwith: /home/user$ mak-C /usr/src/linux SUBDIRS=$PWD modules Thfil<stdlib.h> is not included in the kernel tree. So I tried to pu#includ"/usr/include/stdlib.h" instead. However, <stdlib.h> needs other .h i/usr/includwhich can't be found. Whais thmore convenient way to use .h files from /usr/include into kernel modulsourccode ? -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Frodunk_palmer ayahoo.com Wed Mar 16 14:14:42 2005 From: dunk_palmer ayahoo.co(Duncan Palmer) Date: Wed Apr 18 12:51:15 2007 Subject: Trying to modify thsourccode In-Reply-To: 6667 Message-ID: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Hi Erwann, You can'usanything from the C standard library in thkernel. Your besbet may be to implement a pseudo randonumber generator yourself - you can gea true randoseed for ifrom the in kernel random driver (sedrivers/char/random.c). If you'rnew akernel programming, I'd suggest you hava look a'Linux Device Drivers' published by O'Reilly - there's aonlinversion somewhere, and also atharticles on LWN about porting drivers to 2.6 - http://lwn.net/Articles/driver-porting/ Dunk --- ErwanThoraval <erwann.thoraval@xxxxxxxx> wrote: > Hi, > > I try to modify thloss process in sch_netem.c to > approximatth > 2-states Gilbermodel. > > To do so, I need to ussrand() and drand() from > <stdlib.h>. This is th > firstimfor me to modify a module and I fail. > > I add #includ<stdlib.h> athe beginning of > sch_netem.c. I try to > compilwith: > > /home/user$ mak-C /usr/src/linux SUBDIRS=$PWD > modules > > Thfil<stdlib.h> is not included in the kernel > tree. So I tried to > pu#includ"/usr/include/stdlib.h" instead. > However, <stdlib.h> needs > other .h i/usr/includwhich can't be found. > > Whais thmore convenient way to use .h files from > /usr/includinto > kernel modulsourccode ? > > > -- > ErwanThoraval > IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, > France > Tél: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 > 71 > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > http://lists.osdl.org/mailman/listinfo/netem > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has thbesspam protection around http://mail.yahoo.co Froshemminger aosdl.org Wed Mar 16 14:25:47 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: Trying to modify thsourccode In-Reply-To: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050316142547.52f7291a@xxxxxxxxxxxxxxxxx> OWed, 16 Mar 2005 14:14:42 -0800 (PST) DuncaPalmer <dunk_palmer@xxxxxxxxx> wrote: > Hi Erwann, > > You can'usanything from the C standard library in > thkernel. Your besbet may be to implement a pseudo > randonumber generator yourself - you can gea true > randoseed for ifrom the in kernel random driver > (sedrivers/char/random.c). Usnet_random() igives a 32bit unsigned long. > If you'rnew akernel programming, I'd suggest you > hava look a'Linux Device Drivers' published by > O'Reilly - there's aonlinversion somewhere, and > also atharticles on LWN about porting drivers to > 2.6 - http://lwn.net/Articles/driver-porting/ Theris a new version: http://lwn.net/kernel/LDD3 Froshemminger aosdl.org Wed Mar 16 14:30:20 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:15 2007 Subject: neteand latency variation In-Reply-To: <20050314201603.17994.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050314201603.17994.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050316143020.10e6e8de@xxxxxxxxxxxxxxxxx> OMon, 14 Mar 2005 12:16:03 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > i havtwo interfaces ( A & B ) connected back to > back. oon(interface A) i configure netem to add a > 1ms delay. whei ping froA to B then i get a delay > of 1ms (with a variatioof .03 ms) buwhen i ping > froB to A i gea vairation of 0.25 ms . what > happens is thathlatency starts from 1ms, goes down > to 0.1 ms and thecycles back to 1ms and so on. i am > noablto understand the reason for this behaivior. > is ith1ms resolution of the clock? Probably, ondevicis using NAPI and the other is not. NAPI drivers scalbetter bucan have greaer latency. You could check by jususing a fifo and see if nothing happens. > or is ijust > coincidencthathe variation in one direction is > less thathother? > > thanks. > > > > __________________________________ > Do you Yahoo!? > Yahoo! Small Business - Try our new resources site! > http://smallbusiness.yahoo.com/resources/ Froerwann.thoraval airisa.fr Thu Mar 17 02:05:38 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 12:51:16 2007 Subject: Trying to modify thsourccode In-Reply-To: <20050316142547.52f7291a@xxxxxxxxxxxxxxxxx> References: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> <20050316142547.52f7291a@xxxxxxxxxxxxxxxxx> Message-ID: <42395672.2080304@xxxxxxxx> StepheHemminger wrote: > Theris a new version: > http://lwn.net/kernel/LDD3 Actually, iis http://lwn.net/Kernel/LDD3 -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Frorajahammad agmail.com Thu Mar 17 23:34:32 2005 From: rajahammad agmail.co(Raja Muhammad Hammad) Date: Wed Apr 18 12:51:16 2007 Subject: No Packeloss for RTP ??? Message-ID: <3ddf6cd3050317233466f7e39e@xxxxxxxxxxxxxx> hi I ausing video streaming server and clien(LiveMedia). I have two machines to testhvideo transmission. I used to netem to emulate WAothe client side but it does not work for RTP/UDP video transmission. e.g. # tc qdisc add dev eth0 rooneteloss 10% EveI puthe value 25%, I get no packet loss. If I puhigh valulike 30% and above I get no transmission at all. 10% and abovpretty hugvalues so I do not understand why I dont get any packeloss for RTP/UDP. WheI usping to test the trasnmission, I get packet loss as expected. I ausing Fedora cor2 with kernel 2.6-10 on both machines. Any possiblreasons...? Regards Hammad Frostefanie.wenk arbt-nbg.de Mon Mar 21 04:23:35 2005 From: stefanie.wenk arbt-nbg.d(Stefanie Wenk) Date: Wed Apr 18 12:51:16 2007 Subject: Problewith replace Message-ID: <423EBCC7.6030201@xxxxxxxxxx> Hello everybody! I aworking on a gui for neteand I'm already quite finished with it. Heramy work, netem is used for testing different network scenarios. This demands e.g. a changof thinterface settings in the middle of a running tesvia thgui. Until now I solved thprobleof changing by deleting the qdisc with "tc qdisc del dev DEV root" and adding inewly. Buunfortunaly this means thathinterface is "free" for a short time, and produces a peak iour measures. I tried to solvthis by using "replace", but this doesn'work if ratcontrol is desired. I realised thratcontrol the same way as shown in the netem examples othhomepage with tbf. The error is "invalid argument". Why does replacnowork here? Is there an alternative solution? I would bvery happy for somadvice and some help! Greetings Steffi Froshemminger aosdl.org Mon Mar 28 15:59:34 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:16 2007 Subject: Problewith replace In-Reply-To: <423EBCC7.6030201@xxxxxxxxxx> References: <423EBCC7.6030201@xxxxxxxxxx> Message-ID: <20050328155934.71f445cc@xxxxxxxxxxxxxxxxx> OMon, 21 Mar 2005 13:23:35 +0100 StefaniWenk <stefanie.wenk@xxxxxxxxxx> wrote: > Hello everybody! > I aworking on a gui for neteand I'm already quite finished with it. > Heramy work, netem is used for testing different network scenarios. > This demands e.g. a changof thinterface settings in the middle of a > running tesvia thgui. > Until now I solved thprobleof changing by deleting the qdisc with > "tc qdisc del dev DEV root" and adding inewly. Buunfortunaly this > means thathinterface is "free" for a short time, and produces a peak > iour measures. I tried to solvthis by using "replace", but this > doesn'work if ratcontrol is desired. > I realised thratcontrol the same way as shown in the netem examples > othhomepage with tbf. The error is "invalid argument". Why does > replacnowork here? Aryou replacing thTBF qdisc or the netem discipline. It probably is a bug ithtbf module that can be fixed. > Is theran alternativsolution? Longer term, I intend to makthqdisc into a full kobject and have thelivin /sys like the netdevices do. This would allow qdisc to exporattributes as files thacould be tweaked by things like your GUI. Buno codhas been done, and it has some issues about how to namand nesqdisc objects in the namespace. Currenthoughwas something like: /sys/class/net/eth0/qdisc would bsymlink to rooqdisc in /sys/class/qdisc/netem/1:0 /sys/class/qdisc/netem/1:0/delay would contaihook for delay /sys/class/qdisc/netem/1:0/10 would bsymlink to /sys/class/qdisc/tbf/ Still thinking and too busy to implemenit. > I would bvery happy for somadvice and some help! > Greetings > Steffi Fronistnet_user ayahoo.com Wed Mar 30 10:05:04 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:16 2007 Subject: netewith prio hangs on duplicate Message-ID: <20050330180504.75838.qmail@xxxxxxxxxxxxxxxxxxxxxxx> hi i tried thexamplgiven on the examples page to duplicatselected traffic like tc qdisc add dev eth0 roohandl1: prio tc qdisc add dev eth0 paren1:3 handl3: netem duplicat40% tc filter add dev eth0 protocol ip paren1:0 prio 3 u32 match ip ds11.0.2.2 flowid 1:3 whei ping fro11.0.2.2 to this interface my machine hangs. thsamthing works for drop or delay. i would appreciatif someoncan tell me what i am doing wrong. thanks. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ Fronistnet_user ayahoo.com Tue Mar 1 10:46:42 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: whartnetlink_dump_ifinfo does? Message-ID: <20050301184643.79442.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i anew to netlink. its been somtime since i have been trying to figure out what rtnetlink_dump_ifinfo does. could someone please help me on this. thanks. --------------------------------- Do you Yahoo!? Yahoo! Mail - Find whayou need with new enhanced search. Learn more. -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20050301/fd5ba25c/attachment-0001.htm Fronistnet_user ayahoo.com Tue Mar 1 14:21:29 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: wherdoes tc requests rtnetlink_dump_ifinfo Message-ID: <20050301222129.39635.qmail@xxxxxxxxxxxxxxxxxxxxxxx> hi whei executtc qdisc add dev eth0 root netem rtnetlink_dump_ifinfo is called which traverses the lisof network devices and calls rtnetlink_fill_ifinfo i anoable to figure out what why is the function rtnetlink_dump_ifinfo called and wheriis requested. fromy understanding its is simply dumping thinformation for all thinterfaces. i would appreciatif someoncould could point me to the call. thanks. __________________________________ Do you Yahoo!? Read only thmail you wan- Yahoo! Mail SpamGuard. http://promotions.yahoo.com/new_mail Froerwann.thoraval airisa.fr Wed Mar 2 06:33:46 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 17:37:45 2007 Subject: Using th'loss' option Message-ID: <4225CECA.30802@xxxxxxxx> Hello, Iseems thathe 'loss' option is not included in the 'tc' command. Heris thexample provided by the Netem website: tc qdisc changdev eth0 roonetem loss .1% And heris thresult of this command: RTNETLINK answers: No such filor directory Indeed, running 'tc qdisc changdev eth0 roonetem help' gives: Usage: ... nete[ limiPACKETS ] [ delay TIME [ JITTER [CORRELATION]]] [ drop PERCENT [CORRELATION]] [ duplicatPERCENT [CORRELATION]] [ distributio{uniform|normal|pareto|paretonormal} ] [ gap PACKETS ] No 'loss' optioprovided. I arunning a 2.6.10-gentoo-r7 with iproute2_2.6.10.20050124. Any idea ? Regards, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 Frostephen.earl abt.com Wed Mar 2 06:37:08 2005 From: stephen.earl abt.co(stephen.earl@xxxxxx) Date: Wed Apr 18 17:37:45 2007 Subject: Using th'loss' option Message-ID: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> You need to typ'add' instead of'change' to add thrule. 'change' only works if you already have a rule running on that interface. -----Original Message----- From: netem-bounces@xxxxxxxxxxxxxx [mailto:netem-bounces@xxxxxxxxxxxxxx]OBehalf Of Erwann Thoraval Sent: 02 March 2005 14:34 To: netem@xxxxxxxxxxxxxx Subject: Using th'loss' option Hello, Iseems thathe 'loss' option is not included in the 'tc' command. Heris thexample provided by the Netem website: tc qdisc changdev eth0 roonetem loss .1% And heris thresult of this command: RTNETLINK answers: No such filor directory Indeed, running 'tc qdisc changdev eth0 roonetem help' gives: Usage: ... nete[ limiPACKETS ] [ delay TIME [ JITTER [CORRELATION]]] [ drop PERCENT [CORRELATION]] [ duplicatPERCENT [CORRELATION]] [ distributio{uniform|normal|pareto|paretonormal} ] [ gap PACKETS ] No 'loss' optioprovided. I arunning a 2.6.10-gentoo-r7 with iproute2_2.6.10.20050124. Any idea ? Regards, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 _______________________________________________ Netemailing list Netem@xxxxxxxxxxxxxx http://lists.osdl.org/mailman/listinfo/netem Froerwann.thoraval airisa.fr Wed Mar 2 06:54:02 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 17:37:45 2007 Subject: Using th'loss' option In-Reply-To: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> References: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <4225D38A.9070206@xxxxxxxx> stephen.earl@bt.cowrote: > You need to typ'add' instead of'change' to add thrule. 'change' only works if you already have a rule running on that interface. Indeed, iworks with 'add', thanks ! Nevertheless, th'loss' option is nogiven by the 'help': >Usage: ... nete[ limiPACKETS ] > [ delay TIME [ JITTER [CORRELATION]]] > [ drop PERCENT [CORRELATION]] > [ duplicatPERCENT [CORRELATION]] > [ distributiouniform|normal|pareto|paretonormal}] > [ gap PACKETS ] Maybthis will bfixed soon. Regards, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 Frojw afemmecomp.com Wed Mar 2 07:05:28 2005 From: jw afemmecomp.co(John Wiley) Date: Wed Apr 18 17:37:45 2007 Subject: Using th'loss' option In-Reply-To: <4225D38A.9070206@xxxxxxxx> References: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <4225D38A.9070206@xxxxxxxx> Message-ID: <60948.69.143.124.86.1109775928.squirrel@xxxxxxxxxxxxxxxxx> Erwann, My understanding is thath"loss" option and the "drop" option are synonymous (packeloss and packe"drop" are essentially the same phenomenofronetem's standpoint)... BesRegards! -jw > stephen.earl@bt.cowrote: >> You need to typ'add' instead of'change' to add thrule. 'change' >> only works if you already hava rulrunning on that interface. > > Indeed, iworks with 'add', thanks ! > > Nevertheless, th'loss' option is nogiven by the 'help': > > >Usage: ... nete[ limiPACKETS ] > > [ delay TIME [ JITTER [CORRELATION]]] > > [ drop PERCENT [CORRELATION]] > > [ duplicatPERCENT [CORRELATION]] > > [ distribution > uniform|normal|pareto|paretonormal}] > > [ gap PACKETS ] > > Maybthis will bfixed soon. > > Regards, > -- > ErwanThoraval > IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France > T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 > > Tel. direct: +33 (0) 2 99 84 71 39 > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > http://lists.osdl.org/mailman/listinfo/netem Frojw afemmecomp.com Wed Mar 2 07:07:43 2005 From: jw afemmecomp.co(John Wiley) Date: Wed Apr 18 17:37:45 2007 Subject: Patch Archive... In-Reply-To: <4225D38A.9070206@xxxxxxxx> References: <B1A7C3B14EAAED48B27E84CBE6FFB37307A31552@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <4225D38A.9070206@xxxxxxxx> Message-ID: <60957.69.143.124.86.1109776063.squirrel@xxxxxxxxxxxxxxxxx> All, Ilooking for an archivof the patches that have been floating around herfor thlast weeks... (even a message archive would be helpful!) Is thersuch a thing available? Thmailing list link on the netem web pagaOSDL doesn't seem to work... BesRegards! -jw Fronistnet_user ayahoo.com Wed Mar 2 17:12:00 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: dynamically increasing bands iprio Message-ID: <20050303011201.80951.qmail@xxxxxxxxxxxxxxxxxxxxxxx> is ipossiblto increase bands in prio dynamically. by defaulihas 3 bands, after attaching prio to an interface, cai increasthe number of bands to lets say 10? or is ionly possiblwhen you attach prio to ainterface? thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Thu Mar 3 09:52:57 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: dynamically increasing bands iprio In-Reply-To: <20050303011201.80951.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050303011201.80951.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050303095257.72990772@xxxxxxxxxxxxxxxxx> OWed, 2 Mar 2005 17:12:00 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > is ipossiblto increase bands in prio dynamically. > by defaulihas 3 bands, after attaching prio to an > interface, cai increasthe number of bands to lets > say 10? or is ionly possiblwhen you attach prio to > ainterface? > Thbesway to answer questions like this is to read the source. Thnumber of bands in thprio scheduler is controlled by the constanTCQ_PRIO_BANDS. Iis defined in include/linux/pkt_sched.h to b16. Fronistnet_user ayahoo.com Thu Mar 3 14:35:13 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: htb and netem- unexpected results Message-ID: <20050303223514.94354.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i experimented attaching neteto htb by doing the following tc qdisc add dev eth0 roohandl1: htb default 11 tc class add dev eth0 paren1:0 handl1:1 htb rate 1000mbiceil 1000mbit tc class add dev eth0 paren1:1 handl1:10 htb rate 500mbiceil 500mbit tc class add dev eth0 paren1:1 handl1:11 htb rate 500mbiceil 500mbit tc filter add dev eth0 protocol ip paren1: prio 1 u32 match ip ds11.0.0.1 flowid 1:10 __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Thu Mar 3 14:38:23 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: htb and netem- unexpected results In-Reply-To: <20050303223514.94354.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050303223514.94354.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050303143823.7c942022@xxxxxxxxxxxxxxxxx> OThu, 3 Mar 2005 14:35:13 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > i experimented attaching neteto htb by doing the > following > > > > tc qdisc add dev eth0 roohandl1: htb default 11 > > tc class add dev eth0 paren1:0 handl1:1 htb rate > 1000mbiceil 1000mbit > > tc class add dev eth0 paren1:1 handl1:10 htb rate > 500mbiceil 500mbit > > tc class add dev eth0 paren1:1 handl1:11 htb rate > 500mbiceil 500mbit > > tc filter add dev eth0 protocol ip paren1: prio 1 > u32 match ip ds11.0.0.1 flowid 1:10 > whahappened? Fronistnet_user ayahoo.com Thu Mar 3 14:39:52 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: netewith htb - mordelay than expected Message-ID: <20050303223953.50646.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i experimented attaching neteto htb by doing the following eth0 has ip 11.0.0.2 tc qdisc add dev eth0 roohandl1: htb default 11 tc class add dev eth0 paren1:0 handl1:1 htb rate 1000mbiceil 1000mbit tc class add dev eth0 paren1:1 handl1:10 htb rate 500mbiceil 500mbit tc class add dev eth0 paren1:1 handl1:11 htb rate 500mbiceil 500mbit tc filter add dev eth0 protocol ip paren1: prio 1 u32 match ip ds11.0.0.1 flowid 1:10 uptill now if i ping 11.0.0.1 i gea responstime of 0.1 ms nexi attached neteby tc qdisc add dev eth0 paren1:10 handl20: netem latency 10ms now id i ping 11.0.0.1 i gea responstime of 1000ms i anosure what i did wrong. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Thu Mar 3 18:00:17 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: Re: netewith htb - mordelay than expected Message-ID: <20050304020017.54118.qmail@xxxxxxxxxxxxxxxxxxxxxxx> whai havnoticed is that the delay gets round off to thneares1000th of ms. for 1ms i get 1000ms, for 1001ms i ge2000ms... i anot sure what i am doing wrong --- js si <nistnet_user@xxxxxxxxx> wrote: > i experimented attaching neteto htb by doing the > following > > eth0 has ip 11.0.0.2 > > tc qdisc add dev eth0 roohandl1: htb default 11 > > tc class add dev eth0 paren1:0 handl1:1 htb rate > 1000mbiceil 1000mbit > > tc class add dev eth0 paren1:1 handl1:10 htb > rate > 500mbiceil 500mbit > > tc class add dev eth0 paren1:1 handl1:11 htb > rate > 500mbiceil 500mbit > > tc filter add dev eth0 protocol ip paren1: prio 1 > u32 match ip ds11.0.0.1 flowid 1:10 > > uptill now if i ping 11.0.0.1 i gea responstime > of > 0.1 ms > > nexi attached neteby > > tc qdisc add dev eth0 paren1:10 handl20: netem > latency 10ms > > now id i ping 11.0.0.1 i gea responstime of > 1000ms > > i anosure what i did wrong. > > > > > __________________________________ > CelebratYahoo!'s 10th Birthday! > Yahoo! Netrospective: 100 Moments of thWeb > http://birthday.yahoo.com/netrospective/ > __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Thu Mar 3 19:23:15 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: Re: netewith htb - mordelay than expected Message-ID: <20050304032315.48348.qmail@xxxxxxxxxxxxxxxxxxxxxxx> another thing thai tried was to pua prio in betweehtb and netem. then iworks fine. --- js si <nistnet_user@xxxxxxxxx> wrote: > whai havnoticed is that the delay gets round off > to thneares1000th of ms. for 1ms i get 1000ms, > for > 1001ms i ge2000ms... i anot sure what i am doing > wrong > > > --- js si <nistnet_user@xxxxxxxxx> wrote: > > > i experimented attaching neteto htb by doing the > > following > > > > eth0 has ip 11.0.0.2 > > > > tc qdisc add dev eth0 roohandl1: htb default > 11 > > > > tc class add dev eth0 paren1:0 handl1:1 htb > rate > > 1000mbiceil 1000mbit > > > > tc class add dev eth0 paren1:1 handl1:10 htb > > rate > > 500mbiceil 500mbit > > > > tc class add dev eth0 paren1:1 handl1:11 htb > > rate > > 500mbiceil 500mbit > > > > tc filter add dev eth0 protocol ip paren1: prio > 1 > > u32 match ip ds11.0.0.1 flowid 1:10 > > > > uptill now if i ping 11.0.0.1 i gea response > time > > of > > 0.1 ms > > > > nexi attached neteby > > > > tc qdisc add dev eth0 paren1:10 handl20: netem > > latency 10ms > > > > now id i ping 11.0.0.1 i gea responstime of > > 1000ms > > > > i anosure what i did wrong. > > > > > > > > > > __________________________________ > > CelebratYahoo!'s 10th Birthday! > > Yahoo! Netrospective: 100 Moments of thWeb > > http://birthday.yahoo.com/netrospective/ > > > > > > > > __________________________________ > CelebratYahoo!'s 10th Birthday! > Yahoo! Netrospective: 100 Moments of thWeb > http://birthday.yahoo.com/netrospective/ > __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Fri Mar 4 11:33:22 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: Re: Asymmetrical delay? In-Reply-To: <14106EBD68D5D944AD82BD1672022192013C755C@xxxxxxxxxxxxxxxxxxxxxxxxxxx> References: <14106EBD68D5D944AD82BD1672022192013C755C@xxxxxxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050304113322.33c61e1c@xxxxxxxxxxxxxxxxx> OFri, 4 Mar 2005 11:19:47 -0800 "Hanson, Lars A" <lars.a.hanson@xxxxxxxxxx> wrote: > I havjuscompleted my successful installation of iproute2. One of > thmorvaluable features of NistNET was the ability to do asymmetrical > delays of outbound and inbound traffic. To emulatsatellitlink > latency, thforward link and return link arslightly different > latencies. > > Oa linux router, would I apply onlatency to eth0 and the other to > eth1, and hava cliendevice behind the router generating traffic? I > believusing NAT I would then havtwo filters, one specific to > outbound traffic frothclient address, and inbound traffic from the > defaulgateway. Does thasound correct to you? > Surthawould work. You could even do it with a bridge, without routing. tc add dev eth0 roonetedelay 1000ms tc add dev eth1 roonetedelay 1ms Fronistnet_user ayahoo.com Fri Mar 4 16:05:33 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: htb and netem Message-ID: <20050305000533.19948.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i anosure what htb does when it enqueues a packet and tries to dequeuibut does not get a packet back as ihas been delayed. i think this is causing the problebecausif i ping from multiple shell commands thethresponse time decreases. i am not sure what is causing this. thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Mon Mar 7 09:40:13 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: netewatchdog timer question Message-ID: <20050307174013.32454.qmail@xxxxxxxxxxxxxxxxxxxxxxx> whethnetem watchdog timer expires it calls qdisc_restart. will this call thdequeufunction of throoqdisc or the corresponding netem qdisc whose timer expired? __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froshemminger aosdl.org Mon Mar 7 11:05:26 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: netewatchdog timer question In-Reply-To: <20050307174013.32454.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050307174013.32454.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050307110526.1b67d3fc@xxxxxxxxxxxxxxxxx> OMon, 7 Mar 2005 09:40:13 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > whethnetem watchdog timer expires it calls > qdisc_restart. will this call thdequeufunction of > throoqdisc or the corresponding netem qdisc whose > timer expired? Iwill call throot qdisc. For thsimplcase of just netem, this will call: qdisc_restart q->dequeu(root) netem_dequeuq->qdisc->dequeue pfifo_dequeue Ipulls thnext available packet from the transmit queue and sends ito thdevice. Fronistnet_user ayahoo.com Mon Mar 7 15:55:03 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: filter ids Message-ID: <20050307235504.27556.qmail@xxxxxxxxxxxxxxxxxxxxxxx> is ipossiblto have filter ids so that i can delete a specific filter whenorequired? i tried giving a wholdescription of thfilter to the tc filter del command buideletes all the filters at the specified device. thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Mon Mar 7 18:17:44 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: getting-setting filter handlof basfilter Message-ID: <20050308021744.15792.qmail@xxxxxxxxxxxxxxxxxxxxxxx> is thera way to sethe filter handle of a u32 type filter? or is thera way to gethe handle value after a filter has beeadded? __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Froalexander acheskis.com Mon Mar 7 23:33:23 2005 From: alexander acheskis.co(Alexander Cheskis) Date: Wed Apr 18 17:37:45 2007 Subject: sequential losses In-Reply-To: <20050308021744.15792.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <200503080733.j287XTqi006781@xxxxxxxxxxxxx> Hi, I need to emulatwireless fading effects, by sequential packelost (f.e. 20 consequenpackets losses every 1000 packets) CaI do iby using random loss, only changing random distribution tables, or I need to add additional optioin netem? 10x, Alexander Froerwann.thoraval airisa.fr Tue Mar 8 07:33:36 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 17:37:45 2007 Subject: Debug messages Message-ID: <422DC5D0.8090905@xxxxxxxx> Hi all, I would likto seevery debug message generated by the pr_debug() function. Frothkernel sources, in include/linux/kernel.h: #ifdef DEBUG #definpr_debug(fmt,arg...) \ printk(KERN_DEBUG fmt,##arg) #else #definpr_debug(fmt,arg...) \ do { } whil(0) #endif I havtwo questions: - Wheris DEBUG defined ? - Wherthmessages go ? (in /var/log/kern.log or /var/log/debug ?) Thanks, -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Tel. direct: +33 (0) 2 99 84 71 39 Froshemminger aosdl.org Tue Mar 8 08:34:24 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: sequential losses In-Reply-To: <200503080733.j287XTqi006781@xxxxxxxxxxxxx> References: <20050308021744.15792.qmail@xxxxxxxxxxxxxxxxxxxxxxx> <200503080733.j287XTqi006781@xxxxxxxxxxxxx> Message-ID: <20050308083424.662a4c4b@xxxxxxxxxxxxxxxxx> OTue, 8 Mar 2005 09:33:23 +0200 "Alexander Cheskis" <alexander@xxxxxxxxxxx> wrote: > Hi, > > I need to emulatwireless fading effects, by sequential packelost (f.e. > 20 consequenpackets losses every 1000 packets) > CaI do iby using random loss, only changing random distribution tables, > or I need to add additional optioin netem? > > 10x, > Alexander Probably having a high correlatiovaluwould cause the same effect. Froshemminger aosdl.org Tue Mar 8 08:40:57 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: Debug messages In-Reply-To: <422DC5D0.8090905@xxxxxxxx> References: <422DC5D0.8090905@xxxxxxxx> Message-ID: <20050308084057.08afbc15@xxxxxxxxxxxxxxxxx> OTue, 08 Mar 2005 16:33:36 +0100 ErwanThoraval <erwann.thoraval@xxxxxxxx> wrote: > Hi all, > > I would likto seevery debug message generated by the pr_debug() > function. Frothkernel sources, in include/linux/kernel.h: > > > #ifdef DEBUG > #definpr_debug(fmt,arg...) \ > printk(KERN_DEBUG fmt,##arg) > #else > #definpr_debug(fmt,arg...) \ > do { } whil(0) > #endif > > > I havtwo questions: > > - Wheris DEBUG defined ? Iisn'normally. When I need them I just add '#define DEBUG 1' to thstarof the file. > - Wherthmessages go ? (in /var/log/kern.log or /var/log/debug ?) They comouvia kernel logging, the files are controlled by syslog.conf and ardistribution dependant. I jususe dmesg to look at the last stuff. > Thanks, > > -- > ErwanThoraval > IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France > T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 > > Tel. direct: +33 (0) 2 99 84 71 39 > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > http://lists.osdl.org/mailman/listinfo/netem Frojw afemmecomp.com Tue Mar 8 08:58:27 2005 From: jw afemmecomp.co(John Wiley) Date: Wed Apr 18 17:37:45 2007 Subject: sequential losses In-Reply-To: <20050308083424.662a4c4b@xxxxxxxxxxxxxxxxx> Message-ID: <002801c52400$0ce5e4c0$6600a8c0@ra> Juscreatthe pdf tables based on your expected fading pdf... The correlatiobetween packeloss and "RF Fade" is often not well defined, but you could simply decidthaif your BER exceeds, say, 1e-4, you start taking packehits... We'vdone some work in defining basic relationships betweePER (PackeError Rate) and BER, but this is very often implementatiospecific based on things likinterleaver length, FEC implementation, whether or noyou'vimplemented adaptive coding/modulation or adaptivpower control... Hopthahelps... Best! -jw --- JohWiley Senior Engineer FCI 14170 Newbrook DrivSuit100 Chantilly, VA 20151 703.961.1818x123 (voice) 703.817.1313 (fax) http://www.femmecomp.com > -----Original Message----- > From: netem-bounces@xxxxxxxxxxxxxx > [mailto:netem-bounces@xxxxxxxxxxxxxx] OBehalf Of Stephen Hemminger > Sent: Tuesday, March 08, 2005 11:34 AM > To: Alexander Cheskis > Cc: netem@xxxxxxxxxxxxxx > Subject: Re: sequential losses > > > OTue, 8 Mar 2005 09:33:23 +0200 > "Alexander Cheskis" <alexander@xxxxxxxxxxx> wrote: > > > Hi, > > > > I need to emulatwireless fading effects, by sequential > packelos > > (f.e. 20 consequenpackets losses every 1000 packets) Can > I do iby > > using randoloss, only changing randodistribution > tables, or I need > > to add additional optioin netem? > > > > 10x, > > Alexander > > Probably having a high correlatiovaluwould cause the same effect. > Fronistnet_user ayahoo.com Tue Mar 8 09:54:08 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: u32 filter handles Message-ID: <20050308175408.99634.qmail@xxxxxxxxxxxxxxxxxxxxxxx> does each of thu32 filter has a handlstarting with 800:: i tried add filters with thcommands tc filter add dev eth0 paren1:0 handl700::201 prio 1 u32 which gives aerror. buif i do tc filter add dev eth0 paren1:0 handl800::201 prio 1 u32 thea filter is created. even if i do tc filter add dev eth0 paren1:0 handl::201 prio 1 u32 thea filter with handl800::201 is created. is this somsorof a rule or just a coincidence? thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Fronistnet_user ayahoo.com Tue Mar 8 10:39:58 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: htb deactivataffects netem Message-ID: <20050308183958.98896.qmail@xxxxxxxxxxxxxxxxxxxxxxx> whais thpurpose of deactivating a class in htb wheon dequeuno packet is returned? is it optimizing thcodor some other technical issue? can i disablthis class deactivation as iaffects the working of netefor latency and duplication? thanks. __________________________________ CelebratYahoo!'s 10th Birthday! Yahoo! Netrospective: 100 Moments of thWeb http://birthday.yahoo.com/netrospective/ Frostephen.earl abt.com Fri Mar 11 06:46:42 2005 From: stephen.earl abt.co(stephen.earl@xxxxxx) Date: Wed Apr 18 17:37:45 2007 Subject: PHPNetemGUI 0.7 is available. Message-ID: <B1A7C3B14EAAED48B27E84CBE6FFB37307AF94DF@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> Versio0.7 of my neteweb interface is available for download from: www.smyles.plus.com/phpnetemgui Below is a lisof changes: 2005-03-04 V0.7 -Added aIP address matching featurso you can limit netem to a specific source and destination ip address or address block -Added thability to view thcommands that will be run when you click te 'apply settings' button instead of just running them in the background. This will allow you to copy and modify the commands for possible use on other machines. -Added a buttoto stop and starall running netem rules. The commands get stored in a text file so that you can always stop and start the last set of rules that were used. I acurrently working on version 0.8, which will havallow you to have more than one set of netem rules per interface, each set matching different ip source and destination address values. Fronistnet_user ayahoo.com Mon Mar 14 12:16:03 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: neteand latency variation Message-ID: <20050314201603.17994.qmail@xxxxxxxxxxxxxxxxxxxxxxx> i havtwo interfaces ( A & B ) connected back to back. oon(interface A) i configure netem to add a 1ms delay. whei ping froA to B then i get a delay of 1ms (with a variatioof .03 ms) buwhen i ping froB to A i gea vairation of 0.25 ms . what happens is thathlatency starts from 1ms, goes down to 0.1 ms and thecycles back to 1ms and so on. i am noablto understand the reason for this behaivior. is ith1ms resolution of the clock? or is it just coincidencthathe variation in one direction is less thathother? thanks. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/ Froerwann.thoraval airisa.fr Wed Mar 16 05:46:51 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 17:37:45 2007 Subject: Trying to modify thsourccode Message-ID: <423838CB.4010400@xxxxxxxx> Hi, I try to modify thloss process in sch_netem.c to approximatthe 2-states Gilbermodel. To do so, I need to ussrand() and drand() fro<stdlib.h>. This is the firstimfor me to modify a module and I fail. I add #includ<stdlib.h> athe beginning of sch_netem.c. I try to compilwith: /home/user$ mak-C /usr/src/linux SUBDIRS=$PWD modules Thfil<stdlib.h> is not included in the kernel tree. So I tried to pu#includ"/usr/include/stdlib.h" instead. However, <stdlib.h> needs other .h i/usr/includwhich can't be found. Whais thmore convenient way to use .h files from /usr/include into kernel modulsourccode ? -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Frodunk_palmer ayahoo.com Wed Mar 16 14:14:42 2005 From: dunk_palmer ayahoo.co(Duncan Palmer) Date: Wed Apr 18 17:37:45 2007 Subject: Trying to modify thsourccode In-Reply-To: 6667 Message-ID: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Hi Erwann, You can'usanything from the C standard library in thkernel. Your besbet may be to implement a pseudo randonumber generator yourself - you can gea true randoseed for ifrom the in kernel random driver (sedrivers/char/random.c). If you'rnew akernel programming, I'd suggest you hava look a'Linux Device Drivers' published by O'Reilly - there's aonlinversion somewhere, and also atharticles on LWN about porting drivers to 2.6 - http://lwn.net/Articles/driver-porting/ Dunk --- ErwanThoraval <erwann.thoraval@xxxxxxxx> wrote: > Hi, > > I try to modify thloss process in sch_netem.c to > approximatth > 2-states Gilbermodel. > > To do so, I need to ussrand() and drand() from > <stdlib.h>. This is th > firstimfor me to modify a module and I fail. > > I add #includ<stdlib.h> athe beginning of > sch_netem.c. I try to > compilwith: > > /home/user$ mak-C /usr/src/linux SUBDIRS=$PWD > modules > > Thfil<stdlib.h> is not included in the kernel > tree. So I tried to > pu#includ"/usr/include/stdlib.h" instead. > However, <stdlib.h> needs > other .h i/usr/includwhich can't be found. > > Whais thmore convenient way to use .h files from > /usr/includinto > kernel modulsourccode ? > > > -- > ErwanThoraval > IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, > France > Tél: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 > 71 > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > http://lists.osdl.org/mailman/listinfo/netem > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has thbesspam protection around http://mail.yahoo.co Froshemminger aosdl.org Wed Mar 16 14:25:47 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: Trying to modify thsourccode In-Reply-To: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050316142547.52f7291a@xxxxxxxxxxxxxxxxx> OWed, 16 Mar 2005 14:14:42 -0800 (PST) DuncaPalmer <dunk_palmer@xxxxxxxxx> wrote: > Hi Erwann, > > You can'usanything from the C standard library in > thkernel. Your besbet may be to implement a pseudo > randonumber generator yourself - you can gea true > randoseed for ifrom the in kernel random driver > (sedrivers/char/random.c). Usnet_random() igives a 32bit unsigned long. > If you'rnew akernel programming, I'd suggest you > hava look a'Linux Device Drivers' published by > O'Reilly - there's aonlinversion somewhere, and > also atharticles on LWN about porting drivers to > 2.6 - http://lwn.net/Articles/driver-porting/ Theris a new version: http://lwn.net/kernel/LDD3 Froshemminger aosdl.org Wed Mar 16 14:30:20 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: neteand latency variation In-Reply-To: <20050314201603.17994.qmail@xxxxxxxxxxxxxxxxxxxxxxx> References: <20050314201603.17994.qmail@xxxxxxxxxxxxxxxxxxxxxxx> Message-ID: <20050316143020.10e6e8de@xxxxxxxxxxxxxxxxx> OMon, 14 Mar 2005 12:16:03 -0800 (PST) js si <nistnet_user@xxxxxxxxx> wrote: > i havtwo interfaces ( A & B ) connected back to > back. oon(interface A) i configure netem to add a > 1ms delay. whei ping froA to B then i get a delay > of 1ms (with a variatioof .03 ms) buwhen i ping > froB to A i gea vairation of 0.25 ms . what > happens is thathlatency starts from 1ms, goes down > to 0.1 ms and thecycles back to 1ms and so on. i am > noablto understand the reason for this behaivior. > is ith1ms resolution of the clock? Probably, ondevicis using NAPI and the other is not. NAPI drivers scalbetter bucan have greaer latency. You could check by jususing a fifo and see if nothing happens. > or is ijust > coincidencthathe variation in one direction is > less thathother? > > thanks. > > > > __________________________________ > Do you Yahoo!? > Yahoo! Small Business - Try our new resources site! > http://smallbusiness.yahoo.com/resources/ Froerwann.thoraval airisa.fr Thu Mar 17 02:05:38 2005 From: erwann.thoraval airisa.fr (Erwann Thoraval) Date: Wed Apr 18 17:37:45 2007 Subject: Trying to modify thsourccode In-Reply-To: <20050316142547.52f7291a@xxxxxxxxxxxxxxxxx> References: <20050316221442.42123.qmail@xxxxxxxxxxxxxxxxxxxxxxx> <20050316142547.52f7291a@xxxxxxxxxxxxxxxxx> Message-ID: <42395672.2080304@xxxxxxxx> StepheHemminger wrote: > Theris a new version: > http://lwn.net/kernel/LDD3 Actually, iis http://lwn.net/Kernel/LDD3 -- ErwanThoraval IRISA-INRIA, Campus dBeaulieu, 35042 Rennes cedex, France T?l: +33 (0) 2 99 84 71 00, Fax: +33 (0) 2 99 84 71 71 Frorajahammad agmail.com Thu Mar 17 23:34:32 2005 From: rajahammad agmail.co(Raja Muhammad Hammad) Date: Wed Apr 18 17:37:45 2007 Subject: No Packeloss for RTP ??? Message-ID: <3ddf6cd3050317233466f7e39e@xxxxxxxxxxxxxx> hi I ausing video streaming server and clien(LiveMedia). I have two machines to testhvideo transmission. I used to netem to emulate WAothe client side but it does not work for RTP/UDP video transmission. e.g. # tc qdisc add dev eth0 rooneteloss 10% EveI puthe value 25%, I get no packet loss. If I puhigh valulike 30% and above I get no transmission at all. 10% and abovpretty hugvalues so I do not understand why I dont get any packeloss for RTP/UDP. WheI usping to test the trasnmission, I get packet loss as expected. I ausing Fedora cor2 with kernel 2.6-10 on both machines. Any possiblreasons...? Regards Hammad Frostefanie.wenk arbt-nbg.de Mon Mar 21 04:23:35 2005 From: stefanie.wenk arbt-nbg.d(Stefanie Wenk) Date: Wed Apr 18 17:37:45 2007 Subject: Problewith replace Message-ID: <423EBCC7.6030201@xxxxxxxxxx> Hello everybody! I aworking on a gui for neteand I'm already quite finished with it. Heramy work, netem is used for testing different network scenarios. This demands e.g. a changof thinterface settings in the middle of a running tesvia thgui. Until now I solved thprobleof changing by deleting the qdisc with "tc qdisc del dev DEV root" and adding inewly. Buunfortunaly this means thathinterface is "free" for a short time, and produces a peak iour measures. I tried to solvthis by using "replace", but this doesn'work if ratcontrol is desired. I realised thratcontrol the same way as shown in the netem examples othhomepage with tbf. The error is "invalid argument". Why does replacnowork here? Is there an alternative solution? I would bvery happy for somadvice and some help! Greetings Steffi Froshemminger aosdl.org Mon Mar 28 15:59:34 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:45 2007 Subject: Problewith replace In-Reply-To: <423EBCC7.6030201@xxxxxxxxxx> References: <423EBCC7.6030201@xxxxxxxxxx> Message-ID: <20050328155934.71f445cc@xxxxxxxxxxxxxxxxx> OMon, 21 Mar 2005 13:23:35 +0100 StefaniWenk <stefanie.wenk@xxxxxxxxxx> wrote: > Hello everybody! > I aworking on a gui for neteand I'm already quite finished with it. > Heramy work, netem is used for testing different network scenarios. > This demands e.g. a changof thinterface settings in the middle of a > running tesvia thgui. > Until now I solved thprobleof changing by deleting the qdisc with > "tc qdisc del dev DEV root" and adding inewly. Buunfortunaly this > means thathinterface is "free" for a short time, and produces a peak > iour measures. I tried to solvthis by using "replace", but this > doesn'work if ratcontrol is desired. > I realised thratcontrol the same way as shown in the netem examples > othhomepage with tbf. The error is "invalid argument". Why does > replacnowork here? Aryou replacing thTBF qdisc or the netem discipline. It probably is a bug ithtbf module that can be fixed. > Is theran alternativsolution? Longer term, I intend to makthqdisc into a full kobject and have thelivin /sys like the netdevices do. This would allow qdisc to exporattributes as files thacould be tweaked by things like your GUI. Buno codhas been done, and it has some issues about how to namand nesqdisc objects in the namespace. Currenthoughwas something like: /sys/class/net/eth0/qdisc would bsymlink to rooqdisc in /sys/class/qdisc/netem/1:0 /sys/class/qdisc/netem/1:0/delay would contaihook for delay /sys/class/qdisc/netem/1:0/10 would bsymlink to /sys/class/qdisc/tbf/ Still thinking and too busy to implemenit. > I would bvery happy for somadvice and some help! > Greetings > Steffi Fronistnet_user ayahoo.com Wed Mar 30 10:05:04 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:45 2007 Subject: netewith prio hangs on duplicate Message-ID: <20050330180504.75838.qmail@xxxxxxxxxxxxxxxxxxxxxxx> hi i tried thexamplgiven on the examples page to duplicatselected traffic like tc qdisc add dev eth0 roohandl1: prio tc qdisc add dev eth0 paren1:3 handl3: netem duplicat40% tc filter add dev eth0 protocol ip paren1:0 prio 3 u32 match ip ds11.0.2.2 flowid 1:3 whei ping fro11.0.2.2 to this interface my machine hangs. thsamthing works for drop or delay. i would appreciatif someoncan tell me what i am doing wrong. thanks. __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/
