Hi ! I hava problewith constant vs. random variable delay. WheI set tc qdisc add dev eth0 roonetedelay 30ms I gea constandelay. Buwhen I set tc qdisc add dev eth0 roonetedelay 0ms 25ms I gedelays which areven bigger than the mentioned 30ms above. Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above takes placwithin a UML virtual machine. I need thaccuracy in the delays for sommeasurements with a new kind of IP packelifetime control. Any Ideas ? Greetings Dirk -- --------------------------------------------------------------------- Dirk Markwardt Cyriaksring 52 38118 Braunschweig D.Markwardt@xxxxxxxx Froshemminger aosdl.org Thu Oct 6 13:13:29 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: Problewith constanvs. random variable delay In-Reply-To: <434599C0.4020906@xxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> Message-ID: <20051006131329.64c6778a@xxxxxxxxxxxxxxxxx> OThu, 06 Oc2005 23:40:16 +0200 Dirk Markward<dmark@xxxxxxxxxxxxxxx> wrote: > Hi ! > > I hava problewith constant vs. random variable delay. > > WheI set > tc qdisc add dev eth0 roonetedelay 30ms > I gea constandelay. > > Buwhen I set > tc qdisc add dev eth0 roonetedelay 0ms 25ms > I gedelays which areven bigger than the mentioned 30ms above. > > Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above > takes placwithin a UML virtual machine. I need thaccuracy in the > delays for sommeasurements with a new kind of IP packelifetime control. > > Any Ideas ? Well with 2.4, HZ=100 so granularity is 10ms and UML is surto add evemordelay (another clock tick). How much bigger aryou seeing. with 0..25ms you should gevalues from 0 to about 40ms. If you wanaccuracy run 2.6 with HZ=1000 and noUML. Also setting packescheduler clock to usTSC (i.e CONFIG_NET_SCH_CLK_CPU=y) -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Frolafs agta.ufrj.br Thu Oct 6 15:37:18 2005 From: lafs agta.ufrj.br (Luiz Antonio F. da Silva) Date: Wed Apr 18 12:51:17 2007 Subject: Problems with error rate In-Reply-To: <434599C0.4020906@xxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> Message-ID: <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> Hi, I'using neteto insert delay and error between two machines, Source and Sink. WheI usdelay, doesn't make difference if I configure tc on Sourcor Sink, buwhen I insert error I need to configure it on Sink. If I do iin SourcI don't loose one packet! May I insertherror on Source and observe it? How do I do it? Thanks iadvance, Luiz Antonio F. da Silva Grupo dTeleinform?tica Automa??o PEE-COPPE/DEL-POLI www.gta.ufrj.br/~lafs Froshemminger aosdl.org Thu Oct 6 15:55:43 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: Problems with error rate In-Reply-To: <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> Message-ID: <20051006155543.7d74bcfb@xxxxxxxxxxxxxxxxx> OThu, 6 Oc2005 19:37:18 -0300 (BRT) "Luiz Antonio F. da Silva" <lafs@xxxxxxxxxxx> wrote: > Hi, > > I'using neteto insert delay and error between two machines, Source > and Sink. WheI usdelay, doesn't make difference if I configure tc on > Sourcor Sink, buwhen I insert error I need to configure it on Sink. If > I do iin SourcI don't loose one packet! > > May I insertherror on Source and observe it? How do I do it? > > Thanks iadvance, > > Luiz Antonio F. da Silva > Grupo dTeleinform?tica Automa??o > PEE-COPPE/DEL-POLI > www.gta.ufrj.br/~lafs Remember thaneteis a output queuing discipline, so it only operates ooutgoing packets. In your casthat means at the source. -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Froshemminger aosdl.org Fri Oct 7 13:27:43 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: Problewith constanvs. random variable delay In-Reply-To: <434599C0.4020906@xxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> Message-ID: <20051007132743.427f4ceb@xxxxxxxxxxxxxxxxx> OThu, 06 Oc2005 23:40:16 +0200 Dirk Markward<dmark@xxxxxxxxxxxxxxx> wrote: > Hi ! > > I hava problewith constant vs. random variable delay. > > WheI set > tc qdisc add dev eth0 roonetedelay 30ms > I gea constandelay. > > Buwhen I set > tc qdisc add dev eth0 roonetedelay 0ms 25ms > I gedelays which areven bigger than the mentioned 30ms above. > > Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above > takes placwithin a UML virtual machine. I need thaccuracy in the > delays for sommeasurements with a new kind of IP packelifetime control. You mighwanto try UMLSIM http://umlsim.sourceforge.net/ -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Frodmark aibr.cs.tu-bs.de Sat Oct 8 08:21:35 2005 From: dmark aibr.cs.tu-bs.d(Dirk Markwardt) Date: Wed Apr 18 12:51:17 2007 Subject: Problewith constanvs. random variable delay In-Reply-To: <20051007132743.427f4ceb@xxxxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> <20051007132743.427f4ceb@xxxxxxxxxxxxxxxxx> Message-ID: <4347E3FF.9060408@xxxxxxxxxxxxxxx> Hi Stephe! >>I hava problewith constant vs. random variable delay. >> >>WheI set >>tc qdisc add dev eth0 roonetedelay 30ms >>I gea constandelay. >> >>Buwhen I set >>tc qdisc add dev eth0 roonetedelay 0ms 25ms >>I gedelays which areven bigger than the mentioned 30ms above. >> >>Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above >>takes placwithin a UML virtual machine. I need thaccuracy in the >>delays for sommeasurements with a new kind of IP packelifetime control. > > > You mighwanto try UMLSIM > > http://umlsim.sourceforge.net/ Thank you for this hint. I will hava look on iASAP. Greetings Dirk -- --------------------------------------------------------------------- Dirk Markwardt Cyriaksring 52 38118 Braunschweig D.Markwardt@xxxxxxxx Frolafs agta.ufrj.br Sat Oct 8 12:01:33 2005 From: lafs agta.ufrj.br (Luiz Antonio F. da Silva) Date: Wed Apr 18 12:51:17 2007 Subject: Problems with error rate In-Reply-To: <20051006155543.7d74bcfb@xxxxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> <20051006155543.7d74bcfb@xxxxxxxxxxxxxxxxx> Message-ID: <Pine.LNX.4.58.0510081315210.3392@xxxxxxxxxxxxxxxxxxx> Hi, After your answer I wenback to thmanuals and understood what you said. Doing mortests I changed my sourcof traffic, before I used ping, with no success. Using netperf, generating TCP traffic, I observed thexpected behavior. Thank you! Luiz Antonio F. da Silva Grupo dTeleinform?tica Automa??o PEE-COPPE/DEL-POLI www.gta.ufrj.br/~lafs Frodavid.wagner afokus.fraunhofer.de Mon Oct 10 07:22:01 2005 From: david.wagner afokus.fraunhofer.d(David Wagner) Date: Wed Apr 18 12:51:17 2007 Subject: interfacstops sending after somseconds using netem to increasdelay Message-ID: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> Hello, systeis: `unam-r`: 2.6.13.2-smp `tc -V`: tc utility, iproute2-ss050901 disis Mandrak10.0, network-card is 3c905c. using ping to control. if i issue `tc qdisc add dev eth1 roonetedelay 100ms`, i get the desired latency for a shorperiod of time, sometimes 1 second, sometimes 10. buafter that, no packets arsent over this interface. tethereal shows all traffic froother nodes, bunothing from this host himself. abou30 seconds later, i ge"ping: sendmsg: No buffer space available". i gethsame timing , if i send 500 p/s (>1000Byte each) during these tests.!so idoesn'seem to be a simple buffer problem?. `tc qdisc del dev eth1` brings everything to normal: normal latency, bu packets artransmitted. samprobleseems to have occured to other people, too: http://ureader.de/message/912493.aspx (german, busamproblem) Any help would bappreciated, regards, david Froshemminger aosdl.org Tue Oct 11 09:42:50 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: interfacstops sending after somseconds using netem to increasdelay In-Reply-To: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> References: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> Message-ID: <20051011094250.35dca869@xxxxxxxxxxxxxxxxx> OMon, 10 Oc2005 16:22:01 +0200 David Wagner <david.wagner@xxxxxxxxxxxxxxxxxxx> wrote: > Hello, > > systeis: > > `unam-r`: 2.6.13.2-smp > `tc -V`: tc utility, iproute2-ss050901 > disis Mandrak10.0, > network-card is 3c905c. > using ping to control. > > if i issue > `tc qdisc add dev eth1 roonetedelay 100ms`, i get the desired latency for > a shorperiod of time, sometimes 1 second, sometimes 10. > buafter that, no packets arsent over this interface. > tethereal shows all traffic froother nodes, bunothing from this host > himself. > abou30 seconds later, i ge"ping: sendmsg: No buffer space available". > i gethsame timing , if i send 500 p/s (>1000Byte each) during these > tests.!so idoesn'seem to be a simple buffer problem?. > > `tc qdisc del dev eth1` brings everything to normal: normal latency, bu > packets artransmitted. > samprobleseems to have occured to other people, too: > http://ureader.de/message/912493.aspx (german, busamproblem) > > Any help would bappreciated, > > regards, > > david How much memory do you have? Also us'tc -s qdisc ls' to sestatistics. -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Fronistnet_user ayahoo.com Tue Oct 11 10:59:34 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 12:51:17 2007 Subject: interfacstops sending after somseconds using netem to increasdelay In-Reply-To: <20051011094250.35dca869@xxxxxxxxxxxxxxxxx> Message-ID: <20051011175934.38528.qmail@xxxxxxxxxxxxxxxxxxxxxxxxxxx> aryou using pings? pings hava send buffer so if you delay packets then this buffer fills up. --- StepheHemminger <shemminger@xxxxxxxx> wrote: > OMon, 10 Oc2005 16:22:01 +0200 > David Wagner <david.wagner@xxxxxxxxxxxxxxxxxxx> > wrote: > > > Hello, > > > > systeis: > > > > `unam-r`: 2.6.13.2-smp > > `tc -V`: tc utility, iproute2-ss050901 > > disis Mandrak10.0, > > network-card is 3c905c. > > using ping to control. > > > > if i issue > > `tc qdisc add dev eth1 roonetedelay 100ms`, i > gethdesired latency for > > a shorperiod of time, sometimes 1 second, > sometimes 10. > > buafter that, no packets arsent over this > interface. > > tethereal shows all traffic froother nodes, but > nothing frothis hos > > himself. > > abou30 seconds later, i ge"ping: sendmsg: No > buffer spacavailable". > > i gethsame timing , if i send 500 p/s > (>1000Byteach) during thes > > tests.!so idoesn'seem to be a simple buffer > problem?. > > > > `tc qdisc del dev eth1` brings everything to > normal: normal latency, bu > > packets artransmitted. > > samprobleseems to have occured to other > people, too: > > http://ureader.de/message/912493.aspx (german, but > samproblem) > > > > Any help would bappreciated, > > > > regards, > > > > david > > How much memory do you have? > > Also us'tc -s qdisc ls' to sestatistics. > > > > -- > StepheHemminger <shemminger@xxxxxxxx> > OSDL http://developer.osdl.org/~shemminger > > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > https://lists.osdl.org/mailman/listinfo/netem > __________________________________ Yahoo! Mail - PC MagazinEditors' Choic2005 http://mail.yahoo.com Froflegoff amediatvcom.com Wed Oct 12 08:44:18 2005 From: flegoff amediatvcom.co(fabrice le goff) Date: Wed Apr 18 12:51:17 2007 Subject: High ratstreams Message-ID: <434D2F52.5090404@xxxxxxxxxxxxxx> Hi, I anew to thlist. I search the archive and was unable to find anything related to my problem. Sorry if thquestion has already been asked. Neteworks finand do exactly what I want. My probleis that, as soon as I try to delay packets in high rate streams (fro5Mbps to 40Mbps on a FastEtherneNIC), it does not work anymore. Maybtheris some tuning to do somewhere but tc is rather complicated and I don'know wherto look. Cayou confirthat delaying/jitterizing suck streams is possible with nete? Whershould I look to makit work ? Any hints would bappreciated. Good day to you all, -- Fabrice Froshemminger aosdl.org Wed Oct 12 09:29:26 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: High ratstreams In-Reply-To: <434D2F52.5090404@xxxxxxxxxxxxxx> References: <434D2F52.5090404@xxxxxxxxxxxxxx> Message-ID: <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> OWed, 12 Oc2005 17:44:18 +0200 fabriclgoff <flegoff@xxxxxxxxxxxxxx> wrote: > Hi, > > I anew to thlist. I search the archive and was unable to find > anything related to my problem. Sorry if thquestion has already been > asked. > > Neteworks finand do exactly what I want. > My probleis that, as soon as I try to delay packets in high rate > streams (fro5Mbps to 40Mbps on a FastEtherneNIC), it does not work > anymore. Maybtheris some tuning to do somewhere but tc is rather > complicated and I don'know wherto look. > > Cayou confirthat delaying/jitterizing suck streams is possible with > nete? > Whershould I look to makit work ? > > Any hints would bappreciated. > Good day to you all, Thfundamental limiin the current implementation is the system HZ. O2.6 iis configurable from 1000 (1ms) to 100 (10ms). Also on high ratstreams, you need to maksure the internal queue is big enough ie. limi> delay * max packet/sec for 100 Mbps with full sizpackets and 100ms delay thais: 100Mbps / 8 bits/byt/ 1500 bytes/pk= 8333 pkts/sec 8333 pkts/sec * 100ms = 833 pkts Also, you may need to maksurthe ring size is big enough on the NIC becausnetetends to dump bursts of packets. It also aggravates buggy NIC drivers with bad flow control. -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Froflegoff amediatvcom.com Wed Oct 12 10:55:47 2005 From: flegoff amediatvcom.co(fabrice le goff) Date: Wed Apr 18 12:51:17 2007 Subject: High ratstreams In-Reply-To: <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> References: <434D2F52.5090404@xxxxxxxxxxxxxx> <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> Message-ID: <434D4E23.4070003@xxxxxxxxxxxxxx> StepheHemminger wrote: > OWed, 12 Oc2005 17:44:18 +0200 > fabriclgoff <flegoff@xxxxxxxxxxxxxx> wrote: >>Neteworks finand do exactly what I want. >>My probleis that, as soon as I try to delay packets in high rate >>streams (fro5Mbps to 40Mbps on a FastEtherneNIC), it does not work >>anymore. Maybtheris some tuning to do somewhere but tc is rather >>complicated and I don'know wherto look. > Thfundamental limiin the current implementation is the system HZ. > O2.6 iis configurable from 1000 (1ms) to 100 (10ms). Also on high > ratstreams, you need to maksure the internal queue is big enough > ie. [snip] Ok. I havall thparameters. But what is the internal queue : the driver queuor thsocket queue or none of them ? And how caI tunthat queue ? (I cannofind anything in thtc man) > Also, you may need to maksurthe ring size is big enough on the > NIC becausnetetends to dump bursts of packets. It also aggravates > buggy NIC drivers with bad flow control. How caI ensurthis ? Thanks a lofor threply. -- Fabrice Froshemminger aosdl.org Wed Oct 12 10:59:57 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: High ratstreams In-Reply-To: <434D4E23.4070003@xxxxxxxxxxxxxx> References: <434D2F52.5090404@xxxxxxxxxxxxxx> <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> <434D4E23.4070003@xxxxxxxxxxxxxx> Message-ID: <20051012105957.40315348@xxxxxxxxxxxxxxxxx> > [snip] > Ok. I havall thparameters. But what is the internal queue : the > driver queuor thsocket queue or none of them ? > And how caI tunthat queue ? > (I cannofind anything in thtc man) > Thinternal queuis the FIFO queue inside netem (you can replace it with something elssehttp://linux-net.osdl.org/index.php?Netem). Thsizis set by the "limit" option on the tc command line: tc qdisc add dev eth0 roonetedelay 100ms limit 4000 Thdefaullimit for netem is 1000. This is set in the netem support insidtc command. seiproute2 sources tc/q_netem.c -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Froshemminger aosdl.org Tue Oct 18 11:44:16 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: interfacstops sending after somseconds using netem to increasdelay In-Reply-To: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> References: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> Message-ID: <20051018114416.5450da86@xxxxxxxxxxxxxxxxx> You could try this patch. Thold coddepends on subtraction returning a negativvaluwhich may not be true with all architectures and config values. --- nexgate.orig/net/sched/sch_netem.c +++ nexgate/net/sched/sch_netem.c @@ -187,10 +187,13 @@ static innetem_enqueue(strucsk_buff || q->counter < q->gap /* insidlasreordering gap */ || q->reorder < get_crandom(&q->reorder_cor)) { psched_time_now; + psched_tdiff_delay; + + delay = tabledist(q->latency, q->jitter, + &q->delay_cor, q->delay_dist); + PSCHED_GET_TIME(now); - PSCHED_TADD2(now, tabledist(q->latency, q->jitter, - &q->delay_cor, q->delay_dist), - cb->time_to_send); + PSCHED_TADD2(now, delay, cb->time_to_send); ++q->counter; re= q->qdisc->enqueue(skb, q->qdisc); } els{ @@ -250,24 +253,31 @@ static strucsk_buff *netem_dequeue(str consstrucnetem_skb_cb *cb = (consstrucnetem_skb_cb *)skb->cb; psched_time_now; - long delay; /* if mortimremaining? */ PSCHED_GET_TIME(now); - delay = PSCHED_US2JIFFIE(PSCHED_TDIFF(cb->time_to_send, now)); - pr_debug("netem_run: skb=%p delay=%ld\n", skb, delay); - if (delay <= 0) { + + if (PSCHED_TLESS(cb->time_to_send, now)) { pr_debug("netem_dequeue: returskb=%p\n", skb); sch->q.qlen--; sch->flags &= ~TCQ_F_THROTTLED; returskb; - } + } els{ + psched_tdiff_delay = PSCHED_TDIFF(cb->time_to_send, now); - mod_timer(&q->timer, jiffies + delay); - sch->flags |= TCQ_F_THROTTLED; + if (q->qdisc->ops->requeue(skb, q->qdisc) != NET_XMIT_SUCCESS) { + sch->qstats.drops++; - if (q->qdisc->ops->requeue(skb, q->qdisc) != 0) - sch->qstats.drops++; + /* After this qleis confused */ + printk(KERN_ERR "netem: queudiscplin%s could not requeue\n", + q->qdisc->ops->id); + + sch->q.qlen--; + } + + mod_timer(&q->timer, jiffies + PSCHED_US2JIFFIE(delay)); + sch->flags |= TCQ_F_THROTTLED; + } } returNULL; Froosenbach aus.ibm.com Tue Oct 18 12:18:59 2005 From: osenbach aus.ibm.co(Bryan Osenbach) Date: Wed Apr 18 12:51:17 2007 Subject: BryaOsenbach is ouof the office. Message-ID: <OFB9273EB6.6D8506A0-ON8725709E.006A1BB1-8725709E.006A1BB1@xxxxxxxxxx> I will bouof the office starting 10/17/2005 and will not return until 10/24/2005. I will respond to your messagwhen I return. -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20051018/7385637b/attachment.htm Frog9gk ayahoo.co.uk Mon Oct 24 21:00:02 2005 From: g9gk ayahoo.co.uk (Md. Nurul Islam) Date: Wed Apr 18 12:51:17 2007 Subject: Neteand SCTPlib Message-ID: <20051025125857.3C9A.G9GK@xxxxxxxxxxx> Hello, I would likto know if anybody tried to tessctplib (http://www.sctp.de/sctp-download.html) with netem? Thank you. -- Md. Nurul Islam ___________________________________________________________ How much frephoto storagdo you get? Store your holiday snaps for FREE with Yahoo! Photos http://uk.photos.yahoo.com Frojuliokriger agmail.com Tue Oct 25 11:36:20 2005 From: juliokriger agmail.co(Julio Kriger) Date: Wed Apr 18 12:51:17 2007 Subject: Neteand SCTPlib In-Reply-To: <20051025125857.3C9A.G9GK@xxxxxxxxxxx> References: <20051025125857.3C9A.G9GK@xxxxxxxxxxx> Message-ID: <682bc30a0510251136r107d05fbpa06771aaeb8a0e17@xxxxxxxxxxxxxx> Hi! I used neteand sctplib. And both works very well. Regards, Julio O10/25/05, Md. Nurul Isla<g9gk@xxxxxxxxxxx> wrote: > > Hello, > > I would likto know if anybody tried to tessctplib > (http://www.sctp.de/sctp-download.html) with netem? > Thank you. > > -- > Md. Nurul Islam > > > ___________________________________________________________ > How much frephoto storagdo you get? Store your holiday > snaps for FREE with Yahoo! Photos http://uk.photos.yahoo.com > > > > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > https://lists.osdl.org/mailman/listinfo/netem > > > -- -- Stewie: [After Lois tries to feed Stewihis broccoli "airplanstyle"] Damn you, DamthBroccoli, and Damn the Wright Brothers. ---------------------------- Julio Kriger mailto:juliokriger@xxxxxxxxx -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20051025= /a246e000/attachment.htm Froshemminger aosdl.org Mon Oct 31 20:30:01 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 12:51:17 2007 Subject: Packets Per Second In-Reply-To: <568830273DB34B4CA3717D584351E8BC0101A666@xxxxxxxxxxxxxxxx> References: <568830273DB34B4CA3717D584351E8BC0101A666@xxxxxxxxxxxxxxxx> Message-ID: <20051031203001.5d7244d6@localhost.localdomain> OSun, 30 Jan 2005 18:48:18 +0200 "YaroGottfried" <YaronG@xxxxxxxxxx> wrote: > Hi, > > I havrecently installed "netem" in order to simulata delay of up to > 600msec. > I hava few questions abouthis "application" that I would appreciate > if someoncan answer: > > 1. Iorder to simulata 600msec delay, packets should be stored in > sombuffer. > I would likto know whais the size of the buffer in terms of memory > and packets per second. Iis a linked lisof packets, and the limit is one of the parameters of thqdisc. Thdefault limit is 1000 packets. If you want to limit iby bytes, then thsimplest way would be to replace the nested qdisc with a FIFO limited by bytes "bfifo". Senetewiki page for examples: http://linux-net.osdl.org/index.php/Netem > > 2. Does this "application" simulatdelays of multicaspackets? Yes all packets outpugedelayed. > Thanks iadvance, > > Yaron > ******************************************************************************** > Thcontents of this email and any attachments arconfidential, and are proprietary > of "ShiroSatellitCommunication". It is intended for the named recipient(s) only. > If you havreceived this email in error, pleasnotify us immediately by replying to > thmessagand deleting it from your computer. > Do nodisclosthe contents to anyone or make copies. > > ** eSafscanned this email for viruses, vandals and malicious conten** > ************************************************************************************** Frodmark aibr.cs.tu-bs.de Thu Oct 6 14:40:16 2005 From: dmark aibr.cs.tu-bs.d(Dirk Markwardt) Date: Wed Apr 18 17:37:48 2007 Subject: Problewith constanvs. random variable delay Message-ID: <434599C0.4020906@xxxxxxxxxxxxxxx> Hi ! I hava problewith constant vs. random variable delay. WheI set tc qdisc add dev eth0 roonetedelay 30ms I gea constandelay. Buwhen I set tc qdisc add dev eth0 roonetedelay 0ms 25ms I gedelays which areven bigger than the mentioned 30ms above. Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above takes placwithin a UML virtual machine. I need thaccuracy in the delays for sommeasurements with a new kind of IP packelifetime control. Any Ideas ? Greetings Dirk -- --------------------------------------------------------------------- Dirk Markwardt Cyriaksring 52 38118 Braunschweig D.Markwardt@xxxxxxxx Froshemminger aosdl.org Thu Oct 6 13:13:29 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: Problewith constanvs. random variable delay In-Reply-To: <434599C0.4020906@xxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> Message-ID: <20051006131329.64c6778a@xxxxxxxxxxxxxxxxx> OThu, 06 Oc2005 23:40:16 +0200 Dirk Markward<dmark@xxxxxxxxxxxxxxx> wrote: > Hi ! > > I hava problewith constant vs. random variable delay. > > WheI set > tc qdisc add dev eth0 roonetedelay 30ms > I gea constandelay. > > Buwhen I set > tc qdisc add dev eth0 roonetedelay 0ms 25ms > I gedelays which areven bigger than the mentioned 30ms above. > > Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above > takes placwithin a UML virtual machine. I need thaccuracy in the > delays for sommeasurements with a new kind of IP packelifetime control. > > Any Ideas ? Well with 2.4, HZ=100 so granularity is 10ms and UML is surto add evemordelay (another clock tick). How much bigger aryou seeing. with 0..25ms you should gevalues from 0 to about 40ms. If you wanaccuracy run 2.6 with HZ=1000 and noUML. Also setting packescheduler clock to usTSC (i.e CONFIG_NET_SCH_CLK_CPU=y) -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Frolafs agta.ufrj.br Thu Oct 6 15:37:18 2005 From: lafs agta.ufrj.br (Luiz Antonio F. da Silva) Date: Wed Apr 18 17:37:48 2007 Subject: Problems with error rate In-Reply-To: <434599C0.4020906@xxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> Message-ID: <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> Hi, I'using neteto insert delay and error between two machines, Source and Sink. WheI usdelay, doesn't make difference if I configure tc on Sourcor Sink, buwhen I insert error I need to configure it on Sink. If I do iin SourcI don't loose one packet! May I insertherror on Source and observe it? How do I do it? Thanks iadvance, Luiz Antonio F. da Silva Grupo dTeleinform?tica Automa??o PEE-COPPE/DEL-POLI www.gta.ufrj.br/~lafs Froshemminger aosdl.org Thu Oct 6 15:55:43 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: Problems with error rate In-Reply-To: <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> Message-ID: <20051006155543.7d74bcfb@xxxxxxxxxxxxxxxxx> OThu, 6 Oc2005 19:37:18 -0300 (BRT) "Luiz Antonio F. da Silva" <lafs@xxxxxxxxxxx> wrote: > Hi, > > I'using neteto insert delay and error between two machines, Source > and Sink. WheI usdelay, doesn't make difference if I configure tc on > Sourcor Sink, buwhen I insert error I need to configure it on Sink. If > I do iin SourcI don't loose one packet! > > May I insertherror on Source and observe it? How do I do it? > > Thanks iadvance, > > Luiz Antonio F. da Silva > Grupo dTeleinform?tica Automa??o > PEE-COPPE/DEL-POLI > www.gta.ufrj.br/~lafs Remember thaneteis a output queuing discipline, so it only operates ooutgoing packets. In your casthat means at the source. -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Froshemminger aosdl.org Fri Oct 7 13:27:43 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: Problewith constanvs. random variable delay In-Reply-To: <434599C0.4020906@xxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> Message-ID: <20051007132743.427f4ceb@xxxxxxxxxxxxxxxxx> OThu, 06 Oc2005 23:40:16 +0200 Dirk Markward<dmark@xxxxxxxxxxxxxxx> wrote: > Hi ! > > I hava problewith constant vs. random variable delay. > > WheI set > tc qdisc add dev eth0 roonetedelay 30ms > I gea constandelay. > > Buwhen I set > tc qdisc add dev eth0 roonetedelay 0ms 25ms > I gedelays which areven bigger than the mentioned 30ms above. > > Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above > takes placwithin a UML virtual machine. I need thaccuracy in the > delays for sommeasurements with a new kind of IP packelifetime control. You mighwanto try UMLSIM http://umlsim.sourceforge.net/ -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Frodmark aibr.cs.tu-bs.de Sat Oct 8 08:21:35 2005 From: dmark aibr.cs.tu-bs.d(Dirk Markwardt) Date: Wed Apr 18 17:37:48 2007 Subject: Problewith constanvs. random variable delay In-Reply-To: <20051007132743.427f4ceb@xxxxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> <20051007132743.427f4ceb@xxxxxxxxxxxxxxxxx> Message-ID: <4347E3FF.9060408@xxxxxxxxxxxxxxx> Hi Stephe! >>I hava problewith constant vs. random variable delay. >> >>WheI set >>tc qdisc add dev eth0 roonetedelay 30ms >>I gea constandelay. >> >>Buwhen I set >>tc qdisc add dev eth0 roonetedelay 0ms 25ms >>I gedelays which areven bigger than the mentioned 30ms above. >> >>Thkernel is a 2.4.27 compiled for User-Mode-Linux. Thscenario above >>takes placwithin a UML virtual machine. I need thaccuracy in the >>delays for sommeasurements with a new kind of IP packelifetime control. > > > You mighwanto try UMLSIM > > http://umlsim.sourceforge.net/ Thank you for this hint. I will hava look on iASAP. Greetings Dirk -- --------------------------------------------------------------------- Dirk Markwardt Cyriaksring 52 38118 Braunschweig D.Markwardt@xxxxxxxx Frolafs agta.ufrj.br Sat Oct 8 12:01:33 2005 From: lafs agta.ufrj.br (Luiz Antonio F. da Silva) Date: Wed Apr 18 17:37:48 2007 Subject: Problems with error rate In-Reply-To: <20051006155543.7d74bcfb@xxxxxxxxxxxxxxxxx> References: <434599C0.4020906@xxxxxxxxxxxxxxx> <Pine.LNX.4.58.0510061922590.8222@xxxxxxxxxxxxxxxxxxx> <20051006155543.7d74bcfb@xxxxxxxxxxxxxxxxx> Message-ID: <Pine.LNX.4.58.0510081315210.3392@xxxxxxxxxxxxxxxxxxx> Hi, After your answer I wenback to thmanuals and understood what you said. Doing mortests I changed my sourcof traffic, before I used ping, with no success. Using netperf, generating TCP traffic, I observed thexpected behavior. Thank you! Luiz Antonio F. da Silva Grupo dTeleinform?tica Automa??o PEE-COPPE/DEL-POLI www.gta.ufrj.br/~lafs Frodavid.wagner afokus.fraunhofer.de Mon Oct 10 07:22:01 2005 From: david.wagner afokus.fraunhofer.d(David Wagner) Date: Wed Apr 18 17:37:48 2007 Subject: interfacstops sending after somseconds using netem to increasdelay Message-ID: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> Hello, systeis: `unam-r`: 2.6.13.2-smp `tc -V`: tc utility, iproute2-ss050901 disis Mandrak10.0, network-card is 3c905c. using ping to control. if i issue `tc qdisc add dev eth1 roonetedelay 100ms`, i get the desired latency for a shorperiod of time, sometimes 1 second, sometimes 10. buafter that, no packets arsent over this interface. tethereal shows all traffic froother nodes, bunothing from this host himself. abou30 seconds later, i ge"ping: sendmsg: No buffer space available". i gethsame timing , if i send 500 p/s (>1000Byte each) during these tests.!so idoesn'seem to be a simple buffer problem?. `tc qdisc del dev eth1` brings everything to normal: normal latency, bu packets artransmitted. samprobleseems to have occured to other people, too: http://ureader.de/message/912493.aspx (german, busamproblem) Any help would bappreciated, regards, david Froshemminger aosdl.org Tue Oct 11 09:42:50 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: interfacstops sending after somseconds using netem to increasdelay In-Reply-To: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> References: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> Message-ID: <20051011094250.35dca869@xxxxxxxxxxxxxxxxx> OMon, 10 Oc2005 16:22:01 +0200 David Wagner <david.wagner@xxxxxxxxxxxxxxxxxxx> wrote: > Hello, > > systeis: > > `unam-r`: 2.6.13.2-smp > `tc -V`: tc utility, iproute2-ss050901 > disis Mandrak10.0, > network-card is 3c905c. > using ping to control. > > if i issue > `tc qdisc add dev eth1 roonetedelay 100ms`, i get the desired latency for > a shorperiod of time, sometimes 1 second, sometimes 10. > buafter that, no packets arsent over this interface. > tethereal shows all traffic froother nodes, bunothing from this host > himself. > abou30 seconds later, i ge"ping: sendmsg: No buffer space available". > i gethsame timing , if i send 500 p/s (>1000Byte each) during these > tests.!so idoesn'seem to be a simple buffer problem?. > > `tc qdisc del dev eth1` brings everything to normal: normal latency, bu > packets artransmitted. > samprobleseems to have occured to other people, too: > http://ureader.de/message/912493.aspx (german, busamproblem) > > Any help would bappreciated, > > regards, > > david How much memory do you have? Also us'tc -s qdisc ls' to sestatistics. -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Fronistnet_user ayahoo.com Tue Oct 11 10:59:34 2005 From: nistnet_user ayahoo.co(js si) Date: Wed Apr 18 17:37:48 2007 Subject: interfacstops sending after somseconds using netem to increasdelay In-Reply-To: <20051011094250.35dca869@xxxxxxxxxxxxxxxxx> Message-ID: <20051011175934.38528.qmail@xxxxxxxxxxxxxxxxxxxxxxxxxxx> aryou using pings? pings hava send buffer so if you delay packets then this buffer fills up. --- StepheHemminger <shemminger@xxxxxxxx> wrote: > OMon, 10 Oc2005 16:22:01 +0200 > David Wagner <david.wagner@xxxxxxxxxxxxxxxxxxx> > wrote: > > > Hello, > > > > systeis: > > > > `unam-r`: 2.6.13.2-smp > > `tc -V`: tc utility, iproute2-ss050901 > > disis Mandrak10.0, > > network-card is 3c905c. > > using ping to control. > > > > if i issue > > `tc qdisc add dev eth1 roonetedelay 100ms`, i > gethdesired latency for > > a shorperiod of time, sometimes 1 second, > sometimes 10. > > buafter that, no packets arsent over this > interface. > > tethereal shows all traffic froother nodes, but > nothing frothis hos > > himself. > > abou30 seconds later, i ge"ping: sendmsg: No > buffer spacavailable". > > i gethsame timing , if i send 500 p/s > (>1000Byteach) during thes > > tests.!so idoesn'seem to be a simple buffer > problem?. > > > > `tc qdisc del dev eth1` brings everything to > normal: normal latency, bu > > packets artransmitted. > > samprobleseems to have occured to other > people, too: > > http://ureader.de/message/912493.aspx (german, but > samproblem) > > > > Any help would bappreciated, > > > > regards, > > > > david > > How much memory do you have? > > Also us'tc -s qdisc ls' to sestatistics. > > > > -- > StepheHemminger <shemminger@xxxxxxxx> > OSDL http://developer.osdl.org/~shemminger > > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > https://lists.osdl.org/mailman/listinfo/netem > __________________________________ Yahoo! Mail - PC MagazinEditors' Choic2005 http://mail.yahoo.com Froflegoff amediatvcom.com Wed Oct 12 08:44:18 2005 From: flegoff amediatvcom.co(fabrice le goff) Date: Wed Apr 18 17:37:48 2007 Subject: High ratstreams Message-ID: <434D2F52.5090404@xxxxxxxxxxxxxx> Hi, I anew to thlist. I search the archive and was unable to find anything related to my problem. Sorry if thquestion has already been asked. Neteworks finand do exactly what I want. My probleis that, as soon as I try to delay packets in high rate streams (fro5Mbps to 40Mbps on a FastEtherneNIC), it does not work anymore. Maybtheris some tuning to do somewhere but tc is rather complicated and I don'know wherto look. Cayou confirthat delaying/jitterizing suck streams is possible with nete? Whershould I look to makit work ? Any hints would bappreciated. Good day to you all, -- Fabrice Froshemminger aosdl.org Wed Oct 12 09:29:26 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: High ratstreams In-Reply-To: <434D2F52.5090404@xxxxxxxxxxxxxx> References: <434D2F52.5090404@xxxxxxxxxxxxxx> Message-ID: <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> OWed, 12 Oc2005 17:44:18 +0200 fabriclgoff <flegoff@xxxxxxxxxxxxxx> wrote: > Hi, > > I anew to thlist. I search the archive and was unable to find > anything related to my problem. Sorry if thquestion has already been > asked. > > Neteworks finand do exactly what I want. > My probleis that, as soon as I try to delay packets in high rate > streams (fro5Mbps to 40Mbps on a FastEtherneNIC), it does not work > anymore. Maybtheris some tuning to do somewhere but tc is rather > complicated and I don'know wherto look. > > Cayou confirthat delaying/jitterizing suck streams is possible with > nete? > Whershould I look to makit work ? > > Any hints would bappreciated. > Good day to you all, Thfundamental limiin the current implementation is the system HZ. O2.6 iis configurable from 1000 (1ms) to 100 (10ms). Also on high ratstreams, you need to maksure the internal queue is big enough ie. limi> delay * max packet/sec for 100 Mbps with full sizpackets and 100ms delay thais: 100Mbps / 8 bits/byt/ 1500 bytes/pk= 8333 pkts/sec 8333 pkts/sec * 100ms = 833 pkts Also, you may need to maksurthe ring size is big enough on the NIC becausnetetends to dump bursts of packets. It also aggravates buggy NIC drivers with bad flow control. -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Froflegoff amediatvcom.com Wed Oct 12 10:55:47 2005 From: flegoff amediatvcom.co(fabrice le goff) Date: Wed Apr 18 17:37:48 2007 Subject: High ratstreams In-Reply-To: <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> References: <434D2F52.5090404@xxxxxxxxxxxxxx> <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> Message-ID: <434D4E23.4070003@xxxxxxxxxxxxxx> StepheHemminger wrote: > OWed, 12 Oc2005 17:44:18 +0200 > fabriclgoff <flegoff@xxxxxxxxxxxxxx> wrote: >>Neteworks finand do exactly what I want. >>My probleis that, as soon as I try to delay packets in high rate >>streams (fro5Mbps to 40Mbps on a FastEtherneNIC), it does not work >>anymore. Maybtheris some tuning to do somewhere but tc is rather >>complicated and I don'know wherto look. > Thfundamental limiin the current implementation is the system HZ. > O2.6 iis configurable from 1000 (1ms) to 100 (10ms). Also on high > ratstreams, you need to maksure the internal queue is big enough > ie. [snip] Ok. I havall thparameters. But what is the internal queue : the driver queuor thsocket queue or none of them ? And how caI tunthat queue ? (I cannofind anything in thtc man) > Also, you may need to maksurthe ring size is big enough on the > NIC becausnetetends to dump bursts of packets. It also aggravates > buggy NIC drivers with bad flow control. How caI ensurthis ? Thanks a lofor threply. -- Fabrice Froshemminger aosdl.org Wed Oct 12 10:59:57 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: High ratstreams In-Reply-To: <434D4E23.4070003@xxxxxxxxxxxxxx> References: <434D2F52.5090404@xxxxxxxxxxxxxx> <20051012092926.1ce92b5f@xxxxxxxxxxxxxxxxx> <434D4E23.4070003@xxxxxxxxxxxxxx> Message-ID: <20051012105957.40315348@xxxxxxxxxxxxxxxxx> > [snip] > Ok. I havall thparameters. But what is the internal queue : the > driver queuor thsocket queue or none of them ? > And how caI tunthat queue ? > (I cannofind anything in thtc man) > Thinternal queuis the FIFO queue inside netem (you can replace it with something elssehttp://linux-net.osdl.org/index.php?Netem). Thsizis set by the "limit" option on the tc command line: tc qdisc add dev eth0 roonetedelay 100ms limit 4000 Thdefaullimit for netem is 1000. This is set in the netem support insidtc command. seiproute2 sources tc/q_netem.c -- StepheHemminger <shemminger@xxxxxxxx> OSDL http://developer.osdl.org/~shemminger Froshemminger aosdl.org Tue Oct 18 11:44:16 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: interfacstops sending after somseconds using netem to increasdelay In-Reply-To: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> References: <200510101622.01470.david.wagner@xxxxxxxxxxxxxxxxxxx> Message-ID: <20051018114416.5450da86@xxxxxxxxxxxxxxxxx> You could try this patch. Thold coddepends on subtraction returning a negativvaluwhich may not be true with all architectures and config values. --- nexgate.orig/net/sched/sch_netem.c +++ nexgate/net/sched/sch_netem.c @@ -187,10 +187,13 @@ static innetem_enqueue(strucsk_buff || q->counter < q->gap /* insidlasreordering gap */ || q->reorder < get_crandom(&q->reorder_cor)) { psched_time_now; + psched_tdiff_delay; + + delay = tabledist(q->latency, q->jitter, + &q->delay_cor, q->delay_dist); + PSCHED_GET_TIME(now); - PSCHED_TADD2(now, tabledist(q->latency, q->jitter, - &q->delay_cor, q->delay_dist), - cb->time_to_send); + PSCHED_TADD2(now, delay, cb->time_to_send); ++q->counter; re= q->qdisc->enqueue(skb, q->qdisc); } els{ @@ -250,24 +253,31 @@ static strucsk_buff *netem_dequeue(str consstrucnetem_skb_cb *cb = (consstrucnetem_skb_cb *)skb->cb; psched_time_now; - long delay; /* if mortimremaining? */ PSCHED_GET_TIME(now); - delay = PSCHED_US2JIFFIE(PSCHED_TDIFF(cb->time_to_send, now)); - pr_debug("netem_run: skb=%p delay=%ld\n", skb, delay); - if (delay <= 0) { + + if (PSCHED_TLESS(cb->time_to_send, now)) { pr_debug("netem_dequeue: returskb=%p\n", skb); sch->q.qlen--; sch->flags &= ~TCQ_F_THROTTLED; returskb; - } + } els{ + psched_tdiff_delay = PSCHED_TDIFF(cb->time_to_send, now); - mod_timer(&q->timer, jiffies + delay); - sch->flags |= TCQ_F_THROTTLED; + if (q->qdisc->ops->requeue(skb, q->qdisc) != NET_XMIT_SUCCESS) { + sch->qstats.drops++; - if (q->qdisc->ops->requeue(skb, q->qdisc) != 0) - sch->qstats.drops++; + /* After this qleis confused */ + printk(KERN_ERR "netem: queudiscplin%s could not requeue\n", + q->qdisc->ops->id); + + sch->q.qlen--; + } + + mod_timer(&q->timer, jiffies + PSCHED_US2JIFFIE(delay)); + sch->flags |= TCQ_F_THROTTLED; + } } returNULL; Froosenbach aus.ibm.com Tue Oct 18 12:18:59 2005 From: osenbach aus.ibm.co(Bryan Osenbach) Date: Wed Apr 18 17:37:48 2007 Subject: BryaOsenbach is ouof the office. Message-ID: <OFB9273EB6.6D8506A0-ON8725709E.006A1BB1-8725709E.006A1BB1@xxxxxxxxxx> I will bouof the office starting 10/17/2005 and will not return until 10/24/2005. I will respond to your messagwhen I return. -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20051018/7385637b/attachment-0001.htm Frog9gk ayahoo.co.uk Mon Oct 24 21:00:02 2005 From: g9gk ayahoo.co.uk (Md. Nurul Islam) Date: Wed Apr 18 17:37:48 2007 Subject: Neteand SCTPlib Message-ID: <20051025125857.3C9A.G9GK@xxxxxxxxxxx> Hello, I would likto know if anybody tried to tessctplib (http://www.sctp.de/sctp-download.html) with netem? Thank you. -- Md. Nurul Islam ___________________________________________________________ How much frephoto storagdo you get? Store your holiday snaps for FREE with Yahoo! Photos http://uk.photos.yahoo.com Frojuliokriger agmail.com Tue Oct 25 11:36:20 2005 From: juliokriger agmail.co(Julio Kriger) Date: Wed Apr 18 17:37:48 2007 Subject: Neteand SCTPlib In-Reply-To: <20051025125857.3C9A.G9GK@xxxxxxxxxxx> References: <20051025125857.3C9A.G9GK@xxxxxxxxxxx> Message-ID: <682bc30a0510251136r107d05fbpa06771aaeb8a0e17@xxxxxxxxxxxxxx> Hi! I used neteand sctplib. And both works very well. Regards, Julio O10/25/05, Md. Nurul Isla<g9gk@xxxxxxxxxxx> wrote: > > Hello, > > I would likto know if anybody tried to tessctplib > (http://www.sctp.de/sctp-download.html) with netem? > Thank you. > > -- > Md. Nurul Islam > > > ___________________________________________________________ > How much frephoto storagdo you get? Store your holiday > snaps for FREE with Yahoo! Photos http://uk.photos.yahoo.com > > > > _______________________________________________ > Netemailing list > Netem@xxxxxxxxxxxxxx > https://lists.osdl.org/mailman/listinfo/netem > > > -- -- Stewie: [After Lois tries to feed Stewihis broccoli "airplanstyle"] Damn you, DamthBroccoli, and Damn the Wright Brothers. ---------------------------- Julio Kriger mailto:juliokriger@xxxxxxxxx -------------- nexpar-------------- AHTML attachmenwas scrubbed... URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20051025= /a246e000/attachment-0001.htm Froshemminger aosdl.org Mon Oct 31 20:30:01 2005 From: shemminger aosdl.org (Stephen Hemminger) Date: Wed Apr 18 17:37:48 2007 Subject: Packets Per Second In-Reply-To: <568830273DB34B4CA3717D584351E8BC0101A666@xxxxxxxxxxxxxxxx> References: <568830273DB34B4CA3717D584351E8BC0101A666@xxxxxxxxxxxxxxxx> Message-ID: <20051031203001.5d7244d6@localhost.localdomain> OSun, 30 Jan 2005 18:48:18 +0200 "YaroGottfried" <YaronG@xxxxxxxxxx> wrote: > Hi, > > I havrecently installed "netem" in order to simulata delay of up to > 600msec. > I hava few questions abouthis "application" that I would appreciate > if someoncan answer: > > 1. Iorder to simulata 600msec delay, packets should be stored in > sombuffer. > I would likto know whais the size of the buffer in terms of memory > and packets per second. Iis a linked lisof packets, and the limit is one of the parameters of thqdisc. Thdefault limit is 1000 packets. If you want to limit iby bytes, then thsimplest way would be to replace the nested qdisc with a FIFO limited by bytes "bfifo". Senetewiki page for examples: http://linux-net.osdl.org/index.php/Netem > > 2. Does this "application" simulatdelays of multicaspackets? Yes all packets outpugedelayed. > Thanks iadvance, > > Yaron > ******************************************************************************** > Thcontents of this email and any attachments arconfidential, and are proprietary > of "ShiroSatellitCommunication". It is intended for the named recipient(s) only. > If you havreceived this email in error, pleasnotify us immediately by replying to > thmessagand deleting it from your computer. > Do nodisclosthe contents to anyone or make copies. > > ** eSafscanned this email for viruses, vandals and malicious conten** > **************************************************************************************
