Menu
▾
▴
Re: [Pyxida-users] V/v: Problem with jpcap
|
From: Ngo C. C. <cuo...@gm...> - 2009-04-24 09:23:35
|
Thanks for your response, I wonder for how long we can reach the convergence state ? It should be in maximum 2500 turns ? In fact I plotted the coordinates of all the nodes in 170 hours and it seems not convergent (you could take a look at the coordinates of node onelab03.inria.fr in 170 hours in attached file). For the first 30000-100000 things are good, the coordinates are stable, but after that, about 200000 seconds, it change unstably. Cuong. Jonathan Ledlie wrote: > > Cuong, > > Thanks for being our guinea pig :-) > > It is possible that there are mismatches going on in our Jpcap wrapper. > This might explain both problems. > > So if host A sends two pings to host B and the first one is dropped, we > might be returning the latency value as the total of the send of the > first to the receipt of the second. > > Why don't you look at a few inter-node latency distributions and see if > what you are getting makes sense? > > The list [] would happen if one ping was enclosed by another. > > I started to put some code in that checks the 'ident' received. The way > to fix the problem you see might be to only remove items when the ident > matches what has been sent. > > -jonathan > > On Apr 23, 2009, at 11:17 AM, Ngo Cao Cuong wrote: > >> Hello, >> >> I ran newest Pyxida (from CVS) on Planetlab (about 170 nodes), but >> after some days there's still nodes quit (about 40), and the >> coordinates change unstably, >> I've tracked some errors but don't know how to fix them. Could you >> take a look and help me fix it. >> >> Thank you. >> >> An error : >> >> DE 1240415906095 NCManager : !addPendingNeighbor: >> freedom.ri.uni-tuebingen.de:55904 >> DE 1240415906095 NCManager : app rtt of 37.526696 to >> itchy.cs.uga.edu:55904 >> DE 1240415906095 JpcapPinger : Sending new ping to >> remoteNode=itchy.cs.uga.edu:55904 using ICMP >> DE 1240415906095 JpcapPinger : Creating ICMP ping to >> itchy.cs.uga.edu:55904 >> DE 1240415906095 JpcapPinger : Ethernet header for this ICMP >> ping is jpcap.packet.EthernetPacket@1f0cbfb >> 00:0b:db:90:c6:97->00:0f:34:56:7b:80 (2048) >> DE 1240415906095 JpcapPinger : SENT seq=299 id=5639 orig=0 >> trans=0 recv=0 >> DE 1240415906096 JpcapPinger : sendPacket [raw=true]=0:0 >> /165.91.83.22->/128.192.101.217 protocol(1) priority(0) hop(100) >> offset(0) ident(0)type(8) code(0) sendTS=1240415906095 >> DE 1240415910581 JpcapPinger : in receivePacket >> packet=1240415906:127208 /128.192.101.217->/165.91.83.22 protocol(1) >> priority(0) hop(50) offset(0) ident(25984)type(0) code(0) >> DE 1240415910581 JpcapPinger : Captured a packet from >> src=/128.192.101.217 >> DE 1240415910581 JpcapPinger : Found a matching ping request >> WA 1240415910729 JpcapPinger : timeout ident=299 >> Exception in thread "main" java.lang.AssertionError: PingData >> pd=edu.harvard.syrah.pyxida.ping.JpcapPinger$JpcapPingData@73b72 not >> found in list: [] >> at >> edu.harvard.syrah.pyxida.ping.JpcapPinger.removeRequest(JpcapPinger.java:205) >> >> at edu.harvard.syrah.pyxida.ping.JpcapPinger$2.cb(JpcapPinger.java:389) >> at >> edu.harvard.syrah.sbon.async.CallbacksIF$CB0.callCBResult(CallbacksIF.java:168) >> >> at >> edu.harvard.syrah.sbon.async.CallbacksIF$AbstractCB$1.cb(CallbacksIF.java:71) >> >> at edu.harvard.syrah.sbon.async.EL.handleEvent(EL.java:659) >> at edu.harvard.syrah.sbon.async.EL.handleEventQueues(EL.java:594) >> at edu.harvard.syrah.sbon.async.EL.main(EL.java:406) >> at edu.harvard.syrah.pyxida.Pyxida.main(Pyxida.java:191) >> >> Another type of error: >> >> IN 1240418952240 NCClient : getNeighborToPing: pinged/non >> 32/0 avgTimeSincePing 369498.78125 minTime 1 >> IN 1240418952240 NCClient : getNeighborToPing: lowered >> MIN_UPDATE_TIME_TO_PING 1 >> IN 1240418952240 NCClient : getNeighborToPing: picking from >> grayingNeighbors 32/32 planetlab2.ifi.uio.no:55904 >> DE 1240418952240 NCManager : getUpNeighbor using up: >> itchy.cs.uga.edu:55904 >> DE 1240418952241 NCManager : Sending gossip request to >> planetlab2.ifi.uio.no:55904 >> WA 1240418952570 ObjComm$ReadConnHa: Received another connection from >> remoteAddr=planetlab2.ifi.uio.no:55904 >> WA 1240418952571 ObjComm$ReadConnHa: Closing old connection. >> Exception in thread "main" java.nio.channels.CancelledKeyException >> at sun.nio.ch.SelectionKeyImpl.ensureValid(SelectionKeyImpl.java:55) >> at sun.nio.ch.SelectionKeyImpl.readyOps(SelectionKeyImpl.java:69) >> at edu.harvard.syrah.sbon.async.EL.handleSelectCallbacks(EL.java:770) >> at edu.harvard.syrah.sbon.async.EL.main(EL.java:434) >> at edu.harvard.syrah.pyxida.Pyxida.main(Pyxida.java:191) >> >> Jonathan Ledlie wrote: >>> >>> I've been using jpcap 0.7 and jdk 1.6.0_11. I'm fairly sure the >>> jpcap.jar is 0.7. >>> >>> Maybe you should try jpcap outside of Pyxida and see if that works on >>> your system. >>> >>> -jonathan >>> >>> On Apr 14, 2009, at 7:44 AM, Ngo Cao Cuong wrote: >>> >>>> >>>> Hello, >>>> >>>> Could you tell me with version of jpcap and jdk you currently use with >>>> Pyxida ? Because I've tried compiling with jpcap 0.5.1 source and >>>> there's some errors with (with the lastest code of pyxida), I tried the >>>> jpcap.rar file (I dont know what version it is) on the CVS repository, >>>> no error on compiling but there's a runtime error : >>>> "Exception in thread "Thread-2" java.lang.NoSuchMethodError: >>>> setRoutingOption >>>> at jpcap.JpcapCaptor.nativeOpenLive(Native Method) >>>> at jpcap.JpcapCaptor.openDevice(JpcapCaptor.java:68) >>>> at >>>> edu.harvard.syrah.pyxida.ping.JpcapPinger$1.run(JpcapPinger.java:107) >>>> >>>> I tried with both jpcap 0.6 and 0.7 on its site but there's the same >>>> error. >>>> (I use jdk1.6.0_13). >>>> >>>> Thank you very much for your help. >>>> >>>> Jonathan Ledlie wrote: >>>>> >>>>> Yeah, that's not going to work as is. We've asked a student at >>>>> Tsinghua >>>>> to fix it, but you are welcome to give it a crack. In the >>>>> meantime, UDP >>>>> should work. >>>>> >>>>> -jonathan >>>>> >>>>> On Apr 7, 2009, at 4:36 PM, Cao Cuong Ngo wrote: >>>>> >>>>>> I've tried using the version 0.1.5 on the site(both binary and >>>>>> compiling from code) and it doesn't work with Java 1.5 (only work >>>>>> with >>>>>> 1.6). >>>>>> The version 0.1.4 worked with Java 1.5 and jpcap 0.5.1 but the same >>>>>> error SendTS > RecvTS happened to me as 0.1.5. >>>>>> I've also tried with application ping but Vivaldi never converges >>>>>> because of the instability of rtt measurements. >>>>>> >>>>>> Thank you, >>>>>> Cuong. >>>>>> >>>>>> 2009/4/7, Jonathan Ledlie <jon...@ee...>: >>>>>>> >>>>>>> Using Java 1.5 and jpcap 0.5.1 worked, IIRC. >>>>>>> >>>>>>> I'd recommend using UDP if possible for the time being. >>>>>>> >>>>>>> Another Pyxida user is looking into a fix, but if you find one, that >>>>>>> would be great. We haven't looked and I doubt it is a major >>>>>>> incompatability. >>>>>>> >>>>>>> -jonathan >>>>>>> >>>>>>> On Apr 7, 2009, at 9:27 AM, Peter Pietzuch wrote: >>>>>>> >>>>>>>> Ngo Cao Cuong wrote: >>>>>>>>> I haven't found where we can fix it yet but I'll try. >>>>>>>>> At the moment could you tell me which configuration (of pyxida, >>>>>>>>> jdk, >>>>>>>>> jpcap) we can implement stably on PlanetLab? >>>>>>>> >>>>>>>> Jon, what was the last version that ran successfully on PlanetLab >>>>>>>> with pcap? >>>>>>>> >>>>>>>> Cheers, >>>>>>>> Peter >>>>>>>> >>>>>>>>> Peter Pietzuch wrote: >>>>>>>>>> Ngo Cao Cuong wrote: >>>>>>>>>> >>>>>>>>>>> Thank you for your answer. >>>>>>>>>>> I installed java 1.6.0_13, I tried jpcap 0.5.1, it did work but >>>>>>>>>>> everytime the RecvTS is much bigger than SendTS so the >>>>>>>>>>> latency is >>>>>>>>>>> very >>>>>>>>>>> big (like 124535363633), so every node one planetlab update with >>>>>>>>>>> [0,0,0,0] >>>>>>>>>>> >>>>>>>>>> This is the correct version to use. We are aware of the bug with >>>>>>>>>> Java >>>>>>>>>> 1.6 and the recvTS/sendTS timestamps but haven't fixed it yet. I >>>>>>>>>> think >>>>>>>>>> it has something to do with the fact that the resolution of >>>>>>>>>> nanoTime has >>>>>>>>>> changed in Java 1.6. Can you see if you can fix it? >>>>>>>>>> >>>>>>>>>> Cheers, >>>>>>>>>> Peter >>>>>>>>>> >>>>>>>>>> >>>>>>>>>>> Peter Pietzuch wrote: >>>>>>>>>>> >>>>>>>>>>>> Hello, >>>>>>>>>>>> >>>>>>>>>>>> Ngo Cao Cuong wrote: >>>>>>>>>>>> >>>>>>>>>>>>> Well finally I solved this problem (because we need to install >>>>>>>>>>>>> java 1.6 >>>>>>>>>>>>> instead of 1.5!) but there is always an error with jpcap >>>>>>>>>>>>> "exception in >>>>>>>>>>>>> thread"thread-2" java.lang.nosSuchMethodError: >>>>>>>>>>>>> setRoutingOption >>>>>>>>>>>>> at jpcap.JpcapCaptor.nativeOpenLive(Native Method) >>>>>>>>>>>>> at jpcap.JpcapCaptor.openDevice(JpcapCaptor.java:68) >>>>>>>>>>>>> at edu.harvard.syrah.pyxida.ping.JpcapPinger >>>>>>>>>>>>> $1.run(JpcapPinger.java:95) >>>>>>>>>>>>> >>>>>>>>>>>> This seems like a version conflict between the Java jpcap >>>>>>>>>>>> library and >>>>>>>>>>>> the pcap installation on your machine. Can you check the >>>>>>>>>>>> versions? >>>>>>>>>>>> >>>>>>>>>>>> Cheers, >>>>>>>>>>>> Peter >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>>>> Hello, >>>>>>>>>>>>>> >>>>>>>>>>>>>> I've tried to run 0.1.5 but there is always an error: >>>>>>>>>>>>>> "Exception in thread "main" >>>>>>>>>>>>>> java.lang.UnsupportedClassVersionError: Bad >>>>>>>>>>>>>> version number in .class file" >>>>>>>>>>>>>> >>>>>>>>>>>>>> I've tried to run 0.1.4 too , it did run but sometimes there >>>>>>>>>>>>>> is some >>>>>>>>>>>>>> nodes quit the network with the error of JPcap "RecvTS < >>>>>>>>>>>>>> SendTS" ??? >>>>>>>>>>>>>> (error on the assert RecvTS > SendTS) >>>>>>>>>>>>>> >>>>>>>>>>>>>> Anyone can help me? >>>>>>>>>>>>>> >>>>>>>>>>>>>> Thank you, >>>>>>>>>>>>>> Ngo Cao Cuong >>>>>>>>>>>>>> INRIA Sophia Antipolis, France. >>>>>>>>>>>>>> >>>>>>>>>>>>> ------------------------------------------------------------------------------ >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> This SF.net email is sponsored by: >>>>>>>>>>>>> High Quality Requirements in a Collaborative Environment. >>>>>>>>>>>>> Download a free trial of Rational Requirements Composer Now! >>>>>>>>>>>>> http://p.sf.net/sfu/www-ibm-com >>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>> Pyxida-users mailing list >>>>>>>>>>>>> Pyx...@li... >>>>>>>>>>>>> https://lists.sourceforge.net/lists/listinfo/pyxida-users >>>>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> ------------------------------------------------------------------------------ >>>>>>>>> >>>>>>>>> >>>>>>>>> This SF.net email is sponsored by: >>>>>>>>> High Quality Requirements in a Collaborative Environment. >>>>>>>>> Download a free trial of Rational Requirements Composer Now! >>>>>>>>> http://p.sf.net/sfu/www-ibm-com >>>>>>>>> _______________________________________________ >>>>>>>>> Pyxida-users mailing list >>>>>>>>> Pyx...@li... >>>>>>>>> https://lists.sourceforge.net/lists/listinfo/pyxida-users >>>>>>>> >>>>>>>> >>>>>>>> --Dr Peter R Pietzuch >>>>>>>> Department of Computing, Imperial College London >>>>>>>> South Kensington Campus, 180 Queen's Gate, London SW7 2AZ, UK >>>>>>>> Telephone: +44 (20) 7594 8314 http://www.doc.ic.ac.uk/~prp >>>>>>>> >>>>>>>> ------------------------------------------------------------------------------ >>>>>>>> >>>>>>>> >>>>>>>> This SF.net email is sponsored by: >>>>>>>> High Quality Requirements in a Collaborative Environment. >>>>>>>> Download a free trial of Rational Requirements Composer Now! >>>>>>>> http://p.sf.net/sfu/www-ibm-com >>>>>>>> _______________________________________________ >>>>>>>> Pyxida-users mailing list >>>>>>>> Pyx...@li... >>>>>>>> https://lists.sourceforge.net/lists/listinfo/pyxida-users >>>>>>> >>>>>>> >>>>>> >>>>>> ------------------------------------------------------------------------------ >>>>>> >>>>>> >>>>>> This SF.net email is sponsored by: >>>>>> High Quality Requirements in a Collaborative Environment. >>>>>> Download a free trial of Rational Requirements Composer Now! >>>>>> http://p.sf.net/sfu/www-ibm-com >>>>>> _______________________________________________ >>>>>> Pyxida-users mailing list >>>>>> Pyx...@li... >>>>>> https://lists.sourceforge.net/lists/listinfo/pyxida-users >>>>> >>> > |
